Miljø- og Fødevareudvalget 2021-22
MOF Alm.del Bilag 605
Offentligt
2597473_0001.png
TRANSPORT OF NITROGEN AND
PHOSPHORUS FROM LAND TO SEA
AROUND YEAR 1900
Scientific Report from DCE – Danish Centre for Environment and Energy
No. 498
2022
AU
AARHUS
UNIVERSITY
DCE – DANISH CENTRE FOR ENVIRONMENT AND ENERGY
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0003.png
TRANSPORT OF NITROGEN AND
PHOSPHORUS FROM LAND TO SEA
AROUND YEAR 1900
Scientific Report from DCE – Danish Centre for Environment and Energy
No. 498
2022
Hans Estrup Andersen
1
Karsten Arnbjerg-Nielsen
5
Camilla Bitsch
5
Sarah Brudler
5
Bent T. Christensen
2
Jørgen Eriksen
2
Goswin Heckrath
2
Carl Christian Hoffmann
1
Anker Lajer Højberg
4
Jørgen E. Olesen
2
Birger F. Pedersen
2
Johannes W.M. Pullens
2
Martin Rygaard
5
Gitte Rubæk
2
Mikkel Thelle
6
Hans Thodsen
1
Henrik Tornbjerg
1
Lars Troldborg
4
Flemming Vejen
3
DCE-Danish Centre for Environment and Energy, Aarhus University
2
DCA-Danish Centre for Food and Agriculture, Aarhus University
3
DMI- Danish Meteorological Institute
4
GEUS –Geological Survey of Denmark and Greenland
5
DTU Sustain, Department of Environmental and Resource Engineering, Technical University of Denmark
6
School of Culture and Society, Aarhus University
1
AU
AARHUS
UNIVERSITY
DCE – DANISH CENTRE FOR ENVIRONMENT AND ENERGY
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0004.png
Data sheet
Series title and no.:
Category:
Title:
Editors:
Authors:
Scientific Report from DCE – Danish Centre for Environment and Energy No. 498
Scientific advisory report
Transport of nitrogen and phosphorus from land to sea around year 1900
Signe Jung-Madsen
1
; Hanne Bach
1
Hans Estrup Andersen
1
, Karsten Arnbjerg-Nielsen
5
, Camilla Bitsch
5
, Sarah Brudler
5
,
Bent T. Christensen
2
, Jørgen Eriksen
2
, Goswin Heckrath
2
, Carl Christian Hoffmann
1
,
Anker Lajer Højberg
4
, Jørgen E. Olesen
2
, Birger F. Pedersen
2
, Johannes W.M. Pullens
2
,
Martin Rygaard
5
, Gitte Rubæk
2
, Mikkel Thelle
6
, Hans Thodsen
1
, Henrik Tornbjerg
1
, Lars
Troldborg
4
and Flemming Vejen
3
Centre for Environment and Energy, Aarhus University,
2
DCA-Danish
Centre for Food and Agriculture, Aarhus University,
3
DMI- Danish Meteorological
Institute,
4
GEUS –Geological Survey of Denmark and Greenland,
5
DTU Sustain,
Department of Environmental and Resource Engineering, Technical University of
Denmark &
6
School of Culture and Society, Aarhus University
Aarhus University, DCE – Danish Centre for Environment and Energy ©
http://dce.au.dk/en
June2022
June 2022
Mathias Neumann Andersen
3
, Peter Steen Mikkelsen
5
, Torben Schmidt
3
, Dennis
Trolle
1
, Ingrid K. Thomsen
2
, Jørgen Windolf
1
Markus Venohr from the Leibniz-Institute of Freshwater Ecology and Inland Fisheries
in Germany and Erwin van Boekel from Wageningen Environmental Research,
Netherlands.
Anja Skjoldborg Hansen
Anne Mette Poulsen
The comments can be found here:
http://dce2.au.dk/pub/komm/SR498_komm.pdf
Miljøstyrelsen
Jung-Madsen, S. and Bach H. (red.) 2022. Transport of nitrogen and phosphorus from
land to sea around year 1900. Aarhus University, DCE – Danish Centre for
Environment and Energy, 192 pp. Scientific Report No. 498
http://dce2.au.dk/pub/SR498.pdf
Reproduction permitted provided the source is explicitly acknowledged
Abstract:
The nutrient loads from land to sea around year 1900 in Denmark are estimated using a
delta change modelling approach considering the numerous factors affecting the
nutrient inputs and transport. The estimates are based on available data from that time,
literature, comparative analysis methods and modelling tools. The main factors
investigated are climate, hydrology, land use, agricultural practices and drainage, urban
developments and landscape (e.g. nutrient retention in groundwater, wetland, lakes and
streams). Nutrient loads around the year 1900 were affected by human activity, with
total nitrogen and phosphorous loads being approx. 40% and 25-40% less than present-
day loadings, respectively.
Nutrient transport, nutrient load, nutrient concentration, nitrogen, phosphorus, water
discharge, reference condition, point source, urban development, year 1900, climate,
land use, retention.
978-87-7156-691-8
2245-0203
192
The report is available in electronic format (pdf) at
http://dce2.au.dk/pub/SR498.pdf
Chapters were no authors are listed are written by editors with input from the authors
of the report (preface, summary, introduction and synthesis of results).
1
DCE-Danish
Institutions:
Publisher:
URL:
Year of publication:
Editing completed:
Referees:
International review:
Quality assurance, DCE:
Linguistic QA:
External comments:
Financial support:
Please cite as:
Keywords:
ISBN:
ISSN (electronic):
Number of pages
Internet version:
Supplementary notes
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Contents
Preface
Summary
Sammenfatning
Introduction
1
Climate around year 1900: Temperature and precipitation
Abstract
1.1 Introduction
1.2 Material and methods
1.3 Results
1.4 Evaluation of uncertainty
1.5 Conclusion
1.6 References
2
Climate around the year 1900: Calculation of global radiation
and evaporation for historic sites
Abstract
2.1 Introduction
2.2 Materials and methods
2.3 Results
2.4 Conclusion
2.5 References
3
Modelling discharge to the sea
Abstract
3.1 Introduction
3.2 Material and methods
3.3 Results
3.4 Conclusion
3.5 References
4
Point source emissions of nutrients from urban areas in 1900
Abstract
4.1 Introduction
4.2 Materials and methods
4.3 Results and discussion
4.4 Conclusion
4.5 References
5
Land use, agriculture and nitrate concentrations in root-zone
percolates around year 1900
Abstract
5.1 Introduction
5.2 Agriculture around year 1900
5.3 Estimating the N concentration in root zone percolates
5.4 References
7
9
14
19
21
21
22
23
29
38
41
42
45
45
45
45
47
50
50
51
51
52
52
59
64
64
66
66
66
67
69
75
76
79
79
79
80
85
88
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
6
The National Nitrogen Model and other Nitrogen Sources
Abstract
6.1 The national nitrogen model
6.2 Point sources
6.3 Atmospheric nitrogen deposition year 1900
6.4 Organic nitrogen concentrations
6.5 References
91
91
91
95
95
95
98
100
100
101
101
102
105
106
107
108
109
110
111
112
115
115
115
115
118
123
125
125
125
127
138
142
142
146
147
148
152
153
155
159
159
160
163
7
Modelling nitrogen retention in surface waters
Abstract
7.1 Introduction
7.2 Materials and methods
7.3 Nitrogen retention in natural wetlands
7.4 Nitrogen retention in constructed wetlands
7.5 Nitrogen retention in small watercourses
7.6 Nitrogen retention in larger watercourses
7.7 Nitrogen retention in small lakes
7.8 Nitrogen retention in larger lakes
7.9 Total surface water nitrogen retention around year 1900
compared with the present period
7.10 National Nitrogen Model – nitrogen mass balance
7.11 References
8
Calculating the nitrogen load to the sea
Abstract
8.1 Introduction
8.2 Material and methods
8.3 Results
8.4 References
9
Phosphorus losses from the Danish land area to the sea
around year 1900
Abstract
9.1 Introduction
9.2 Material and methods
9.3 Results and discussion
9.4 Conclusion
9.5 References
10 Uncertainty and sensitivity
10.1
10.2
10.3
10.4
Quantitative uncertainty and sensitivity - nitrogen
Qualitative uncertainty and sensitivity -nitrogen
Uncertainties on phosphorus load around year 1900
References
11 Synthesis of results
12 Perspectives of results
12.1 Comparisons to other root zone nitrate leaching studies
12.2 Comparisons to other river load studies
12.3 References
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Appendix to chapter 4 (4.1 and 4.2).
Appendix to chapter 5 (5.1)
166
169
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0009.png
Preface
In order to assess the status and set environmental goals for waterbodies un-
der the WFD, it is necessary to determine the border between “good” and
“moderate” ecological status of the waterbody. The boundary is set by defin-
ing a “reference condition” reflecting an undisturbed/pristine condition, add-
ing to this an acceptable deviation to allow for some human impact. The WFD
describes the reference conditions as:
The values of the biological quality elements
for the surface water body reflect those normally associated with that type under un-
disturbed conditions, and show no, or only very minor, evidence of distortion (WFD,
Annex V. p 38 (https://eur-lex.europa.eu/resource.html?uri=cellar:5c835afb-2ec6-
4577-bdf8-756d3d694eeb.0004.02/DOC_1&format=PDF)
The ecological status in coastal waters is based on the status of the “ecological
quality elements”, here eelgrass and chlorophyll-a. For eelgrass, historical in-
formation from around the year 1900 has been used to establish the reference
condition. This type of information is not available for chlorophyll-a, hence it
has been necessary to use models to estimate a reference situation. As the con-
centration of chlorophyll-a is tightly linked to the concentration of nutrients
in the water, reference load of nutrients from land to the sea are needed in
order to run the models.
In October 2017, Aarhus University published the report “Estimation of nitro-
gen concentrations from root zone to marine areas around the year 1900” (Jen-
sen (ed.) 2017)
1
, which investigates the factors that could influence the annual
mean nitrogen (N) concentrations from source to sea around the year 1900
and provides an estimation of the nitrogen concentration level in stream run-
off to the sea at that time in Denmark. The main factors investigated were
climate, hydrology, land use, agricultural practices, drainage and landscape
(retention). The result was provided as a range of concentrations for the whole
country but did not include point sources and the contribution of organic ni-
trogen. However, if the results are to be used in modelling studies for target
setting for, for instance, chlorophyll-a, a more detailed spatial distribution of
N concentrations and water runoff is required.
Therefore, the Danish Environmental Protection Agency (Danish EPA) re-
quested that DCE, Aarhus University initiated a follow-up project where the
analysis and estimations are made at a more detailed spatial scale and include
an in-depth analysis of the contribution of nutrients from the larger cities in
Denmark (point sources) to the modelled N concentrations. Furthermore,
there was a wish to include phosphorus in the analysis. The goal of the project
was to obtain a geographically distributed dataset of nutrient inputs to the sea
from around the year 1900 with the hope that it could be used in defining the
reference condition for target setting according to the WFD.
The project was initiated in 2019 but has, due to difficulties in obtaining reli-
able climate data for the years around the year 1900 and thus reliable mod-
elled water discharge, been delayed several times. However, by early 2020 the
first preliminary results from the project were presented at the national event
“Plantekongres” in January 2020, showing that nutrient concentrations
around year 1900 were significantly influenced by agricultural activities, ur-
ban wastewater etc. Based on the presentations, it was concluded that the es-
timated supply of nutrients to the sea around the year 1900 cannot be assumed
7
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
to have been unaffected or only slightly affected by human activity (Timmer-
mann, 2020
2
). Hence, the nutrient concentration and load around year 1900 in
Denmark cannot be associated with an undisturbed situation such as required
when establishing reference conditions in relation to the WFD. One of the
main reasons for this conclusion is that the nutrient concentration levels esti-
mated in this study are significantly higher compared with the current meas-
urements of nutrient concentrations in streams with low human impact. Con-
sequently, DCE recommended using an alternative approach to estimate nu-
trient loads representing reference conditions regarding the WFD (Timmer-
man, 2020
2
).
The current report documents the substantial work that has been carried out
in this project and for the first time presents detailed analysis of agricultural
practices, point source pollution and model-based estimates of climate condi-
tions, water discharge and nutrient flows around the year 1900 in Denmark.
During the project period, meetings with an advisory group of external inter-
ested parties were organized by the Danish EPA. At these meetings results
from the project were presented and discussed. Further, several project status
meetings were held with the Danish EPA.
In June 2022 a draft report was sent in international scientific review. The re-
viewers were Dr. Markus Venohr from the Leibniz-Institute of Freshwater
Ecology and Inland Fisheries in Germany and ir.EMPM Erwin van Boekel
from Wageningen Environmental Research, in the Netherlands. Following the
review a draft report was send to the Danish EPA who provided written com-
ments to the report.
Acknowledgement:
The authors of the report would like to thank Tinna Christensen for assistance
with graphical work and Karin Balle Madsen for setting up the report and also
the many colleagues at DCE, DCA, DMI, GEUS and DTU who have partici-
pated in discussions of the result during the project. We especially want to
thank the two reviewers Marcus Venohr and Erwin Van Boekel for the many
useful comments and suggestions that have led to an improved product.
1
Jensen
P.N. (ed.), 2017. Estimation of nitrogen concentrations from root zone to marine areas
around the year 1900. Scientific Report from DCE No. 241. Aarhus University, Danish Centre for
Environment and Energy.
2
Timmermann,
K. (2020). Referencetilførsler af kvælstof til brug for Vandplan 3. DCE – National
Center for Miljø og Energi
8
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Summary
Background
The current report aims to describe the nutrient loads from land to sea around
the year 1900. The nutrient loads are calculated considering the various fac-
tors affecting the nutrient inputs and transport based on available data from
that time, literature, comparative analysis methods and modelling tools. The
main factors investigated are climate, hydrology, land use, agricultural prac-
tices and drainage, urban development and landscape (e.g. nutrient retention
in groundwater, wetlands, lakes and streams). The ambition was to use as
much data and information from the time as possible, taking into considera-
tion the quality and representativeness, and use modelling and GIS tools to
provide a geographically distributed estimate of total nitrogen (TN) and total
phosphorus (TP) concentrations and loads from the root zone to the sea. To
be able to do this, the datasets on climate and hydrology were analysed and
expanded as part of the project, and runoff was modelled using the national
water resources model (DK- model).
The agricultural practices and land use in different areas of the country were
analysed in detail, and the root zone concentration of nitrogen was deter-
mined for different land use categories. This was based on old farmland sta-
tistics from Denmark and Northern Germany, which at the time included land
areas that are now Danish, and root zone nitrogen concentrations extracted
from farming experiments resembling year 1900 farm management practices.
Phosphorus inputs to the surface waters from the landscape, including agri-
cultural activities, were estimated for the main input categories: soil drainage,
land reclamation, grazing animals, soil erosion and manure storage.
Furthermore, a detailed analysis of the contribution of nutrients from cities
(point sources) was included. This analysis estimated nitrogen as well as
phosphorus inputs.
To model the concentration and transport of nitrogen in the freshwater load
to the sea, the national nitrogen model (NMN) was used. The model requires
geographically distributed input of climate variables, such as temperature
and water runoff, the input of nitrogen from the root zone and point sources
as well as data on the depth of groundwater and the proportion and location
of wetlands, lakes and streams, to estimate the retention in the surface water
system.
The phosphorus analysis used an approach of ”background” or ”nature” wa-
ter concentration levels of phosphorous, on top of which the relevant addi-
tional diffuse and point sources were added and retention in lakes subtracted.
The transport of phosphorus through the catchments was simulated using the
same water discharge as for the nitrogen modelling, but the model as such is
much simpler than the nitrogen model because of the way phosphorus be-
haves in the environment.
Among other things, the analysis showed that in the year 1900, weather con-
ditions were colder and drier than today, more land was in agricultural use
but less was tile drained, and due especially to a larger proportion of wetlands
the retention of nutrients in the landscape was higher.
9
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
The conclusion from the chapters on climate, hydrology, nutrient input nutri-
ent transport from land to sea and uncertainties are given in the synthesis of
results in chapter 11 and are also presented below.
Climate
The climate was colder and drier around year 1900 compared with the pre-
sent-day. The estimated average annual precipitation around year 1900 was
about 60 mm, or 7% lower than today. Digitized climate data from around
year 1900 at observation points across the country, including temperature,
wind and rainfall, were used to find monthly values of bias-corrected precip-
itation. The correction approach was evaluated for the period 1917-1950 using
water balance modelling of discharge. At national level, a water balance error
of 3% indicated reasonable correction estimates, but large regional differences
in error level was found. In order to obtain spatially distributed corrected pre-
cipitation for the period 1890-1910 a delta change climate factor approach was
used. In this approach national monthly correction factors were calculated
based on corrected precipitation for 1890-1910 compared with 1989-2010.
These national factors were then applied to the present daily corrected pre-
cipitation assuming a similar geographical distribution of precipitation
around year 1900 as in the present time reference period (1989-2010) to pro-
vide a spatially distributed daily time series of precipitation for the period
1890-1910.
To be able to model nitrate leaching, global radiation and potential evapotran-
spiration must be calculated. By using the measured minimum and maximum
air temperatures for 1890-1950, the global radiation and potential evapotran-
spiration can be calculated and used in the simulation of nitrate leaching in
this period. The modelled global radiation and potential evapotranspiration
around year 1900 are in good agreement with values measured at Foulum
from 1987-2013.
Hydrology
The total discharge based on the precipitation estimates and drainage density
estimated for the historic time was in average 292 mm/yr for the period 1890-
1910 compared with 333 mm/yr for the present period (1990-2010) as calcu-
lated by the hydrological model (DK model, chapter 3). Subsequently, apply-
ing a delta change method, the total discharge was recalculated to 335 mm/yr
for the present period and 297 mm/yr for the year 1900 period (chapter 8).
This means that for the total average, annual discharge was about 11% lower
around year 1900 compared with the present time. The change in discharge
for the two periods largely reflects the changes in precipitation, but is ampli-
fied in some areas by the lower density of drainage in the historical period.
The calculated change in discharge when comparing the present time to the
time around year 1900 ranged between 0 and 20% for most of the country,
which agrees with the trend analysis of long discharge time-series presented
in Jensen (ed.) (2017). However, for the western part of Zealand, the data in-
dicate an increase in discharge between the two periods of approximately
30%, while the simulation resulted in a decrease in discharge of approx. 5%.
The methodology used implies that the total discharge to the sea may resem-
ble the conditions around year 1900, but it cannot be expected to reproduce
the local conditions at that time.
10
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Nutrient inputs
The nutrients reaching the sea from the land mainly originated from agricul-
tural activities and dwellings across the country around year 1900. In addi-
tion, runoff from erosion along the streams contributed to the nutrient content
in freshwater, particularly phosphorus.
Agriculture
The area in agricultural use increased dramatically during the last half of the
19th century and accounted for close to 3/4 of the area under Danish admin-
istration around year 1900. Crop production differed significantly from cur-
rent agriculture for virtually all growth factors: inferior crop varieties, higher
weed pressure, lack of chemical crop protection and inferior plant nutrient
supply, including the absence of mineral fertiliser. The main sources of nutri-
ents were solid farmyard manure, liquid manure and nitrogen fixation by leg-
ume crops. The number and categories of livestock as well as the farm struc-
ture and management practices around year 1900 also differed from today’s
practices.
Parish-level statistics from around year 1900 for the area under current Danish
administration were unified into eight categories (winter and spring crops,
grass, root crops, fallow, nature and forest), and for each category a nitrogen
concentration was ascribed to the root zone percolate. The nitrogen root zone
concentrations were set using data from studies of organic farming as a proxy
for the past time situation. Literature data were found for the remaining cate-
gories. These values were applied in the nitrogen modelling. The calculation
of the area-weighted average nitrogen concentration for land in agricultural
use (78% of the land area) resulted in a value of 12 mg N/l, while the value
for the entire land area was 9.6 mg N/l in root zone percolate (inorganic ni-
trogen).
The estimation of agricultural sources for phosphorus considered factors such
as soil drainage, land reclamation, grazing animals, soil erosion and manure
storage. These factors were difficult and uncertain to determine, leading to an
estimated range from 56 to 196 ton P annually around year 1900.
Point sources
Sewer systems were increasingly implemented in towns, but wastewater
treatment did not exist in year 1900. Therefore, towns were significant point
sources around year 1900, with 4,261 ton N/yr and 764 ton P/yr emitted in
excrements from humans and animals and industrial wastewater. The find-
ings indicate that the majority of the nutrients from point sources discharged
directly to receiving waters (55%), but emissions to landfills (20%) and agri-
cultural soil (25%) were significant as well. The total contribution from inland
and direct point sources to water was estimated to 471 ton P, about 65–70% of
the present-day value (704 ton P, average 2014–2018) and 2,531 ton N, about
47% of the present-day TN point sources (5,400 ton N, 2020).
Other sources, including background nutrient concentration
Nitrogen inputs from the atmosphere were estimated by multiplying EMEP
simulations for year 2000 by 0.3 (Jensen (ed.), 2017). Organic nitrogen origi-
nating from landscape sources and internal surface water sources was in-
cluded to be able to calculate total nitrogen concentrations. Estimates based
on literature studies assume that the organic nitrogen concentration around
year 1900 was about 20% below the current levels. Furthermore, it is assumed
11
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
that the current geographical distribution of organic N is valid for the time
around year 1900.
A literature review and measurements from largely undisturbed streams
were used to estimate background TP stream concentrations. An area-
weighted TP median value at 0.052 mg/l was estimated.
Nutrient transport to the sea
Nitrogen percolates through the soil and reaches the groundwater where re-
duction of nitrogen under oxygen-free conditions (retention) can take place
before the remaining nitrogen ends up in surface waters (wetlands, lakes,
streams). The National Nitrogen Model simulates transport and retention in
groundwater based on water discharge and nitrogen percolate input. The sur-
face water component calculates the nitrogen retention in wetlands, streams
and lakes, while also considering point source inputs, atmospheric inputs and
the contribution of organic nitrogen. Landscape changes between the time
around year 1900 and the present time were handled by modifying the current
landscape maps based on various information sources on the past landscape
related to rivers and lakes. For wetlands, different maps were used.
The phosphorus analysis was based on total phosphorus considerations and
used an approach of a ”background” or ”nature” concentration level, on top
of which the relevant additional agricultural and point sources were added
and retention in lakes was subtracted. The transport and routing of phospho-
rus through the catchments were simulated using the same water discharge
as in the nitrogen modelling.
The nitrogen retention in inland surface water was shown to be higher in the
present period (28,000 ton N) than in the 1900 period (26,000 ton N) due to a
larger present-day nitrogen load. However, the relative nitrogen retention
was higher around the year 1900, as 43% of the load was removed compared
with 33% for the present period.
The total nitrogen load is modelled to be approximately 36,000 ton N/yr
around year 1900, which is approximately 40% less than for the present period
(59,000 ton N/yr). The nitrogen concentration is modelled to be around 2.8
mg N/l around year 1900 compared with 4.1 mg N/l in the present period.
The national nitrogen model yields regional results, which are utilised for es-
timating regional year 1900 nitrogen and freshwater loads.
The estimated average stream water phosphorus concentration around year
1900 was 0.062–0.075 mg P/l, equivalent to 60–70% of the present-day stream
water TP concentration (0.1 mg P/l). The TP values were calculated for nine
geographical regions across the country. The total TP loading to the sea, in-
cluding background, other diffuse sources, inland and direct point sources
and subtracted phosphorus retention in lakes, was estimated to 1,200–1,340
ton P, 60–65% of (or 35-40% less than) the present-day phosphorus loading
(2,021 ton/yr).
12
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Uncertainties
Working with a period 120 years ago naturally makes most aspects of calcu-
lating the national nitrogen and phosphorus load more uncertain than when
calculating it for the present period. An in-depth analysis of uncertainties of
data layers, variables, model and model assumptions was not a part of the
present study however, some considerations regarding uncertainty and sen-
sitivity (the effect of a given parameter on the result) have been made.
Most of the parameters used to estimate the nitrogen loads around the year
1900 are considered to have “medium” uncertainty (on a three-step scale from
low to high). The uncertainty of the nitrogen loads is influenced by a variety
of factors, the most important being the uncertainty of the estimates of pre-
cipitation, run-off, root zone concentration of nitrogen and retention in sur-
face and groundwater.
Overall, the model concept used to calculate the year 1900 nitrogen load is
considered relatively robust and the overall uncertainty at national scale ac-
ceptable. However, the uncertainty increases with decreasing geographical-
and timescales.
Most of the parameters used to estimate phosphorus loads are considered
“medium” to “highly” uncertain. The uncertainty of phosphorous loads is es-
pecially influenced by the uncertainties of the estimates of precipitation, run
off, P input from point sources and the background TP concentration. Despite
the many uncertainties the results of this study are believed to be the best
possible estimate of the year 1900 phosphorous loads. Furthermore, the re-
sults are supported by historical lake measurements that also find the histor-
ical TP-concentrations to be lower than today but considerably higher than
the background concentration, though.
Perspectives
Many of the European studies that are compared with the present study re-
port nitrogen concentrations around the year 1900 that are considerably lower
than in the present study. The reasons for this are probably differences in
landscape, land use, farming practices and runoff between the investigated
areas and Denmark. It probably also reflects the degree to which agricultural
practices and nutrient dynamics are included in the studies. In the present
study, the year 1900 root zone leaching is calculated, and nitrogen fixation,
the main source of nitrogen in Danish agriculture in the year 1900, is consid-
ered, which is not the case in most other studies.
13
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Sammenfatning
Baggrund
Denne rapport har til formål at beskrive næringsstofbelastningen fra land til
kyst omkring år 1900 under hensyntagen til de forskellige faktorer, der påvir-
ker næringsstoftilførslen og transporten. Næringsstoftilførslen og transporten
er beregnet ud fra tilgængelige data fra omkring år 1900, litteratur, kompara-
tive analysemetoder og modelleringsværktøjer. De vigtigste faktorer, der un-
dersøges, er klima, hydrologi, arealanvendelse, landbrugspraksis og dræning,
byudvikling og landskabsforhold (fx retention af næringsstoffer i grundvand,
vådområder, søer og vandløb). Det var hensigten at bruge så mange data og
oplysninger fra perioden som muligt, kvaliteten og repræsentativiteten taget
i betragtning, og at anvende modellerings- og GIS-værktøjer til at give et geo-
grafisk fordelt skøn over total kvælstofkoncentration (TN) og total fosforkon-
centration (TP) i det vand, der transporteres til havet, samt et skøn over næ-
ringstoftransporten fra rodzonen til kyst. For at kunne gøre dette blev datasæt
om klima og hydrologi analyseret og udvidet som en del af projektet, og af-
strømningen blev modelleret ved hjælp af den nationale vandressourcemodel
(DK-modellen).
Landbrugspraksis og arealanvendelse i forskellige dele af landet blev analy-
seret i detaljer, og rodzonekoncentrationen af kvælstof blev fastsat for forskel-
lige kategorier af arealanvendelse. Analyserne blev foretaget på baggrund af
gamle landbrugsstatistikker fra Danmark samt for Nordtyskland, idet Nord-
tyskland omkring år 1900 omfattede arealer, der nu er danske. Desuden ind-
går for kvælstof målte rodzonekoncentrationer, der stammer fra landbrugs-
forsøg, hvor landbrugspraksis minder om den, der var gældende omkring år
1900. Tilførslen af fosfor fra landskabet til overfladevandet, herunder land-
brugsaktiviteter, blev anslået for de vigtigste kilder: jordafvanding, landind-
vinding, græssende dyr, jorderosion og opbevaring af gødning.
Desuden indgik en detaljeret analyse af bidraget af næringsstoffer fra byer
(punktkilder). Denne analyse estimerede kvælstof samt fosfortilførsel fra by-
erne.
Den nationale kvælstofmodel (NMN) blev anvendt til at modellere koncen-
trationen og transporten af kvælstof til havet. Modellen kræver geografisk di-
stribueret input af klimavariabler, såsom temperatur og vandafstrømning, til-
førsel af kvælstof fra rodzonen og punktkilder, data om grundvandets dybde
og mængden samt placeringen af vådområder, søer og vandløb, for at vurdere
den kvælstofretention, der forekommer i overfladevandsystemet.
Fosforanalysen anvendte ”baggrunds-” eller ”naturlig” -fosforkoncentration
i vandløbsvandet, hvortil de relevante landbrugs- og punktkilder blev tilføjet,
og tilbageholdelse i søer blev trukket fra. Transport af fosfor i oplandet blev
simuleret ved brug af den samme vandafstrømning som for kvælstofmodel-
leringen, men modellen som sådan er meget enklere end kvælstofmodellen
grundet den måde, hvorpå fosfor agerer i miljøet.
Analysen viste blandt andet, at vejrforholdene i år 1900 var koldere og tørrere
end i dag, mere jord blev anvendt til landbrug, men mindre blev drænet, og
14
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
især på grund af en større andel af vådområder var retentionen af nærings-
stoffer i landskabet højere. Konklusionerne fra kapitlerne om klima, hydro-
logi, næringsstoftilførsel og næringstoftransport til havet gengivet nedenfor
stammer fra syntesen, kapitel 11.
Klima
Klimaet var koldere og tørrere omkring år 1900, end det er i dag. Den anslåede
gennemsnitlige årlige nedbør omkring år 1900 var ca. 60 mm, eller 7 % lavere
end i dag. Digitaliserede klimadata fra omkring år 1900, herunder temperatur,
vind og nedbør, blev brugt til at beregne daglige værdier af bias-korrigeret
nedbør. Korrektionsberegningerne blev evalueret ved at sammenligne simu-
leret og målt afstrømning for perioden 1917-1950. På landsplan indikerede en
vandbalancefejl på 3 % rimelige estimater, men de regionale variationer i
denne fejl viste sig at være for store. For at opnå rumlige estimater af korri-
geret nedbør for 1890-1910 blev der i stedet defineret månedlige delta change
klimafaktorer, som blev beregnet ved at sammenholde korrektionsfaktorer
for 1890-1910 og referenceperioden 1989-2010. Det blev antaget, at disse fak-
torer er regionalt repræsentative. De nationale klimafaktorer blev herefter an-
vendt til at korrigere daglig nedbør for referenceperioden, idet det blev anta-
get, at den rent klimatiske regionale nedbørfordeling for denne periode er den
samme som for nedbør omkring år 1900. For at kunne modellere nitratud-
vaskning skal den globale ind- og udstråling og potentielle evapotranspira-
tion beregnes. Ved at anvende de målte minimums- og maksimumstempera-
turer for 1890-1950 kan den globale ind- og udstråling og potentielle
evapotranspiration beregnes og anvendes i simuleringen af nitratudvaskning
i denne periode. Den modellerede globale ind- og udstråling og potentielle
evapotranspiration omkring år 1900 stemmer godt overens med værdier målt
på Foulum fra 1987-2013.
Hydrologi
Den samlede vandafstrømning beregnet med en hydrologisk model (DK-mo-
dellen) baseret på disse nedbørsestimater og ændringer i dræntætheden var i
gennemsnit på 292 mm/år for perioden 1890-1910 sammenlignet med 333
mm/år for den nuværende periode (1990-2010) (kapitel 3). Efterfølgende er
vandafstrømningen genberegnet ved brug af en “delta change”-tilgang, hvor-
ved den samlede vandafstrømning opgøres til 335 mm/år for den nuværende
periode og 297 mm/år for tiden omkring år 1900 (kapitel 8). Det betyder, at
den årlige vandafstrømning i gennemsnit var ca. 11 % lavere omkring år 1900,
end den er i nutiden. Ændringen i vandafstrømningen mellem de to perioder
afspejler i vid udstrækning ændringerne i nedbøren, men forstærkes i nogle
områder af den lavere dræningstæthed i den historiske periode. Den bereg-
nede ændring i vandafstrømning i forhold til tiden omkring år 1900 var mel-
lem 0-20 % for det meste af landet, hvilket er i overensstemmelse med analy-
sen af lange tidsserier for vandafstrømning, der blev præsenteret i Jensen (ed.)
(2017). For den vestlige del af Sjælland indikerer observerede data imidlertid
en øget vandafstrømning mellem de to perioder på ca. 30 %, mens modelsi-
muleringen resulterede i et fald i vandafstrømningen på ca. 5 %. Den an-
vendte metode indebærer, at den samlede vandafstrømning forventes at af-
spejle forholdene omkring år 1900, hvorimod metoden ikke kan forventes at
reproducere de lokale forhold på det pågældende tidspunkt.
15
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Næringsstoftilførsel
De næringsstoffer, der nåede havet fra land, stammede hovedsagelig fra land-
brugsaktiviteter og bebyggelser over hele landet omkring år 1900. Derudover
bidrog afstrømning fra erosion langs vandløbene til næringsindholdet i fersk-
vand, især fosforindholdet.
Landbrug
Landbrugsarealet steg dramatisk i sidste halvdel af 1800-tallet og tegnede sig
for omkring tre fjerdedele af det område, der var under dansk administration
omkring år 1900. Afgrødeproduktionen afveg betydeligt fra det nuværende
landbrug for stort set alle vækstfaktorer: ringere afgrødesorter, højere
ukrudtstryk, manglende kemisk plantebeskyttelse og ringere næringsstoffor-
syning til planter, herunder fravær af mineralsk gødning. De vigtigste kilder
til næringsstoffer var fast husdyrgødning, gylle og kvælstoffiksering ved
bælgfrugter. Antallet og kategorierne af husdyr samt landbrugsstruktur og
forvaltningspraksis omkring år 1900 afveg også fra praksis i dag.
Sognestatistikker fra omkring år 1900 for arealet under nuværende dansk ad-
ministration blev samlet i otte kategorier (vinter- og forårsafgrøder, græs, ro-
dafgrøder, brak, natur og skov). Hver kategori fik tilknyttet en rodzonekon-
centration af kvælstof. For landbrugskategorierne blev kvælstofkoncentratio-
nen i rodzonen fastsat med baggrund i data fra undersøgelser af økologisk
landbrug, idet disse forudsættes at kunne anvendes som proxy for rodzone-
koncentrationen omkring år 1900. For de resterende kategorier blev rodzone-
koncentrationen fastsat ud fra litteraturdata. Disse værdier blev anvendt i
kvælstofmodelleringen. Beregningen af den arealvægtede gennemsnitlige
kvælstofkoncentration for arealer anvendt til landbrug (78 % af landarealet)
resulterede i en værdi på 12 mg N/l, mens værdien for hele landarealet var
9,6 mg N/l i rodzoneudvaskning (uorganisk kvælstof).
Vurderingen af fosfortilførslen fra landbrugskilder tog hensyn til faktorer
som dræning, landindvinding, græssende dyr, jorderosion og opbevaring af
gødning. Der var en del usikkerhed tilknyttet disse faktorer, da de var van-
skelige at fastslå, hvilket førte til et anslået interval for fosfortilførslen fra 70-
200 ton P årligt omkring år 1900.
Punktkilder
Kloaksystemerne blev i stigende grad anlagt i byerne, men stadig uden spil-
devandsrensning omkring år 1900. Byerne var derfor vigtige punktkilder om-
kring år 1900, med 4.261 ton N/år og 764 ton P/år udledt med ekskrementer
fra mennesker og dyr samt med industrielt spildevand. Resultaterne viser, at
størstedelen af næringsstofferne fra punktkilder udledtes direkte til det mod-
tagende vand (55 %), men tilførslerne fra lossepladser (20 %) og landbrugsjord
(25 %) var også betydelige. Det samlede bidrag fra indirekte og direkte punkt-
kilder til vand blev anslået til 471 ton P, ca. 65–70 % af den nutidige værdi (704
ton P, gennemsnit 2014-2018) og 2531 ton N, ca. 47% af nutidens TN-punkt-
kilder (5.400 ton N, 2020).
Andre kilder, herunder baggrundsnæringsstofkoncentration
Kvælstoftilførsel fra atmosfæren blev estimeret ved at multiplicere EMEP-si-
muleringer for år 2000 med 0,3 (Jensen (ed.), 2017). Organisk kvælstof fra
landskabskilder og overfladevandskilder blev medtaget for at kunne beregne
de samlede kvælstofkoncentrationer. Skøn baseret på litteraturundersøgelser
antager, at den organiske kvælstofkoncentration omkring år 1900 lå ca. 20 %
16
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
under det nuværende niveau. Desuden antages det, at den nuværende geo-
grafiske fordeling af organisk kvælstof er gyldig for tiden omkring år 1900.
En gennemgang af litteraturen og målinger fra stort set uforstyrrede vandløb
bruges til at estimere baggrundskoncentrationen af TP i det vand, der strøm-
mer til havet. En arealvægtet medianværdi på 0,052 mg/l TP blev anslået.
Næringsstoftransport til havet
Kvælstof siver gennem jorden og når grundvandet, hvor der under iltfrie for-
hold kan ske reduktion i kvælstofindholdet (retention), før det resterende
kvælstof ender i overfladevand (vådområder, søer, vandløb). Den nationale
kvælstofmodel simulerer transport og retention i grundvandet baseret på
vandafstrømning og kvælstofudvaskning. Overfladevandskomponenten be-
regner kvælstofretentionen i vådområder, vandløb og søer, samtidig med at
der tages højde for punktkildetilførsler, atmosfæriske tilførsler og bidraget fra
organisk kvælstof. Landskabsændringer fra år 1900 og nu blev håndteret ved
at ændre de nuværende landskabskort baseret på forskellige informationskil-
der om det tidligere landskab relateret til floder og søer. For vådområder blev
der anvendt et andet kort.
Fosforanalysen anvendte ”baggrunds”- eller ”naturlig” -koncentration af fos-
for i vandløbsvandet, hvortil de relevante landbrugs- og punktkilder blev til-
føjet og retention i søer trukket fra. Transport af fosfor gennem oplandet blev
simuleret ved brug af den samme vandafstrømning som i kvælstofmodelle-
ringen.
Den absolutte mængde af kvælstof, der fjernes ved retention, viste sig at være
højere i den nuværende periode (28.000 ton N) end i 1900-talsperioden (26.000
ton N), hvilket skyldes den større kvælstofbelastning i dag. Den relative kvæl-
stofretention var imidlertid højere omkring år 1900, idet 43 % af belastningen
blev fjernet sammenlignet med 33 % i den nuværende periode.
Den totale kvælstoftilførsel (TN) er modelleret til at være ca. 36.000 ton N/år
omkring år 1900, hvilket er ca. 40 % mindre end i den nuværende periode.
Kvælstofkoncentrationen er modelleret til at være omkring 2,8 mg N/l om-
kring år 1900 sammenlignet med 4,1 mg N/l i den nuværende periode. Den
nationale kvælstofmodel giver regionale resultater, som anvendes til at esti-
mere regionale kvælstoftilførsler samt kvælstofkoncentrationer i det vand,
der strømmer til havet, omkring år 1900.
Den gennemsnitlige fosforkoncentration i vand fra vandløb omkring år 1900
blev anslået til at være 0,062–0,075 mg P/l, svarende til 60–70 % af den nuvæ-
rende TP-koncentration (0,1 mg P/l). TP-værdierne er fastsat for ni geografi-
ske regioner i hele landet. Den samlede TP-belastning til havet, herunder bag-
grundsbelastning, andre diffuse kilder, indirekte og direkte punktkilder og
fratrukket fosforretention i søer, blev anslået til 1.200-1.340 ton P, 60–65 % af
(eller 35-40 % mindre end) nutidens fosforbelastning (2,021 ton).
Usikkerhed
Arbejdet med en periode for 120 år siden gør naturligvis de fleste aspekter af
opgørelsen af den nationale kvælstof- og fosforbelastning mere usikre, end
når man opgør den for den nuværende periode. En grundig analyse af usik-
kerheder på datalag, variable, model- og modelantagelser var ikke en del af
17
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
nærværende undersøgelse; dog er der foretaget nogle overvejelser om usik-
kerhed og følsomhed (effekten af en given parameter på resultatet).
De fleste af de parametre, der bruges til at estimere kvælstofbelastninger om-
kring år 1900, anses for at have en ”middel” usikkerhed (på en tretrinsskala
fra lav til høj). Usikkerheden på kvælstofbelastningerne er påvirket af en
række faktorer, hvor de vigtigste er usikkerheden omkring estimeret nedbør,
afstrømning, rodzonekoncentrationen af kvælstof og retention af kvælstof i
overflade- og grundvand.
Overordnet anses det modelkoncept, der er brugt til at beregne kvælstofbe-
lastningen i år 1900, som relativt robust og den overordnede usikkerhed på
nationalt plan som acceptabel. Usikkerheden stiger dog med en faldende geo-
grafisk og tidslig skala.
De fleste af de parametre, der bruges til at estimere fosforbelastninger, betrag-
tes som ”middel” til ”meget”" usikre. Usikkerheden på fosforbelastninger er
især påvirket af usikkerheden på estimaterne for nedbør, afstrømning, P-til-
førsel fra punktkilder og baggrundskoncentrationen af TP. På trods af de
mange usikkerheder menes resultaterne af denne undersøgelse at være det
bedst mulige estimat for fosforbelastningen i år 1900. Yderligere er resultatet
af denne undersøgelse understøttet af historiske målinger i søer, der også fin-
der historiske TP-koncentrationer, der er lavere end i dag, men betydeligt hø-
jere end baggrundskoncentrationen.
Perspektivering
Mange af de europæiske undersøgelser, der sammenlignes med nærværende
undersøgelse, rapporterer om kvælstofkoncentrationer omkring år 1900, der
er betydeligt lavere end i nærværende undersøgelse. Årsagerne hertil er for-
mentlig forskelle i landskabet, arealanvendelsen, landbrugspraksis og af-
strømning mellem de undersøgte områder og Danmark. Det afspejler for-
mentlig også, i hvilken grad landbrugspraksis og næringsstofdynamik indgår
i undersøgelsen. I nærværende undersøgelse beregnes rodzoneudvaskning
for år 1900, og kvælstoffiksering, hovedkilden til kvælstof i dansk landbrug i
år 1900, tages i betragtning, hvilket ikke er tilfældet i de fleste andre undersø-
gelser.
18
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Introduction
The current report aims to describe the nutrient load from land to sea around
the year 1900 and builds upon a previous project in which different factors
influencing the nitrogen content from the field to the sea around the year 1900
were investigated (Jensen (ed.), 2017)
1
. The main factors investigated in 2017
were climate, hydrology, land use, agricultural practices, drainage and land-
scape effects (retention). The analysis showed, among other things, that it was
colder and drier in the year 1900, more land was in agricultural use but less
was tile drained, and due to especially a larger proportion of wetlands the
retention of nutrients in the landscape was higher. Combined, this resulted in
a best estimate at the time of the nitrogen concentration in the water running
to the sea of 1-2 mg N/l.
The aim of the current project was to further develop this analysis and expand
it to also include phosphorus, enabling the calculation of a spatial distributed
estimate of the nitrogen and phosphorus concentrations from the root zone to
the sea and the resulting load to the sea.
To be able to do this, the datasets on climate and hydrology were expanded
and the runoff was modelled using the national water resources model (DK-
model).
The agricultural practices and land use in different areas of the country were
analysed in detail, and the root zone-concentration of nitrogen was deter-
mined for different land use categories. Furthermore, a detailed analysis of
the contribution of nutrients from cities (point-sources) was included.
To model the concentration and transport of nitrogen in the freshwater load
to the sea, the national nitrogen model (NMN) was used. The model requires
geographically distributed input of climate variables such as temperature and
water runoff, the input of nitrogen from the root zone and point sources as
well as data on the depth of groundwater and the amount and location of
wetlands, lakes and streams in order to estimate the retention occurring in the
surface water system.
The phosphorus analysis focused on assessing the total phosphorus concen-
trations and used an approach of a ”background” or ”nature” concentration
level on top of which the relevant additional agricultural and point sources
were added and the retention in lakes was subtracted. The transport and rout-
ing of phosphorus through the catchments were simulated using the same
water discharge as for the nitrogen modelling, but the model as such is much
simpler than the nitrogen model because of the way phosphorus behaves in
the environment
In the following chapters (1-7) each of the different factors influencing the ni-
trogen concentrations around the year 1900 is described and discussed and
finally incorporated into the NNM model (chapter 8) to model the nitrogen
concentration and load to the sea.
The method used for estimation of the phosphorus concentration (chapter 9)
differs from the approach used for nitrogen because the two nutrients and
their behaviour in the environment are significantly different. Phosphorus is
19
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
discussed in a separate chapter, but the analysis is based on the same climate,
hydrology, landscape data and point source analysis as for nitrogen.
In the last common chapters (10-12), uncertainties of the chosen approaches,
a synthesis of results, discussion and perspectives are presented.
Report structure
Chapter 1 and 2: Describe the climatic conditions, temperature and precipita-
tion around the year 1900 as well as the evapotranspiration and global radia-
tion at that time. The data are used as input to the models calculating water-
runoff (DK model) and nitrogen-load (NNM-model).
Chapter 3: Describes the water run-off around the year 1900 modelled with
the DK model. Stream discharge is used as input to the NMN model described
in chapter 6-8 and to calculate P transport in chapter 9.
Chapter 4: Describes the nutrient input from the larger Danish cities and re-
lated activities around the year 1900 (point sources). The results of the analysis
are used as input to the NNM (chapter 6-8) and to calculate the P-transport in
chapter 9. Inland point sources are added to the surface water module of the
NNM, along with N and P from other sources and retention is calculated on
the total load. Coastal point sources are treated as loads directly to the sea and
hence there is no retention on coastal/direct loads.
Chapter 5. Describes the land use and agricultural practices around the year
1900 and how the root zone concentration of nitrogen around that time is es-
timated. The results are used as input to the NMN model chapter 6-8.
Chapter 6: Describes the NMN model concept, and how the model is adjusted
to represent conditions around year 1900 based on the results from chapter 1-
5. Further, the input parameters atmospheric nitrogen deposition and the
level of organic nitrogen in freshwater are estimated.
Chapter 7: Describes changes in the landscape since the year 1900, how it af-
fects surface water retention and how the retention is calculated in the differ-
ent types of surface water systems (lakes, streams and wetlands)
Chapter 8: Incorporates the results from chapter 1-7 into the NNM to calculate
the concentration and transport of nitrogen to the sea.
Chapter 9: Describes the phosphorus concentrations and transport around the
year 1900 and how the concentrations and loads to the sea are estimated.
Chapter 10: Describes the uncertainty analysis for the various factors and
methods used.
Chapter 11: Presents a synthesis extracting the main results and providing an
overall overview.
Chapter 12: Presents the perspectives and conclusion of the study.
1
Jensen
(ed.), 2017. Estimation of nitrogen concentrations from root zone to marine areas around
the year 1900. Scientific Report from DCE No. 241. Aarhus University, Danish Centre for Envi-
ronment and Energy.
20
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
1 Climate around year 1900: Temperature
and precipitation
Author: Flemming Vejen
1
Quality assurance: Torben Schmidt
1
1
DMI
- Danish Meteorological Institute
Abstract
Purpose:
In the present chapter, climate data for the year 1900 and a valida-
tion period 1917-1950 are established. Spatially distributed temperature and
bias-corrected precipitation are used as input to the calculation of global radi-
ation and evaporation in chapter 2 and of the modelling discharge to the sea
in chapter 3.
Materials and Methods:
Based on digitised observations, spatially distributed
meteorological variables are provided to the hydrological modelling of stream
discharge. Rain gauge observations were corrected for shelter effects and biases
caused by wind and wetting loss. In order to calculate the bias correction, a
number of simplifications and assumptions are necessary due to, for instance,
the facts that rain rate and precipitation type are not measured. The corrected
precipitation was validated by comparing the simulated and observed water
discharge for the validation period 1917-1950 (see also chapter 3).
Results and discussion:
A general water balance error of 3% for the verification
period 1917-1950 suggests that the national values for corrected precipitation
are realistic. However, large variations in the fit between observed and simu-
lated discharge were seen at a regional scale. Since the true wind speed and
shelter conditions at station level are unknown, various approaches for calcula-
tion of corrected precipitation were tested but with no reasonable results. This
uncertainty could not be addressed within the current project; thus, a delta
change approach was applied instead. Based on this, approach the spatial dis-
tribution of the amount of rainfall in the reference period 1989-2010 was trans-
ferred to the period 1890-1910 by using monthly climate factors (delta change)
derived from the relations in corrected rainfall between the two periods.
Verification of climate data shows that monthly and annual values of temper-
ature and precipitation are reasonably well in line with official climate values,
but with a larger uncertainty at local level, however. It is difficult to verify
wind speed, but it is known that manual observation can be a source of un-
certainty. In a parallel project, manual wind speed observations have been
corrected for homogeneity problems, and a trend adjustment has been ap-
plied using geostrophic wind speed.
Conclusions:
Based on uncorrected wind data, delta change was calculated
to 0.931 per year (773 mm yearly corrected rainfall which is 57 mm less than
during 1989-2010). The corrected wind data show that the uncorrected wind
around year 1900 was probably overestimated, resulting in a 2.7% too large
value of delta change corresponding to 20 mm/yr., a reasonable level given
the different assumptions and uncertainties.
21
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
1.1
Introduction
The objective of this chapter is to provide this project with spatially distrib-
uted estimates of required meteorological variables to support the hydrolog-
ical modelling of water discharge. The detailed content of the chapter has been
previously published in a scientific brief from DCE, AU (Vejen, 2021). As no
discharge data are available before 1917, it is not possible to validate the mod-
elled discharge around year 1900. Therefore, the period 1917-1950 is chosen
as validation period.
A limited number of digitised daily or monthly data in the database of the
Danish Meteorological Institute (DMI) has motivated digitisation of a large
number of climate data. The specific goal of the climate data activity was:
1) to collect and digitise historical meteorological data from 1890 to 1950.
2) to develop an approach for bias correction of historical rain gauge data.
3) to establish data series of bias-corrected rainfall, air temperature and wind
speed for a suitable number of inland stations evenly distributed in Den-
mark, also with data for Southern Jutland before 1920.
4) to validate the uncertainty of calculated climate variables around year 1900
and to compare the results with official monthly and yearly climate values
from DMI.
Based on these data, 10×10 km
2
fields of bias-adjusted precipitation and 20×20
km
2
temperature fields are established for calculation of evaporation, which
acts as climate data input to the hydrological model run for the period 1917-
1950 for evaluation of the water balance (see chapter 3). Finally, the goal is to
refine or develop a novel approach for bias correction of rain gauge under-
catch if the modelled water balance is not sufficiently accurate.
Measurement of precipitation is recognized as a challenging task, and uncer-
tainties regarding rain gauge measurements have been widely reported. Pre-
cipitation measurements are affected by systematic errors, which lead to un-
derestimation of the actual precipitation for rain (e.g., Sevruk, 1979) and, es-
pecially, for snow (e.g., Groisman and Legates, 1994; Yang et al., 1995). Nu-
merous field experiments have shown that wind speed is the most important
environmental factor for this undercatch, or bias, of rain gauge measurements
(e.g., Sevruk and Hamon, 1984).
The interaction between a rain gauge, the wind flow and liquid or solid par-
ticles falling through the air is complex. The design and geometry of a rain
gauge are of great importance for its aerodynamic properties and ability to
measure precipitation, e.g., demonstrated by wind tunnel experiments
(Nespor, 1996), and modelling of the air flow around rain gauges also showed
systematic differences related to gauge geometry (e.g., Colli et al, 2018).
Precipitation measured by a rain gauge is subject to other systematic errors
such as evaporation and wetting losses, the magnitude of which depends on
the type of rain gauge (Sevruk and Hamon, 1984).
Precipitation is an important parameter in hydrological modelling and stud-
ies of the water cycle, and sustainable water balance monitoring requires
availability of accurate precipitation data. It is necessary to apply a correction
for the different losses in measured precipitation to acquire reliable calcula-
tions of the water balance (e.g., Plauborg et al., 2002).
22
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Organised by the World Meteorological Organization (WMO), great efforts
are made to conduct field tests and establish models for correcting sources of
error in measured precipitation. A comprehensive correction model was de-
veloped, which elegantly combines sub-models for rain, sleet and snow in the
same equation (Allerup et al., 1997) by which bias correction is conducted if
information is available about wind speed, rain intensity, dry air temperature
and the proportion of precipitation fallen as snow.
An important part of the work is to apply bias correction to the observed pre-
cipitation, but several data required for this were not available around year
1900, such as shelter information at rain gauge stations and rain rate. Wind
speed was manually observed, and the observation frequency was low. A
method was developed to overcome these challenges. The corrected precipi-
tation estimates included in the hydrological modelling are based on a range
of meteorological data, including manually recorded wind speed. It was pos-
sible to correct the wind speed for a number of errors and the improved wind
data was finally used to examine the sensitivity of the corrected precipitation
estimates.
1.2
Material and methods
1.2.1 Data
Monthly wind speed and daily temperature data for the period 1890-1950, and
daily rain gauge data for the period 1913-1950, were published in analogue form
in monthly or annual weather reports (DMI 1890-1950), and it is assumed that
these data were subject to quality assurance before publication, even though no
documentation for methodology is found. Opposite to this, rainfall data from
1890 to 1913 are available in the form of original observer reports (Rigsarkivet).
Experience at DMI shows that these data have not been subject to quality con-
trol. Quality assurance of such a huge amount of data is an extremely compre-
hensive task, which is beyond the scope of this project, so only simple checks
have been carried out, i.e., values exceeding certain thresholds are automati-
cally flagged as suspicious or in error. It is assumed that after this simple quality
control, these data have required quality, i.e., observation errors and scanning
mistakes have been identified and flagged.
Since about 20% of the stations are not registered in DMI’s metadata database,
the approximate position of the stations could only be determined from the sta-
tion name. There is therefore minor uncertainty in the position of some of the
stations. It is considered that this is acceptable as the alternative would be ex-
clusion of data and increased uncertainty of estimated precipitation fields.
Daily values of maximum and minimum temperature and measured precipita-
tion at rain gauge stations have been digitized. The framework of the project
has not allowed digitization of the daily wind speed, but monthly mean values
have been digitized.
While the precipitation stations are evenly distributed, the wind speed sta-
tions are primarily coastal stations. Until 1913, there were only eight stations
where wind speed was measured, including one inland station. Later, the
number increased to 11-12, of which approx. six inland stations are available.
See example for 1900 and 1935 in Figure 1.1. Stations with temperature meas-
urements are evenly distributed with both coastal and inland stations, and the
number of stations is identical for almost the entire period – 15-17 – with up
23
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0026.png
to nine inland stations (Figure 1.1). The number of digitized precipitation sta-
tions increases from approx. 80 in 1890 to about 130 in 1913, including stations
from Southern Jutland, but already in 1919 there were about 270 stations, a
number that remained almost constant until 1950. Figure 1.2 shows the spatial
distribution of precipitation stations in the two periods 1890-1910 and 1914-
1950. In 1890-1910, there is a lack of precipitation data on the northern part of
Jutland, and there are regions with a relatively sparse rain gauge network.
The rain gauge density is much higher in the period 1914-1950, but the net-
work is slightly in-homogeneous with smaller areas of lower coverage.
Figure 1.1.
Stations with wind speed and temperature in 1900 (blue) and 1935 (red).
Figure 1.2.
Stations with precipitation in the period 1890-1910 (left) and 1914-1950 (right). Maps are from Google Earth.
Up to the 1910s, the amount of precipitation was measured with the so-called
Fjord rain gauge for measurement of rain, and for snow measurements, a zinc
bucket was used (hereafter called snow gauge). In the period 1910-1925, the
Fjord gauge was gradually replaced the Danish Hellmann gauge (Brandt,
1994), which functioned as the cornerstone of the DMI’s rainfall network until
24
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0027.png
2011. For the Hellmann gauge the installation height was 1.5 m (Brandt, 1994),
while the Fjord gauge was installed with its orifice placed at 2 m height above
ground level (DMI, 1875). It is not known for how long time this practice con-
tinued, but presumably it gradually stopped during the transition to the Hell-
mann gauge.
Wind speed (V) was visually observed throughout the period. Until the end of
1910, it was indicated by the so-called Danish Land Scale, which has seven lev-
els (0-6) (Kristensen and Frydendal, 1991). Hereafter, the well-known Beaufort
Scale (level 0-12) was used. At DMI in Copenhagen, wind speed was given in
m/s for the period 1890-1893. When correcting for wind-induced bias in pre-
cipitation measurement, V must be given in m/s. Therefore, a method was de-
veloped for conversion of Danish Land Scale and Beaufort to m/s (Vejen, 2021).
1.2.2 Adjustment, correction, and control of data
Several aspects and parameters need to be considered before the early obser-
vations are used to calculate the needed wind and temperature fields and pre-
cipitation field for the hydrological modelling. The methods and data used
are briefly described in the following. The aspects considered in this work are
listed in Table 1.1 where the topic and overall approach are mentioned. De-
tailed explanations are found in Vejen (2021).
Table 1.1.
List of issues and overall approaches to tackle shortcomings of climatological data from the period 1890-1950. Fur-
ther details in Vejen (2021).
Topic
Wind
Conversion of wind speed measurements to m/s
Control of wind speed observation data
Correction for shelter conditions
Rain
Estimation of wetting and evaporation loss
Estimation of rain rate
Calculation of precipitation type
Bias correction of rain gauge observations
Correction and control of rainfall measurements for wind-in-
duced bias
Evaporation and wetting loss according to the Hellman gauge
(Allerup and Madsen, 1979, 1980)
Based on earlier climatological measurements
Model based on air temperature
Argued not to cause too high uncertainty
Comprehensive model used in Denmark (Allerup et al., 1997)
Conversion model developed
Homogeneity checks with observations from more recent times
Assuming similar shelter practice in 1890-1950 as today
Approach
Adjustment of monthly wind speed to reflect precipitation days Correction based on a monthly correction factor
For correction of precipitation measurements, in Denmark, a model combin-
ing the correction of solid, mixed and liquid precipitation into one expression
is used (Allerup et al., 1997), where the correction factor K
α
for the precipita-
tion type
α
is given by the following correction model:
����
=
���� ⋅ ����
⋅
⋅
⋅ ⋅
+ (1
− ����) ⋅ ����
⋅
⋅
⋅ ⋅
Here,
α
= index indicating the proportion of precipitation fallen as snow (0=rain,
1=snow),
V
= wind speed at gauge level,
T
= air temperature,
I
= rain rate,
β
0
,
β
1
,
β
2
,
β
3
= empirical constants for snow (Allerup et al., 1997) and
γ
0,
γ
1,
γ
2,
γ
3
, c
=
empirical constants for rain (Allerup and Madsen, 1980; Førland et al., 1996). The
corrected precipitation amount,
P
c
, is in principle given by
P
c
=
K
α
P
m
where
P
m
=
measured precipitation, but also a wetting loss must be included in the correc-
tion (further details in Vejen et al., 2014). The design of a rain gauge is of im-
portance for its aerodynamic properties (Sevruk and Klemm, 1989; Sevruk et al.,
25
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
1989). The design of the Fjord and snow gauges is different from the Hellmann
gauge. Despite the fact that their measuring ability may differs, it is assumed that
the correction model can be used for both gauges.
For the Hellmann gauge, the wetting loss is a function of season and precipi-
tation type (Allerup and Madsen, 1979, 1980; Vejen et al., 2000) and amounts to
approx. 5% per year. The evaporation loss is negligible (1.5-2.0 mm/yr). The
undercatch for the Fjord gauge due to wetting and evaporation is assumed to
be the same as for Hellmann, although its design probably causes a larger
evaporation loss than that of Hellmann.
Correction of precipitation requires input on wind speed and temperature at
gauge level during precipitation, on rain rate, wetting loss and precipitation
type, which is illustrated in the process diagram in Figure 1.3. For the period
1890-1950, there are certain data limitations. The wind speed is manually ob-
served, and only monthly average values are digitally available. Air temper-
ature is digitally available only as a maximum,
T
max
, and a minimum temper-
ature,
T
min
. No information on rain rate and precipitation type is available.
Seasonal-dependent climatological values of rain rate, which are based on
measurements of precipitation in Denmark over the period 1959-1974 at four
stations (Madsen and Allerup, pers. comm.), are used and assumed repre-
sentative of 1890-1950.
Daily average temperature,
T
avg
, is calculated using daily observations of
T
min
and
T
max
and is assumed to represent conditions during precipitation. A widely
used model for the determination of precipitation type,
t,
uses the air tempera-
ture, or in our case daily average temperature,
T
avg
, as an indicator of precipita-
tion type, i.e.,
t
= snow if
T
avg
≤
0 °C,
t
= rain if
T
avg
> 2 °C, and otherwise sleet,
although this method may cause bias between estimated and observed
t
(e.g.,
Feiccabrino et al., 2015). A more realistic probability function for
t
can be ob-
tained by including air humidity (e.g., Harder and Pomeroy, 2013), but this pa-
rameter is not available in the data set. Snow index
α
is calculated from
T
avg
by
α=1
for snow,
α=0
for rain and
α=-0.5
T
avg
+1
for mixed precipitation.
Manually observed values of
V
are transformed to m/s based on information
found in WMO (1970) Kaufeld (1981) and Kristensen and Frydendal (1991).
From analyses of the general level of wind speed for manual observations be-
fore and automatic measurements after 1950, it seems reasonable to assume
that V represents 10 m above ground level.
V
is adjusted down to the height
of the rain gauge using the logarithmic wind law as recommended by the
WMO (2008). As there is no information on the properties of the ground sur-
face, the effective roughness length is used in the wind law in the form of a
general value of 0.25 like that used in Refsgaard et al. (2011).
Monthly wind speed does not reflect the conditions during precipitation pe-
riods since it includes dry days and days with stable weather and low wind
speed. Correction for this bias is applied using a seasonal-dependent adjust-
ment factor (Vejen, 2021).
It is common practice to adjust the wind speed for local shelter conditions.
The shelter adjusted wind speed,
V
shlt
, is found by adjusting the wind speed
at gauge height
h, V
h
, by the expression
V
shlt
=
λ
V
h
, where
λ
is a shelter correction
factor given by
λ
=1-c·η (Sevruk, 1988; WMO 2008). Here,
η
= the average height
angle to the top of the shelter given in degrees and
c
= a constant (c=0.024). At
26
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0029.png
DMI, it is practice to use a weighted shelter index,
η
wgt
, which is calculated by
weighting height angles with the statistical frequency of winds from eight di-
rections. Thus
η
wgt
replaces
η
in the expression for
λ
.
Figure 1.3.
Process diagram for
calculation of grid values of cor-
rected precipitation, which is
used as input to the hydrological
modelling of chapter 3.
P
m
=
measured precipitation,
T
max
and
T
min
= maximum and minimum
temperature,
V
(manual) = manu-
ally observed wind speed, wetting
= wetting loss, type
α
= index in-
dicating precipitation type,
T
avg
=
daily average temperature,
T
(grid) = gridded temperature,
V
(m/s, 1.5 m) = wind speed trans-
formed to m/s and 1.5 m level,
V(shlt) = shelter corrected wind
speed,
V
(grid) = gridded wind
speed,
η
wgt
= weighted shelter in-
dex,
P
c
= corrected precipitation,
P
c
(grid) = gridded corrected pre-
cipitation,
η
= shelter index,
P
∆corr
= precipitation corrected using cli-
mate factors. See section 1.2.2
and 1.3.4 to 1.3.6 for explanation
DMI first began to measure height angles at rain gauge stations in the 1960s.
Since shelter conditions can have great significance for the local wind speed
and thus the correction level, it is necessary to make assumptions about height
angles for the period 1890-1950. Already in the 19th century, the wind's effect
on precipitation measurements and the importance of shelter were well
known (Brandt, 1994). It is assumed that in the period 1890-1950, the same
practice was used as today to ensure good shelter conditions at rain gauge
stations (neither too open nor overprotected).
1.2.3 Calculation of 20×20 km
2
fields for wind speed and temperature
For consistency reasons, the same interpolation principles are used as in
KlimagridDK where the weighting is calculated in relation to
1/r
2
. Here,
r
=
the distance between a grid point and a weather station (Scharling, 1999).
For the period 1890-1950, however, the number of stations with wind and
temperature data is quite limited, and especially for wind there are very few
inland stations. Since calculation of the spatial distribution of
V
and
T
from so
few stations is highly uncertain, the interpolated values of
T
and
V
are ad-
justed with a method inspired by Olesen et al. (2000), in which distributed
variables are calculated even when the spatial resolution of the observation
material is limited. The method combines observations with fields that for
each month describe the normal relative spatial variation of
V
and
T.
These
27
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0030.png
fields are incorporated into the interpolation technique by which daily 20×20
km
2
fields of
T,
and monthly values of
V,
are calculated.
The established relative normal fields,
μ,
in a 20×20 km
2
resolution are based
on data for the period 1989-2017. The relative values,
μ(i,j),
are used to adjust
interpolated values of temperature,
T(i,j),
or wind speed,
V(i,j),
in the formula
below given as
F(i,j).
In practice, the interpolated surface is lowered or raised
depending on the pseudo-climatological value in an arbitrary point
(i, j).
Es-
pecially for wind speed, it has the advantage that interpolation, which is
largely based on coastal stations, is forced to lower values inland.
F
(
i
,
j
)
=
μ
(
i
,
j
)
w
g
(
i
,
j
)
F
g
g
=
1
N
N
w
g
(
i
,
j
)
g
=
1
, where the weight
w
g(i,j)
is given by:
w
g
(
i
,
j
)
1
=
2
r
(
i
,
j
)
Here,
w
= a weighting function,
g
= a gauge station
g, N
= number of stations,
F
= measured (or observed) value of wind speed
V,
or temperature
T
at station
g, F(i,j)
= value of
T
or
V
for a grid cell
(i,j),
and
μ(i,j)
= the weighting function
for the relative spatial normal value for grid cell
(i,j).
The weighting for the relative spatial distribution is given by:
μ
(
i
,
j
)
=
w
g
(
i
,
j
)
g
=
1
N
R
g
R
(
i
,
j
)
N
w
g
(
i
,
j
)
g
=
1
Here,
R
g
= the relative climate value at gauge station of temperature
T
or wind
speed
V, R(i,j)
= the relative climate value of
T
or
V
for a grid cell
(i,j),
and
w
= the weighting function previously defined.
It is assumed that the relative spatial distribution based on data from 1989-
2010 is consistent with the period 1890-1950 since the spatial distribution is
determined by physical and meteorological factors as well as terrain condi-
tions that are considered relatively unchanged over the period. Even though
systematic changes in urbanisation and vegetation over the period are seen,
the overall meteorological conditions are assumed to be relatively constant.
However, the use of monthly normal fields for simulating spatial variations
at daily level increases the uncertainty of the interpolated values.
1.2.4 Calculation of 10×10 km
2
fields of observed and bias-corrected
precipitation
Almost the same interpolation method as previously described is used for
wind and temperature. The only difference is that rainfall is not adjusted with
a relative normal field because it is not needed due to the large number of rain
gauge stations. The interpolated field value for precipitation sum,
P,
is given
by:
28
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0031.png
N
P
(
i
,
j
)
=
w
g
(
i
,
j
)
P
g
w
g
(
i
,
j
)
,
g
=
1
g
=
1
1
given by:
w
g
(
i
,
j
)
=
2
.
r
(
i
,
j
)
N
where the weighting function
w
g(i,j)
is
1.3
Results
This section analyses and evaluates the calculations of temperature, wind and
precipitation, and the basic assumptions for bias correction of precipitation.
The evaluation is a challenge as there is no independent data for testing the
daily results. However, official national values of the meteorological variables
(DMI database) are available; thus, the evaluation can be done by examining
whether the interpolation of the meteorological variables can roughly repro-
duce the official monthly and yearly national climate values over the period.
At DMI, official national values of
T
and
P
m
were previously (from the 1950s up
to 2006) calculated as a simple average of station values with data from Jutland
weighted with 7/10 and data from the islands with 3/10, and after 2006 based
on grid interpolation of station data (Cappelen, 2019). Before the 1950s, the
methods used were not published. Since grid values calculated during this pro-
ject are spatially distributed, in contrast to the official national values before
2006, minor differences between official values and grid calculations are ex-
pected. The possibility of assessing the calculations of
P
m
and
T
relative to cli-
mate values is greater than for
V
since
P
m
and
T
values are historically based on
a larger network of stations those of
V,
for example the official normal for
T
for
1886-1925 is based on 30 evenly distributed stations (Det Statistiske Departe-
ment, 1964). The normal, or pseudo normal, for
V
for the periods 1931-1960 and
1961-1990 is based on a relatively limited number of coastal stations (Lysgaard,
1969; Cappelen, 2000), and official national values do not exist.
In the calculation of national averages of the meteorological variables in this
project, grid cells in coastal regions are given lower weight depending on the
fraction of land area.
1.3.1 Evaluation of temperature
While the official values are based on many stations available at that time
(Cappelen, 2019), the grid method is based on a smaller number of stations.
As the calculation method for the two datasets are quite different, the evalua-
tion is used to determine whether this has an impact on the results. Official
national daily values are not available for the period 1890-1950; thus, the grid
estimates of
T
are evaluated monthly. Visual inspection of Figure 1.4 (left)
leaves the expression that there is no obvious bias between official and grid-
ded values, except a slight underestimation of the coldest months by the grid
method. The monthly grid values of
T
are close to the official values. If the
results are inspected carefully, an interesting difference between the two pe-
riods 1890-1919 and 1920-1950 is seen (Table 1.2 and Figure 1.4 (right)). The
grid values before 1920 are biased towards higher values and those after to-
wards slightly lower values. The explanation of the jump in the general tem-
perature level is that the relatively warm southern part of Jutland was not a
part of Denmark before 1920 and therefore not included in the official values,
but it is included in the estimates of grid temperatures.
29
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0032.png
Table 1.2.
Statistics on gridded and official national values of temperature
T
for the two
periods 1890-1919 and 1920-1950; bias of
T
given as
T
bias
=
∑(T
grid
−T
official
) and absolute
ias as
T
bias
=
∑(|T
grid
−T
official
|).
Year
Period
1890-1919
1920-1950
T
grid
7.55
7.73
T
official
7.45
7.79
T
bias
0.098
-0.055
Month
T
absbias
0.263
0.228
R
2
0.998
0.998
Figure 1.4.
Left: National monthly grid values of temperature vs official climate values. Right: National yearly grid values of tem-
perature vs official climate values.
1.3.2 Evaluation of measured precipitation
Figure 1.5 shows monthly and yearly gridded values of measured rainfall com-
pared with official values. The scatter is probably due to differences in calcula-
tion methods. The annual values (Table 1.3) show that the grid values for the
period 1890-1919 marginally underestimate the measured precipitation com-
pared with official values by 0.56 mm per month and with an absolute bias of
2.9 mm.
Table 1.3.
Statistics on gridded and official national values of measured precipitation
P
m
for the two periods 1890-1919 and 1920-1950. Bias of
P
m
given as
P
m(bias)
=
∑(P
m(grid)
–
P
m(official)
) and absolute bias as
P
m(bias)
=
∑(|P
m(grid)
–P
m(official)
|).
Year
Period
1890-1919
1920-1950
P
m(grid)
643.0
670.4
P
m(official)
646.8
659.8
P
m(bias)
-0.56
0.74
Month
P
m(absbias)
2.87
2.91
R
2
0.980
0.978
30
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0033.png
Figure 1.5.
Left: National monthly grid values of measured precipitation vs official climate values. Right: National yearly grid
values of measured precipitation vs official climate values.
1.3.3 Evaluation of wind speed
Like gridded temperatures, gridded wind speed is based on only few stations.
The aim of an evaluation would be to investigate whether the calculation
method results in biases compared with official national values, but this is not
possible due to lack of official monthly or annual wind speeds.
Prior to 1953, practically all wind data were manually observed, and in the
1950s wind speed was measured in m/s only at a few stations, so evaluation
can be done only based on a few long time series of
V.
Evaluation is done
partly by looking at the continuity of
V
at the transition from Beaufort to m/s,
and partly by considering the co-variation and trends in the manual series in
relation to the geostrophic wind velocity,
V
g
.
The inherent uncertainty of manual wind observation is probably the cause of
the homogeneity breaks detected for some of the wind series. An inter-com-
parison of the overall temporal trends in wind speed for the period 1890-1950
showed differences. Some series show increasing wind speed during the pe-
riod, while others demonstrate a decreasing trend, and for many of the series
a homogeneity break is also seen around the transition from manual to auto-
matic measurement. This is a clear indication of the uncertainty of the manual
observations, and adjustment for the homogeneity breaks using classical tech-
niques is probably too uncertain.
Instead, the idea is to use geostrophic wind as an independent source for the
general temporal trends during 1890-1950. After correction of precipitation
and water balance was finished a project funded by DHI
(www.dhigroup.com) made it possible to develop a methodology for homog-
enisation of all wind series in the period 1890-1950.
The method adopts that in Alexandersson et al. (1998) of using geostrophic
wind as an indicator of trends in wind climate. Analyses of
V
g
were conducted
to obtain a picture of the overall trend and to compare trends of
V
g
with the
corresponding Beaufort/Danish Land Scale trends. The method is based on
the assumption that a relation exists between geostrophic wind and the true
31
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0034.png
wind near the ground, even though it is affected by, for instance, stability,
friction and gradient winds (e.g., Luthardt and Hasse, 1981). Geostrophic
wind is not a perfect measure of true wind (Alexandersson et al., 1998), and
V
g
is therefore used mainly for interpretation of trends.
Based on three long air pressure series, daily values of
Vg
were calculated for
the period 1890-2010. Based on monthly values of
Vg,
a polynomial model
was established, which describes, with good approximation, the temporal
trends of
Vg.
Although uncertainty is associated with the calculation of
V
g
,
and certain spatial differences between a point value of
Vg
and wind at indi-
vidual stations would be expected, it is assumed that the model can give a
reasonable impression of the overall trends in wind climate for all stations in
the period 1890-1950. The model is then adapted to the level of wind speed
for each individual wind series to ensure that the model, in addition to adjust-
ment for the general trend errors, also adjusts for the break around the transi-
tion to automatic wind speed measurements.
Reasonable results have been obtained by adjusting all wind series as homo-
geneity breaks and trend errors are significantly reduced without changing
the variability in monthly wind speed. Figure 1.6 shows the monthly values
of
V
for Denmark with and without trend correction for 1890-1950 and 1989-
2010. Because national grid values are not available in the intermediate pe-
riod, representative wind series are shown for stations where
V
has been
measured with anemometer since 1953. The stations chosen are all inland sta-
tions, while the grid estimates for Denmark are averaged over 20×20 km
2
grid
cells. No obvious homogeneity break is seen at the transition from Beaufort to
m/s in the 1950s.
Figure 1.6.
Left: Monthly wind speed for the period 1890-2010, national averages of 20×20 km
2
grid values and selected station
values. Red line: Grid values not corrected for trends. Blue line: Grid values corrected for trends using analyses of geostrophic
wind velocity,
V
g
. Black line: Grid values for the period 1989-2010. Coloured bullets: Station values. Stations are 06030 FSN
Aalborg, 6060 FSN Karup, 6110 FSN Skrydstrup, 6180 Copenhagen Airport. Right: Comparison between averages of monthly
national grid averages before and after correction for trend and homogeneity errors.
It seems from Figure 1.6 that Danish Land Scale (1890-1910) probably overes-
timates and the Beaufort Scale probably underestimates
V,
at least in parts of
the period up to 1950. This underlines the importance of wind adjustments as
the uncertainty of manual wind observations would otherwise deleteriously
affect the bias correction of precipitation, as discussed later.
32
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
1.3.4 Evaluation of shelter assumptions
Of the parameters needed for bias correction of rainfall, the lack of information
on shelter conditions is particularly critical as the wind speed at gauge height
must be adjusted by the shelter correction index,
η
wgt
.
The shelter conditions were generally not known at rain gauge stations until the
1970s. Since shelter can have great significance for the local wind speed and
thus the correction level, it has been necessary to make assumptions about
shelter conditions for the period 1890-1950. Since the wind’s effect on precip-
itation measurements and the importance of shelter were well known already
in the 19th century (Brandt, 1994), it is assumed that in the period 1890-1950
the same practice was used as today to ensure good shelter conditions (neither
too open nor overprotected).
Assumption of a nationwide constant shelter index may result in significant
over- or underestimation of bias-corrected rainfall, locally, regionally or nation-
wide. Streamflow values are calculated by the national water resource model,
the DK model, for several catchments using input of bias-corrected precipitation
data. To get an idea of what would be the most appropriate index, the calculation
of bias-corrected precipitation is iterated over various assumptions about shelter
index,
η
wgt
(see Figure 1.3). The effect on hydrological modelling is then evalu-
ated using streamflow gauge stations, which are available for the evaluation pe-
riod 1917-1950. The water balance error, which is the absolute difference be-
tween calculated and measured streamflow at several streamflow gauge sta-
tions, is calculated for several catchments and is compared for the different shel-
ter index assumptions. The goal is to minimise the water balance error.
A variety of shelter indices have been used, ranging from nationwide values to
more regionally variable indices. It was found that a value of
η
wgt
=12 nation-
wide for the period 1917-1950 led the total observed and simulated discharge
at national scales to agree within a 3% error (Table 3.10).
Despite a good fit at national level, water balance calculations for the ap-
proaches showed major errors at regional scale with an excessive water deficit
in Western Jutland and excess of water in the eastern part of Denmark (Table
3.10). The assumption of equal nationwide shelter conditions does not hold.
A variety of experiments have been carried out with 40×40 km
2
modelling of
regional shelter variations, but all approaches still yield large regional water
balance errors.
Thus, more experiments are required, but this is beyond the framework of this
project. Hence, it was decided to use a climate factor approach for calculation
of precipitation climate around year 1900, since the corrected precipitation for
1917-1950 worked well at a national scale.
1.3.5 Definition of delta change climate factor for precipitation (Δφ
m
)
National values of corrected precipitation are used to calculate monthly delta
change climate factors,
Δφ
m
, defined as the ratio of historical precipitation
(1890-1910),
P
hist
, to present precipitation (1990-2010),
P
pres
:
Δφ
m
=P
hist
/P
pres
Δφ
m
is based on corrected precipitation for 1890-1910 calculated using the
same bias correction method and the same model setup and assumptions as
33
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0036.png
for 1917-1950. Historical and present precipitation is averaged over all grid
cells, which have all been given equal spatial weight.
If it is assumed that the relative regional variations of the precipitation
amount are identical for the two periods, it may be reasonable to use the delta
change climate factors for projection of regional variations of the present cli-
mate to the period 1890-1910. This would imply that the temporal climate de-
velopment is the same all over the country, which does not appear to be the
case since the stream discharges change differently between regions. Thus, the
use of a general delta change factor,
Δφm,
results in regional variations of the
uncertainty in the calculated rainfall climate for this period, and probably also
around year 1900. Despite the differences in response, it seems reasonable to
use the general
Δφm
to adjust for the identified climate change. The use of the
climate factor in the hydrological modelling is described in section 3.3.2
1.3.6 Analyses of the effect of trend-adjusted wind speed on corrected
precipitation
It is difficult to quantify how much the uncertainty of wind corrections (shel-
ter, trends and homogeneity) contributes to the uncertainty of corrected rain-
fall. The results in section 1.3.3 suggest that the uncorrected
V
is somewhat
too high during the period of Danish Land Scale observations (1890-1910). For
the period of Beaufort observations (1911-1950)
V
is on average close to the
level without trend correction, but with underestimation in certain periods.
The idea of using geostrophic wind for correction of manually observed wind
was introduced
after
the modelling of discharge was finished. Therefore, this
section investigates how wind correction propagates the estimated climate
factors.
It is clear that a higher trend-corrected wind speed will result in an increased
amount of bias-corrected precipitation for the period 1917-1950 and that more
precipitation will cause too high values of modelled discharge compared with
the water balance error of 3% reported earlier. Since the wind speed is the only
variable changed, the increased corrected precipitation can be counterbalanced
by adjusting the shelter index until the precipitation amount 1917-1950 is ap-
proximately equal to the amount causing a water balance error of 3%.
The increase in corrected precipitation can be practically eliminated if the
shelter index is changed from 12 to 15, i.e., the new index compensates for the
changes in
V
and reproduces corrected precipitation close to the original val-
ues that resulted in a water balance error of 3%. As shown in Table 1.4 and
Figure 1.7, this is true for the years after the 1910s but not for the periods 1890-
1910 and 1900-1920. For approach 1 (no trend correction of
V
and
η=12),
P
c
is
much higher than for approach 2 (trend correction of
V
and
η=15).
For exam-
ple, for the period 1890-1910,
P
c
is 793.1 mm for approach 1 but only 758.5 mm
for approach 2.
An explanation of these results may be related to observation practice until
the 1910s. Until the 1910s, rainfall was measured 2 m above ground level
(DMI, 1875), but as for Hellmann the shelter index assumes that the orifice of
the rain gauge is at 1.5 m height.
The different measurement heights may have caused the average shelter in-
dex to be lower than 15 up to the 1910s and the rain gauge more exposed to
34
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0037.png
the impact of the wind. It can be calculated theoretically that for a typical gar-
den, the shelter index,
η
wgt
, will change from 15 to 13 if precipitation is meas-
ured in 2.0 m level instead of at 1.5 m. It is presumably more correct to use
η
wgt
= 13 in the period 1890-1915 and then
η
wgt
= 15 in the period 1916-1950
(approach 3). The results of the three approaches are shown in Table 1.4 and
Figure 1.7.
The question is whether the landscape was more open around year 1900, es-
pecially in the western part of Jutland, which would produce a lower shelter
index. For example, a smaller number of plantations and lower height of the
vegetation than today would be expected. Probably, the assumption of a con-
stant shelter index in the whole period 1890-1950 does not hold. For example,
it can be calculated that
η
wgt
=10 would practically eliminate the quite large
changes in corrected precipitation occurring in the 1910s (Table 1.4).
Table 1.4.
Measured annual precipitation (P
m
) and results of corrected precipitation (P
c
,) based on approach 1 (no trend correc-
tion of
V,
η=12),
approach 2 (trend correction of
V,
η=15)
and approach 3 (as approach 2 but with
η=13
for the period 1890-
1915, see text for explanation. The differences between approach 1 and 2 and 1 and 3 are also shown (mm and %).
Period
Measured precip,
P
m
634.1
645.5
674.9
673.0
665.3
671.9
Corrected precip,
P
c
(approach 1-3)
1
1890-1910
1900-1920
1910-1930
1920-1940
1930-1950
1916-1950
793.1
794.5
801.8
789.8
786.5
794.1
2
758.5
772.7
798.8
793.8
786.2
794.7
3
769.7
781.2
801.9
793.8
786.2
794.7
1 vs 2
-34.6 (-4.5 %)
-21.8 (-2.8 %)
-3.0 (-0.4 %)
3.9 (0.5 %)
-0.3 (-0.0 %)
0.5 (0.1 %)
Difference
1 vs 3
-23.4 (-3.0 %)
-13.3 (-1.7 %)
0.1 (0.0 %)
3.9 (0.5 %)
-0.3 (-0.0 %)
0.5 (0.1 %)
Figure 1.7.
Measured and cor-
rected annual rainfall in Denmark
1890-2010. Results of corrected
precipitation are shown for ap-
proach 1 (no trend correction of
V
and
η=12),
approach 2 (trend cor-
rection of
V
and
η=15)
and ap-
proach 3 (trend correction of
V
and
η=13).
Corrected precipita-
tion is also shown for the period
1989-2010 (Vejen et al., 2014).
Figure 1.8 shows a 20-year moving average of measured and corrected pre-
cipitation for shelter approach 1, 2 and 3 and for recent years. The trend curve
35
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0038.png
is based on measured precipitation 1935-2010 and is extended backwards in
time to 1890. The overall trend in both measured and corrected rainfall does
not appear to be linear over the period 1890-2010. The amount of corrected
rainfall seems to be relatively constant during 1915-1950, with a change to-
wards a wetter climate in the years after 1950. Before 1915, the change in type
of rain gauge and the lower shelter index caused higher uncertainty in the
level of corrected precipitation. Calculation of measured and corrected grid
precipitation for the period 1951-1988 are missing, but from a climate perspec-
tive it could be interesting to analyse the changes during this period to see
when the corrected precipitation started to increase.
Figure 1.8.
20-year moving average of measured and corrected annual rainfall in Den-
mark 1890-2010. Note that each point indicates the middle of the time interval. Large blue
dots: measured precipitation. Open black circles: approach 1, corrected precipitation using
shelter index 12 and
V
mw
. Red dots: approach 2, corrected precipitation using shelter in-
dex 15 and
V
mw
corrected for trends according to geostrophic wind. Large black dots: ap-
proach 3, as approach 2 except that
η=13.
Black squares:
P
c
for 1989-2010. Small blue
dots: trend line based on measured precipitation 1935-2010. Small black dots: difference
between measured and expected precipitation according to the trend line.
1.3.7 Comparison of measured and corrected precipitation 1891-2010
Figure 1.9 and Table 1.5 summarize the change in measured and corrected
precipitation over the period 1890-2010 given as (unofficial) 30-year climate
normal. It is seen that the measured amount of precipitation has increased by
approx. 97 mm from around 1900 (1891-1920) until today (1989-2010),
whereas the corrected value only has increased by 53 mm (approach 1) or 70
mm (approach 3). At the same time, the annual correction level of bias cor-
rected precipitation for approach 1 / approach 3 has fallen from 23.6/21.1%
for 1891-1920 over 18.0/18.1% for 1921-1950 to 14.6% for 1989-2010, which pri-
marily is coupled to changes in wind, temperature, and rain climate.
The amount of corrected precipitation has been at a rather constant level, at
least throughout the period between 1910s and 1950, although the measured
36
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0039.png
amount has increased. This is partly due to a short warmer period, especially
in the 1930s, with relatively little snow and lower correction level. The
changes towards a wetter climate thus seem to have occurred after 1950. As
pointed out by Førland and Hanssen-Bauer (2000), the omission of bias ad-
justment of precipitation entails a risk of interpreting on virtual climate
change. On the other hand, the reported uncertainties of the wind speeds are
critical for the corrected precipitation estimates, especially for the period 1890-
1910 where the Danish Land Scale was in use.
Figure 1.9.
Annual correction
level for Denmark 1890-1950 for
approach 3 and for the reference
period 1989-2010.
The trend in bias-adjusted precipitation and correction level is related to a
change towards a gradually warmer climate with less observed snowfall. The
changes in temperature, climate and precipitation type do not occur gradually
at a constant rate but are complexly linked, creating large year-to-year varia-
tions in the correction level.
Table 1.5.
Measured (P
m
) and corrected precipitation (P
c
) for approach 1 and 3 (for explanation see Table 1.4), and correction
level
K
α
(%) for approach 1 and 3 for Denmark for different 30-year periods. In the calculation of national averages, grid cells in
coastal regions are given lower weight depending on the fraction of land area.
1891-1920
Measured,
P
m
Corrected,
P
c
[1]
Corrected,
P
c
[3]
K
α
% [1]
K
α
% [3]
643.8
796.0
779.4
23.6
21.1
1901-1930
657.4
793.3
786.5
20.7
19.6
1911-1940
670.3
792.9
794.9
18.3
18.6
1921-1950
670.9
791.4
792.3
18.0
18.1
1931-1960
670.1
-
-
-
-
1961-1990
711.5
-
-
-
-
14.6
849.3
1989-2010
740.9
1.3.8 The climate around year 1900 and calculation of delta change
climate factor (Δφ
m
)
Table 1.6 shows estimated annual and monthly climate values around year
1900 (based on data from 1890-1910) compared with the conditions in 1989-
2010. The measured precipitation has increased by 92.9 mm but the corrected
one only by 57.2 mm (approach 1), corresponding to a decrease in the annual
correction level from 22.9% to 15.0%. This change is related to several changes
37
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0040.png
in climatic parameters. It has become warmer, which is particularly true dur-
ing winter, where
T
for January and February has increased by approx. 2 °C.
Opposite to this, the temperature change in the three summer months is only
+0.4 °C. The temperature change in the winter months is associated with a
marked decrease in the proportion of precipitation falling as snow. Around
1900, almost 50% of the corrected precipitation fell as snow in the three winter
months, while the proportion of snow in 1989-2010 was approx. 26%. The cli-
mate has only changed marginally in the direction of lower wind speeds.
Table 1.6.
The climate around 1900 (period 1890-1910) is compared with the reference period 1989-2010. Monthly and yearly
values of measured and bias-corrected precipitation,
P
m
and
P
c
, and correction level
K
a
=100(P
c
-P
m
)/P
m
(%) are shown for ap-
proach 1 (η
wgt
=12). Also shown are temperature
T
month
and the proportion of corrected precipitation fallen as snow (%) and trend
corrected wind speed,
V
(*), for the period 1890-1910 using geostrophic wind data. The delta change factor,
Δφ
m
, of each month
for approach 1 used for calculation of water flow is also included. Note that due to rounding of the monthly values, the sum of
these is not equal to the annual rainfall sum.
K
α
(%) is calculated as the average of individual months.
1890-1910 (approach 1)
P
m
J
F
M
A
M
J
J
A
S
O
N
D
Year
44.9
33.1
41.6
40.1
46.1
50.2
61.7
88.1
53.6
69.6
49.4
50.1
628.8
P
c
71.7
54.6
58.7
48.3
51.8
55.7
68.0
96.8
59.6
78.1
60.0
69.6
772.9
K
a
(%)
59.6
64.6
41.1
20.4
12.3
10.8
10.2
9.8
11.3
12.2
21.3
38.9
22.9
V (*)
5.8
5.6
5.6
5.3
4.9
4.8
4,7
5.0
4.8
4.9
5.0
5.6
5.2
T
-0.3
-0.4
2.1
5.8
10.7
14.7
16.3
15.5
12.6
8.4
3.9
1.1
7.5
Snow %
52.0
58.1
32.9
3.2
0.0
0.0
0.0
0.0
0.0
0.2
12.3
35.9
16.2
P
m
61.0
50.6
46.9
37.7
44.4
61.6
61.4
77.5
70.8
79.9
66.9
63.0
721.7
P
c
74.4
66.9
57.1
44.2
49.2
67.6
67.0
83.9
77.4
88.0
77.5
77.0
830.1
1989-2010
K
a
(%)
22.1
32.3
21.7
17.1
10.8
9.6
9.1
8.2
9.3
10.2
15.9
22.2
15.0
V
5.9
5.9
5.6
4.9
4.6
4.6
4.3
4.4
4.8
5.1
5.3
5.2
5.1
T
1.7
1.7
3.3
7.0
11.2
14.2
16.8
16.8
13.4
9.2
5.1
2.1
8.6
Snow %
18.3
32.2
12.8
1.7
0.0
0.0
0.0
0.0
0.0
0.1
6.2
21.1
7.7
Δφ
m
0.963
0.816
1.028
1.094
1.053
0.824
1.014
1.155
0.771
0.888
0.773
0.905
0.931
It is interesting to study the annual variation in rainfall. As today, the spring
months were also the driest around 1900, and the wettest months occurred in
late summer or autumn. Around 1900, August was the wettest month of the
year with 96.8 mm (corrected), while in 1989-2010 it was October with 88.0
mm. The spring months March-May were slightly wetter than today, while
August had more precipitation. Almost all the other months were drier than
in 1989-2010. These differences are reflected in the delta change climate factors
(Δφ
m
) shown in Table 1.6. As stated earlier, the magnitude of these factors may
be affected by the uncertainty of wind speeds converted from Danish Land
Scale.
1.4
Evaluation of uncertainty
For the period 1890-1910, absence of discharge measurements makes it impos-
sible to use hydrological modelling for evaluation of the precipitation esti-
mates, especially for spatial distribution. However, it is assumed that if the
model setup and its assumptions work for 1917-1950, for which discharge
measurements are available for verification, it does so for 1890-1910.
38
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0041.png
Figure 1.10 shows a comparison between
T
and
K
α
(%) for 1890-1910 relative
to the reference period 1989-2010. There seems to be a tendency for the cor-
rection level 1890-1910 to be slightly higher than for the other two periods.
This is especially true at the lowest temperatures, that is, primarily during the
winter months.
Figure 1.10.
A comparison of
temperature and correction level
(%) for monthly values of bias-
corrected precipitation between
the reference period 1989-2010
and the periods 1890-1910 and
1914-1950.
Other factors that may affect uncertainty are the basic assumptions for the
calculations.
A measure of the uncertainty of corrected rainfall is composed of several sig-
nificant contributions, which are further elaborated upon in Vejen (2021):
•
Stochastic uncertainty of the correction model.
•
Spatial uncertainty (values of
T, V
and
α)
is allocated to a precipitation sta-
tion from nearest grid cell.
•
Spatial uncertainty of gridded precipitation due to in-homogeneity of rain
gauge station network.
•
Uncertainty of methods for calculation of meteorological variables.
•
Other sources of uncertainty, for instance adjustment for shelter effect and
trend correction of wind speed.
Due to the challenges of determining the shelter index (section 1.3.4), the cli-
mate factor,
Δφ
m
, was used as a pragmatic approach to calculate corrected rain-
fall for 1890-1910. The result in Figure 1.11 shows the regional distribution of
corrected rainfall around 1900 compared with more recent values. In all regions,
systematically smaller amounts of rainfall are seen. However, using a climate
factor contributes to regional uncertainty as the change in the regional correc-
tion level depends on the magnitude of the corrected rainfall, as shown in Fig-
ure 1.12. The relative change is largest in the western part and lowest in the
eastern part of Denmark.
The use of monthly delta change climate factors results in a fraction between
the historical and recent precipitation that varies from relatively high values in
the eastern part of Denmark to quite low values in the western part, i.e., there
39
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0042.png
are regional variations in the percentage change in the precipitation amount
(Figure 1.11). This variation is only between 6 or 8% lower than today, which
is a rather small regional difference.
Figure 1.11.
Left: Mean precipitation per year 1890-1910 based on approach 1. Right: Mean corrected precipitation per year
1997-2017.
Figure 1.12.
Mean change in cor-
rected precipitation between
1890-1910 and 1997-2017.
The use of a national delta change climate factor at a regional scale may result
in systematic regional bias, which may have an impact on the uncertainty of
runoff calculations at the spatial scale shown in the Figure (10×10 km
2
). It may
40
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
be necessary to average for larger regions and to redefine the delta change
climate factor, for example by incorporation of regional variability.
1.5
Conclusion
The objective is to provide this project with spatially distributed estimates of
required meteorological variables to support the calculation of evaporation
and hydrological modelling. To make this possible, historical meteorological
data over the period 1890-1950 have been digitized. Furthermore, an approach
for bias correction of historical rain gauge data has been developed, and based
on data series of bias-corrected precipitation, air temperature and wind data,
the climate around year 1900 has been estimated and validated. Various con-
ditions contribute to the uncertainty of the corrected rainfall. Particular atten-
tion has been paid to the wind speed based on manual observations. Some of
the wind series are affected by homogeneity breaks and deviation from the
general trends in
V.
The calculations and assumptions are evaluated for the period 1917-1950 us-
ing water balance modelling of discharge. At national level, a water balance
error of 3% has been found, but this covers large regional differences in error
level. Of the many assumptions, the use of a general shelter index seems to
have a large impact on the regional uncertainty. Analyses of the relationship
between monthly temperature and the correction level for the different peri-
ods support the impression of reasonable estimates at national level. How-
ever, it requires more model experiments with fine tuning of the different
basic assumptions to reduce the regional uncertainty to an acceptable level.
It has therefore been necessary to calculate corrected rainfall for 1890-1910 by
using a delta change climate factor. In this approach, national monthly correc-
tion factors were calculated based on corrected rainfall for 1890-1910 com-
pared with 1990-2010. These national factors were then applied to the present
daily corrected precipitation data to provide a spatially distributed daily time
series of precipitation for the period 1890-1910.
Based on the assumptions in this study, a delta change value of 0.931 per year
was found, which corresponds to approximately 773 mm per year or 57 mm
(7%) less than the precipitation amount in the reference period 1989-2010.
Much more of the precipitation around year 1900 consisted of snow, which is
related to a colder climate; 16.2% of the total amount was snow compared
with 7.7% in the reference period. This also explains the higher correction fac-
tor of 22.9% per year compared with 15.0% for the reference period.
The uncertainty of bias-corrected precipitation depends, among other things,
on the reliability of wind speed data. Around year 1900, wind speed was ob-
served with the so-called Danish Land Scale, which can be considered as a
kind of half Beaufort. Manual wind observations are subject to different errors
and larger uncertainty compared with modern automatic instruments. At the
end of the project, tests could be made for correction of wind data for trend
errors using a long time series of on geostrophic wind speed. It was found
that the wind speed used for bias correction was probably overestimated. As
a result, the corrected precipitation and delta change values were a little too
high. Use of the corrected wind speed resulted in a delta change value of 0.906
(-2.7%,) which corresponds to around 20 mm less precipitation per year com-
pared with the original precipitation estimate, which is an acceptable differ-
ence given the many assumptions and uncertainties.
41
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
1.6
References
Alexandersson H., Schmith, T., Iden, K. and Tuomenvirta, H., 1998. Long-
term variations of the storm climate over NW Europe. Global Atmosphere
and Ocean Systems 6, 97-120.
Allerup, P. and Madsen, H., 1979. Accuracy of point precipitation measure-
ments. Danish Meteorological Institute, Climatological Papers, No. 5, Copenha-
gen 1979, 84pp.
Allerup, P. and Madsen, H., 1980. Accuracy of point precipitation measure-
ments. Nordic Hydrology 11, 57-70.
Allerup, P., Madsen, H. and Vejen, F., 1997. A Comprehensive Model for Cor-
recting Point Precipitation. Nordic Hydrology 28, 1-20.
Brandt, M. L., 1994. The North Atlantic Climatological Dataset (NACD). In-
strumenter og rekonstruktioner. En illustreret gennemgang af arkivmateriale.
DMI Technical Report 94-19, København 1994, 74pp.
Cappelen, J., 2000. The Climate of Denmark - Key Climatic Figures 1990-1999.
Technical Report, TR00-08, DMI, Copenhagen 1999, 47pp.
Cappelen, J., 2019. Denmark – DMI Historical Climate Data Collection 1768-
2018. Technical Report, TR19-02, DMI, Copenhagen 2019, 112pp.
Colli, M., Pollock, M., Stagnaro, M., Lanza, L. G., Dutton, M. and O’Connell,
E., 2018. A computational fluid-dynamics assessment of the improved perfor-
mance of aerodynamic rain gauges. Water Resources Research 54, 779-796.
Det Statistiske Departement, 1964. Folketal, areal og klima. Statistiske under-
søgelser nr. 10, Det Statistiske Departement 1964, 271pp.
DMI, 1875. Meteorologisk Aarbog for 1874. Udgivet af det danske meteorolo-
giske institut, Kjøbenhavn 1875.
DMI (Danish Meteorological Institute), 1890-1950. Maanedsoversigt over
Vejrforholdene. Udgivet af det Danske Meteorologiske Institut. Published
monthly and collected in yearbooks, 1890-1950.
Feiccabrino, J., Graff, W., Lundberg, A., Sandström, N. and Gustafsson, D.
2015. Meteorological knowledge useful for the improvement of snow rain sep-
aration in surface based models. Hydrology 2, 266-288. doi:10.3390/hydrol-
ogy2040266.
Førland (ed), E.J., Allerup, P., Dahlström, B., Elomaa, E., Jónsson, T., Madsen, H.,
Perälä, J., Rissanen, P., Vedin, H. and Vejen, F., 1996. Manual for operational cor-
rection of Nordic precipitation data. Nordic Working Group on Precipitation.
First Edition. DNMI, Report Nr. 24/96, 66pp.
Førland, E.J. and Hanssen-Bauer, I. 2000. Increased precipitation in the Norwe-
gian arctics: True or false? Climatic Change 46, 485-409.
Groisman, P. Ya., and Legates, D.R., 1994. The accuracy of United States pre-
cipitation data. Bulletin of the American Meteorological Society 75, 215-227.
42
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Harder P. and Pomeroy J., 2013. Estimating precipitation phase using a psy-
chrometric energy balance method, Hydrological Processes 27, 1901-1914,
doi:10.1002/hyp.9799.
Kaufeld, L., 1981: The Development of a New Beaufort Equivalent Scale. Me-
teorolgische Rundschau 34, 17-23.
Kristensen, L. and Frydendal, K., 1991. Danmarks vindklima fra 1870 til nuti-
den. Havforskning fra Miljøstyrelsen, Nr. 2, 1991, Miljøministeriet, Miljøsty-
relsen, 57pp.
Luthardt, H. and Hasse, L., 1981. On the relationship between surface and ge-
ostrophic wind in the German Bight area. Beitrage zur Physik der At-
mosphäre 54, 222-232.
Lysgaard, L., 1969. Foreløbig oversigt over Danmarks Klima, lufttryk, vind-
forhold, lufttemperatur, solskin, nedbørforhold og luftfugtighed hovedsage-
lig i perioden 1931-60. Det Danske Meteorologiske Institut, Meddelelser Nr.
19, 109pp.
Nespor, V., 1996. Investigation of Wind-Induced Error of Precipitation Meas-
urements Using a Three-Dimensional Numerical Simulation. Zürcher Geog-
raphische Schriften, Heft 63, Geographisches Institut ETH, Zürich 1996,
117pp.
Olesen, J.E., Bøcher, P. K, and Jensen, T., 2000. Comparison of scales of climate
and soil data for aggregating simulated yields of winter wheat in Denmark.
Agricultural Ecosystems and Environment 82, 213-228.
Plauborg F., Refsgaard, J.C., Henriksen, H. ., Blicher-Mathiesen, G. and Kern-
Hansen, C., 2002. Vandbalance på mark- og oplandsskala. DJF rapport. Mark-
brug nr. 70.
Refsgaard, J.C., Stisen, S., Højberg, A.L., Olsen, M., Henriksen, H.J., Børgesen,
C.D., Vejen, F., Kern-Hansen, C. and Blicher-Mathiesen, G., 2011. Vandba-
lance i Danmark - Vejledning i opgørelse af vandbalance ud fra hydrologiske
data for perioden 1990-2010. GEUS rapport 2011/77, 75pp.
Scharling, M., 1999. Klimagrid – Danmark. Nedbør, lufttemperatur og poten-
tiel fordampning, 20×20 & 40×40 km. Metodebeskrivelse. DMI Technical Re-
port 99-12, Copenhagen 1999, 48pp.
Sevruk, B., 1979. Accuracy of Point Precipitation Measurement. Laboratory of
Hydraulic Research, Hydrology and Glaciology, Federal Institute of Technol-
ogy, Zürich, Switzerland. Commision for Hydrology, WMO 1979, 121pp.
Sevruk, B. and Hamon, W.R. (WMO), 1984. International Comparison of Na-
tional Precipitation Gauges with a Reference Pit Gauge. WMO, Instruments
and Observing Methods, Report No. 17, WMO/TD-No. 38, Secretariat of the
World Meteorological Organisation, Geneva, Switzerland, 13pp.
Sevruk, B. (1988): Wind speed estimation at precipitation gauge orifice level.
WMO, Instruments and Observing Methods, Report No. 33, WMO/TD–No.
222, Paper presented at the WMO Technical Conference on Instruments and
43
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Methods of Observation (TECO-1988), Leipzig, German Democratic Republic,
16-20 May 1988, 4pp.
Sevruk, B. and Klemm, S., 1989. Catalogue of national standard precipitation
gauges. WMO, Instruments and Observing Methods, Report no. 39, WMO/TD-
No. 313, 1989, 50pp.
Sevruk, B., Hertig, J.-A., and Spiess, R., 1989: Wind field deformation above pre-
cipitation gauge orifices. Atmospheric Deposition (Proceedings of the Baltimore
Symposium, May 1989), IAHS Publ. No. 179, 65-70.
Vejen, F., Madsen, M., og Allerup, P., 2000: Korrektion for fejlkilder på måling
af nedbør. Korrektionsprocenter ved udvalgte stationer. 1989-1999. Danish
Meteorological Institute, Technical Report No. 00-20, Copenhagen 2000, 53pp.
Vejen, F., Vilic, K., Jensen, H. and Kern-Hansen, C., 2014. Korrigeret Nedbør
1989-2010, 2011-2012 & 2013. Databeskrivelse & Resultater. Konsulentopgave
udført for DCE – Nationalt Center for Miljø og Energi, Aarhus Universitet.
DMI Technical Report 14-13, Copenhagen 2014, 228pp.
Vejen, F., 2021. Climate around year 1900. Aarhus University, DCE - Danish
Centre for Environment and Energy, 42 s. – Scientific briefing no. 2021|52
https://dce.au.dk/fileadmin/dce.au.dk/Udgivelser/No-
tater_2021/N2021_52.pdf
Yang, D., Goodison, B.E., Metcalfe, J.R., Golubev, V.S., Elomaa, E., Günther,
T., Bates, R., Pangburn, T., Hanson, C.L., Emerson, D., Copaciu, V. and
Milkovic, J., 1995. Accuracy of Tretyakov precipitation gauge: Result of WMO
intercomparison. Hydrological Processes 9, 877-895.
WMO, 1970. The Beaufort Scale of Wind Force. WMO Commission for Mari-
time Meteorology, Marine Sciences Affairs Report NO. 3, 22pp.
WMO, 2008. Guide to Meteorological Instruments and Methods of Observation.
WMO-No. 8, 7
th
edition. Secretariat of the World Meteorological Organization,
Geneva, Switzerland.
44
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0047.png
2
Climate around the year 1900: Calculation
of global radiation and evaporation for his-
toric sites
Author: Johannes W.M. Pullens
1
and Jørgen E. Olesen
1
Quality assurance: Mathias Neumann Andersen
1
1
DCA,
Aarhus University, Department of Agroecology
Abstract
Since no data on global radiation and evapotranspiration were available for
the period 1890-1950, they had to be calculated by using the measured mini-
mum and maximum air temperature for the same period. Firstly, the air tem-
perature and the distance to the nearest coastline were used to calculate the
global radiation. The modelled global radiation was used to calculate the ref-
erence evaporation under Danish conditions. The modelled global radiation
and reference evapotranspiration in the year 1900 are in good agreement with
values measured at Foulum from 1987-2013.
2.1
Introduction
To be able to simulate the historical water discharge and nitrogen leaching,
measurements of daily global radiation and reference evaporation are needed.
However, the historical dataset of climate variables (chapter 1) does not have
any recordings of such data. Therefore, these parameters must be estimated
by using the available data (temperature and precipitation).
2.2
Materials and methods
2.2.1 Calculation of global radiation
To calculate the daily global radiation for the sites during the years in which
no global radiation measurements were conducted, Hargreaves’ radiation for-
mula was used (Allen et al., 1998-FAO report 56, chapter 3, formula 50
therein):
����
=
����
(����
− ����
)
����
R
s
is the daily global radiation [MJ/m
2
/d],
R
a
is daily extra-terrestrial radia-
tion [MJ/m
2
/d],
T
max
is maximum daily air temperature [°C],
T
min
is the mini-
mum daily air temperature [°C], and
K
Rs
is the adjustment coefficient [°C
-1
].
K
Rs
ranges from
≈
0.16 for inland locations to
≈
0.19 for locations close to the
coast, where the air mass is influenced by the water bodies (Allen et al., -FAO
report 56, chapter 3). The extra-terrestrial radiation,
R
a
, can be calculated by
using the latitude of the site (Allen et al., FAO report 56, chapter 3, formula 21
therein).
All analyses were conducted using the R statistical software v3.5.0 (R
Development Core Team. R Foundation for Statistical Computing, 2018).
45
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0048.png
To determine the
K
Rs
values for Denmark, data from 10 sites from 1987 until
2013 were used (Figure 2.1). For all 10 sites, the minimum and maximum tem-
perature and the global radiation were recorded and subsequently used in
Hargreaves’ radiation formula. To determine the
K
Rs
values based on the dis-
tance to the closest coastline for these sites, a linear regression is conducted.
The purpose of this regression is to calculate the
K
Rs
values for the historical
sites based on distance to the coast. By using the
K
Rs
values and the air tem-
perature, global radiation can be calculated.
Figure 2.1.
Location of the 10
sites used in this stud, where
global radiation was measured
over the period 1987-2013.
2.2.2 Calculation of reference evapotranspiration (ET0) for historical
sites
The daily reference evapotranspiration was calculated by means of the Makkink
formula calibrated for Denmark (Makkink, 1957; Mikkelsen and Olesen, 1991):
��������
=
���½
+
���½
��������
����(����
+
����)
where
ET
0
is the daily reference evapotranspiration [mm],
β
M0
and
β
M1
are
constants [unitless],
λ
is the latent heat of vaporisation [2.465 MJ/kg,
γ
is the
psychrometric constant [0.667 mb/°C],
s
is the slope of the vapour pressure
curve [mb/°C], and
S
i
is the daily global radiation [MJ/m
2
/day]. The global
radiation is calculated in the previous step,
β
M0
is set to 0, and
β
M1
is set to 0.7
calculated for Danish conditions (Aslyng and Hansen, 1982)
46
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0049.png
The slope of the vapour pressure curve (s) is calculated by using the formula
defined by FAO (Allen et al., 1998):
4098 0.6108������������
17.27����
����
+ 237.3
(���� + 273.3)
����
=
where
T
is the mean daily air temperature (°C).
2.3
Results
Table 2.1.
Mean
K
Rs
values per site in the period 1987-2013.
Site
Aarslev
Askov
Flakkebjerg
Foulum
Jyndevad
Oedum
Roskilde
Silstrup
Tylstrup
Tystofte
K
Rs
0.17
0.16
0.17
0.17
0.15
0.16
0.17
0.19
0.17
0.18
These sites follow the previously described trend of K
Rs
and distance to the
closest coastline (Allen et al., 1998- FAO report 56, chapter 3); therefore, a lin-
ear model was fitted to the distance to the coast and the K
Rs
value (Figure 2.2).
Figure 2.2.
Relation between K
Rs
values and distance to coast for
all 10 sites.
The monthly average, minimum and maximum of the global radiation mod-
elled for Foulum for the period 1890-1950 (Table 2.2) are slightly higher than
the measured values for Foulum between the years 1987-2013 (Table 2.3);
however, the range is the same as are the annual average and the seasonal
fluctuations.
47
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0050.png
Table 2.2.
Modelled monthly minimum, maximum and average global radiation (MJ/m
2
/d) with standard deviation in the period
1890-1950 for Foulum.
Month
January
February
March
April
May
June
July
Augustus
September
October
November
December
Annual
Minimum
5.8
11.3
27.9
56.5
83.9
90.9
110.1
69.2
46.7
16.5
7.4
5.6
5.6
Maximum
51.3
95.9
167.1
240.4
301.2
311.4
295.9
253.8
182.0
108.2
54.0
31.3
311.4
Average
20.3
42.0
84.9
146.5
202.1
217.9
203.0
159.8
108.0
56.6
24.9
14.4
107.0
Standard deviation
6.6
13.2
23.8
31.5
34.4
35.2
34.4
29.2
23.6
16.9
7.8
3.9
77.6
Table 2.3.
Measured monthly minimum, maximum and average global radiation (MJ/m
2
/d) with standard deviation in the period
1987-2013 for Foulum.
Month
January
February
March
April
May
June
July
August
September
October
November
December
Annual
Minimum
0.9
4.3
8.6
18.1
25.1
21.6
21.6
16.4
7.8
0
1.7
1.7
0
Maximum
59.6
96.8
165.0
215.1
265.2
270.4
262.7
220.3
167.6
104.5
55.3
38.0
270.4
Average
14.8
32.6
70.8
115.8
159.4
167.1
159.7
127.8
83.8
44.2
18.3
10.9
84.0
Standard deviation
10.6
20.0
37.2
50.4
59.8
62.8
55.9
44.6
36.2
23.7
11.7
7.5
70.3
The
K
Rs
for each grid cell was calculated based on the distance from the mid-
point of the cell to the coast (Figure 2.3).
48
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0051.png
Figure 2.3.
Distribution of K
Rs
values on a 20*20 km grid.
The monthly measured and modelled reference evapotranspiration are in
good agreement (Table 2.4 and Table 2.5). The annual values are also in good
agreement with each other.
Table 2.4.
Modelled monthly minimum, maximum and average reference evapotranspiration (mm/d) with standard deviation in
the period 1890-1950 for Foulum.
Month
January
February
March
April
May
June
July
August
September
October
November
December
Annual
Minimum
0.1
0.2
0.4
0.8
1.2
1.3
1.6
1.0
0.7
0.2
0.1
0.1
0.1
Maximum
0.8
1.4
2.5
3.5
4.4
4.6
4.3
3.7
2.7
1.6
0.8
0.5
4.6
Average
0.3
0.6
1.2
2.1
3.0
3.2
3.0
2.3
1.6
0.8
0.4
0.2
1.6
Standard deviation
0.1
0.2
0.4
0.5
0.5
0.5
0.5
0.4
0.4
0.3
0.1
0.1
1.1
49
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0052.png
Table 2.5.
Measured monthly minimum, maximum and average reference evapotranspiration (mm/d) with standard deviation in
the period 1987-2013 for Foulum.
Month
January
February
March
April
May
June
July
August
September
October
November
December
Annual
Minimum
0
0.1
0.1
0.3
0.4
0.4
0.4
0.3
0.2
0
0
0
0
Maximum
0.6
1.2
2.5
4.2
6.2
5.9
5.7
5.0
3.4
2.0
0.9
0.5
6.2
Average
0.2
0.4
1.0
1.9
3.0
3.3
3.4
2.7
1.6
0.9
0.3
0.1
1.6
Standard deviation
0.1
0.3
0.5
0.9
1.2
1.3
1.2
1.0
0.7
0.4
0.2
0.1
1.5
2.4
Conclusion
Climate data are an important input to models. However, the required data
have not always been measured. Examples of this are global radiation and
reference evapotranspiration that are needed as input to hydrological models.
By using the measured minimum and maximum air temperature from Danish
meteorological stations for the 1890-1950, the global radiation and reference
evapotranspiration can be calculated and used in the simulation of nitrate
leaching over this period. Such calculated values of global radiation and es-
pecially reference evapotranspiration during 1890-1950 were found to be in
good agreement with values measured at Foulum during 1987-2013.
2.5
References
Allen, R., Pereira, L., Raes, D. and Smith, M., 1998. Crop evapotranspiration-
Guidelines for computing crop water requirements-FAO Irrigation and
drainage Paper 56, FAO 1–15 (1998). Rome. Retrieved from
http://www.fao.org/docrep/X0490E/x0490e00.htm
Aslyng, H.C. and Hansen, J., 1982. Water balance and crop production
simulation. Water Balance and Crop Production Simulation. Retrieved from
https://www.cabdirect.org/cabdirect/abstract/19830748610
Makkink, G. 1957. Ekzameno de la formulo de Penman. Journal of Agricul-
tural Science 5, 290–305.
Mikkelsen, H.A. og Olesen, J.E., 1991. Sammenligning af metoder til
bestemmelse af potentiel vandfordampning - Beretning nr. S 2157-1991.
Forskningscenter Foulum, 8830 Tjele.
R Development Core Team. R Foundation for Statistical Computing, 2018. A
language and environment for statistical computing. Vienna, Austria: R
Foundation for Statistical Computing. Retrieved from http://www.r-
project.org/.
50
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
3
Modelling discharge to the sea
Author: Anker Lajer Højberg
1
Quality assurance: Lars Troldborg
1
1
GEUS,
Geological Survey of Denmark and Greenland
Abstract
Purpose:
In the present chapter, the groundwater flow and the stream dis-
charge from land to sea around year 1900 are simulated. This simulation is
used to estimate the historical recharge to the groundwater system and the
groundwater transport between the root zone and the surface water system.
The simulation is used as input to the national nitrogen model for assessing
the total nitrogen load to the sea presented in chapter 8.
Materials and methods:
The National Water Resources Model (the DK
model) that describes groundwater and surface water flows, including the in-
teraction between the two media, has been used to simulate the historical
stream discharge. Simulation was carried out for the period 1917-1950, for
which historical observations exist, and afterwards for the period 1890-1910.
Comparison between observed and simulated discharges was used to evalu-
ate the historical precipitation established from observations and corrected for
wind and shelter effects.
The DK model is originally developed to represent current conditions, but in
the present project it has been modified to represent historical conditions as
well. This entailed use of historical climate data, adjustment of groundwater
abstraction and point discharges and incorporation of land use, as described
in chapter 5. Artificial drainage, being an important flow and transport path-
way from the fields to the surface water system, has almost doubled since
1900. Using historical data on the estimated production of tile drains, drainage
in the model was modified to represent the conditions around year 1900.
Results and discussion:
Using precipitation established from observational
data corrected for wind and shelter effects for the period 1917-1950 showed
that the total observed and simulated discharge at national scale agreed
within a 3% water balance error. However, large variations in the fit between
observed and simulated discharges were observed for the different regions in
Denmark. Alternative models for correcting the observed precipitation were
tested but without a satisfactory result. Hence, use of observed and corrected
precipitation data results in large regional uncertainty, but since the true wind
speed and shelter index are unknown, addressing this uncertainty was be-
yond the framework of the current project. A delta change approach was
therefore applied using national monthly delta change factors to correct cur-
rent precipitation to a historical level. This approach is more robust at a na-
tional scale and is less dependent on the distribution of wind speed and shel-
ter index. The inherent assumption is, however, that the relative change in
precipitation has been the same for the entire country. Based on the delta
change approach, it was estimated that the historical precipitation was 7% less
on a yearly basis but with variations between the different months. With the
51
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
change in land use, evapotranspiration, and drainage, the change in precipi-
tation resulted in a simulated discharge for 1900 that was approx. 12% lower
than the present discharge.
Conclusions:
Using a delta change approach to establish a historical time se-
ries of precipitation, the discharge to the sea was estimated using the national
water resources model that was modified to represent the conditions around
year 1900. The discharge estimate can be used to assess the change in dis-
charge at the national scale but will not be valid at local scale.
3.1
Introduction
Nitrate leaching from the root zone is transported through the groundwater
to the freshwater system and by the streams further to the marine recipients.
Additionally, the amount of nitrate leaching to the root zone in the period
around year 1900 is calculated from an estimated nitrate concentration (chap-
ter 5) and the amount of water infiltrating the subsurface. A valid representa-
tion of the freshwater system, including estimates of the net precipitation,
transport pathways and potential reduction of nitrate during transport is thus
essential for calculating the total nitrogen load to the coast.
Previous analysis of long time series of observed stream discharge and pre-
cipitation indicates that there has been a significant change in the precipitation
rate between the two periods considered (Jensen (ed.) 2017), with less precip-
itation/stream discharge for the historical period. To assess and quantify the
effect of this change, a hydrological model is utilised to simulate the freshwa-
ter cycle for both the current and historical conditions. The model results are
then used as input to the national nitrogen model to estimate the total nitro-
gen load to the coast in chapter 8.
3.2
Material and methods
Freshwater discharge to the sea is calculated by the national water resources
model (DK model) (Højberg et al., 2013; Henriksen et al., 2003). The model
construction represents the current conditions, i.e. current climate, ground-
water abstractions, land use and farming practice, which together with model
calibration and validation are described in detail in Højberg et al. (2015a) and
will not be described further here. To represent the historical conditions
around year 1900, several adjustments to the model have been made, which
are described in the current chapter.
With the DK model modified to represent the historical conditions, climate
data estimated in chapter 1 and 2 were used as forcing variables in the model
to simulate the historical stream discharge. The simulation was compared
with discharges observed at 41 stream discharge stations with data within the
period 1917-1950. This comparison was used to evaluate the corrected precip-
itation data, i.e., observed precipitation corrected for wind and shelter effects.
The focus of this comparison was how well the total discharge volume could
be represented. When a model to correct the historical precipitation was de-
veloped, this was used to establish a time series for the period 1890-1910 that
was used in the DK-model to calculate discharges for the same period.
52
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0055.png
3.2.1 DK model
The DK model is a physically based and fully distributed model describing
both the groundwater and the surface water systems and their interlinkage
(Højberg et al., 2013; Henriksen et al., 2003) (www.vandmodel.dk), and is de-
veloped in the model system MIKE SHE/MIKE11 (Havnø et al., 1995). Actual
evapotranspiration and the amount of water that recharges the saturated zone
are calculated using a two-layer water balance method based on a formulation
presented in Yan and Smith (1994). Simulation of flow and transport in the
groundwater system relies on a comprehensive three-dimensional interpreta-
tion of the subsurface hydrogeology, while one-dimensional approximation
is used to describe the stream flow for which stream dimensions are charac-
terised from stream cross-sections. The DK-model consists of seven sub-mod-
els as illustrated in Figure 3.1.
Figure 3.1.
The seven sub-mod-
els in the national water re-
sources model (DK-model).
The model has been developed continuously during the past 20 years, with
focus on updating the model basis, i.e., detailing the input data – the hydro-
geological interpretation that describes the water flow in the subsurface and
model performance through improved parameterisation and calibration
schemes. The model has been used and tested in numerous studies to analyse
the impact of varying conditions that are central for the present study. These
53
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0056.png
include climate change impacts as well as changes in land use and manage-
ment. The model is regularly updated based on new data and knowledge. The
model version applied in the present study is the DK-model2014 (Højberg et
al., 2015a), which is the same model version used to develop the national ni-
trogen model (Højberg et al., 2015b).
3.2.2 Modification to the DK-model to represent 1900 conditions
Most data needed to construct and run the national hydrological model for
present conditions can be obtained directly from national databases. This is
not the case for the period around year 1900, and it has therefore been neces-
sary to make some assumptions of how to represent the historical conditions
in the hydrological model, as summarised in Table 3.7 and described below.
Table 3.7.
Summary of modification to the hydrological model to represent the conditions around year 1900.
Present*
Climate
Institute (DMI).
Land use
Data on land use and vegetation on agricul-
tural fields are obtained from national data-
bases.
Drainage
All agricultural areas are assumed drained.
Drains are distributed in accordance with the total drained
area, and the probabilities of drainage are determined from
the physical properties
Abstraction
Abstraction for irrigation and drinking wa-
ter/other uses are based on yearly data re-
ported to national databases.
charge
national database.
It is assumed that there was no irrigation and that all other ab-
straction was 20% of the current amount with the same geo-
graphical distribution as at present.
the current amount with the same geographical distribution as
at present.
* (see Højberg et al., 2015a for details)
Year1900
potential evapotranspiration calculated from temperature, see
chapter 1 and 2.
Land use is obtained from statistics at parish level, chapter 5
National data from the Danish Meteorological Precipitation is estimated by the delta change approach and
Point source dis- Data on discharge to streams is obtained from It is assumed that the discharge by point sources was 20% of
Climate
Development of the historical times series for precipitation and potential
evapotranspiration is described in chapter 1 and 2. Time series for potential
evapotranspiration calculated based on historical data have been used di-
rectly in the model setup.
The initial method for calculating precipitation, as described in chapter 1, was
to develop an approach to correct the historical data and test this approach by
using the data in the hydrological model and test the results of this against
the observed discharge for the period 1917-1950, from which such data exist.
By testing different approaches, it was found that, due to limited and uncer-
tain historical data on wind speed and shelter index for rain gauge stations,
calculation of distributed historical precipitation was beyond the resources of
the current project. While the spatial pattern is very sensitive to the local tem-
poral and spatial conditions, the national long-term estimate, i.e. the change
in precipitation for the entire country averaged over a period, is a much more
robust estimate. It was thus decided to apply a delta change approach using
a national delta change factor per month. This was achieved by calculating an
average total precipitation for the entire country for each month for the peri-
ods 1890–1910 and 1990–2010 as well as the monthly fractions between the
54
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
two periods. A historical precipitation dataset was then generated by multi-
plying the present daily precipitation data with the corresponding monthly
correction factors, i.e. all days in January are multiplied by the same factor, all
days in February by another factor etc.
Land use
Land use was estimated on parish level statistics in which land use has been
reduced to eight categories as described in chapter 5. Since only the statistics
and not the exact locations are known, land uses have been distributed ran-
domly within each parish. However, such random distribution may result in
undesirable placements, with agricultural land being located in areas with
upwelling groundwater, i.e. wet areas that are wetlands, meadows or similar
and not cultivated farmland. The land uses have therefore been corrected to
avoid such inconsistency. This has been achieved by identifying agricultural
land uses in areas with upwelling groundwater and switching the land use
code with the nearest grid cell with no upwelling groundwater and a land use
code different from agriculture.
Under the current conditions, a large fraction of the precipitation can be
routed directly to the sewage system and further to the streams in urban areas.
This process is assumed much less important around year 1900. However,
some impermeable surfaces existed in the larger cities. Direct runoff to the
surface water system has thus been maintained in the larger cities. This has
been achieved by assuming the structure of the larger cities to be identical
with that of the present in terms of location and relative size, but with a much
lower runoff fraction.
The crop development in terms of leaf area index (LAI) and root zone depth
controls the actual evapotranspiration. It is assumed that a lower nitrogen ap-
plication in the historical period resulted in a lower LAI, where the maximum
LAI at full crop development was reduced from 5 to 3, while the maximum
root zone depths were unchanged from present conditions.
Drainage
Drainage of the land can be achieved through natural drainage systems, such
as ditches or by artificial tile drains. Due to the spatial resolution of the hy-
drological model (500 x 500m grids), it is not possible to represent all ditches
directly in the model. Drainage by both ditches and tile drains are therefore
represented in the model by model-drains to mimic the fast runoff from these
areas, preventing accumulation of water on the surface.
In the version of the DK-model representing the current conditions, drainage
is implemented in the entire model. However, drain flow is only generated if
the groundwater table rises above the drain levels. The rationale for including
drains in the entire model area is that the exact location of tile drains is largely
unknown, and it is assumed that all agricultural areas are drained if the
groundwater table is high during part of the year, thereby limiting access to
the fields with heavy machinery. The same assumption is not valid for the
period around year 1900, but drainage of agricultural land by tile drains was
already significant with approximately 660,000 hectares in 1907 (Aslyng, 1980)
and a distribution in the different regions as shown Table 3.8. No data are
55
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0058.png
available for the southern part of Jutland, as this was under German admin-
istration up to 1920, and drainage has thus been estimated based on the drain-
age density in West and East Jutland as discussed further below.
Table 3.8.
Tile drainage statistics from 1907 (Aslyng 1980).
Zea-land Bornholm Lolland-
Falster
Drained area
(1000 ha)
235
19
85
109
Funen
East
Jutland
127
West
Jutland
33
North Jut-
land
51
South
Jutland
32*
Entire
country
659
* No data are available for South Jutland, and the number has thus been estimated based on the drainage density in West and
East Jutland, see text for further explanation.
The location of tile drains areas around year 1900 is unknown, and the follow-
ing assumption was made for the distribution of tile drains:
1. Agricultural land in lowland areas are drained by ditches and not tile
drains.
2. Agricultural land located in areas with a high density of ditches are solely
drained by ditches and not tile drains.
3. Tile drains are sequentially distributed in the remaining agricultural land
according to the area’s need for drainage, i.e. areas with a high potential
need for drainage are drained before areas with less need for drainage.
Agricultural land in lowland areas has been identified by combining the land
use map with agricultural land and the national map of “extended” lowlands.
Areas with a high density of ditches have been identified from the FOT GIS
theme for streams, which has been combined with the numerical grid of the
DK-model, Figure 3.2. Grids with more than five stream reaches and a total
length of more than 1,500 m has been designated as areas where drainage was
carried out by ditches.
Figure 3.2.
Illustration of a grid in
the DK-model with a high density
of streams in the FOT dataset
where natural drainage by
ditches has been assumed.
For the remaining agricultural land, tile drainage has been distributed in ac-
cordance to the drained area within each region. For the Danish islands, the
delineation of the regions coincides with the land areas of the islands. The extent
56
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0059.png
of the regions south-, west-, east- and north-Jutland is not reported in the drain-
age reference, i.e. Aslyng, 1980. The regions have thus been defined by:
1. South Jutland, the area between the border to Germany up to 1920 and the
current border.
2. North Jutland, identical with the current region “Nordjylland”.
3. The Mid Jutland ridge was used to divide the remaining part of Jutland
into an eastern and a western part. This division separates Jutland into a
predominantly sandy part in the west, with generally low drainage inten-
sity, and an eastern part dominated by clay/till and high drainage density.
The regions in Jutland are illustrated in Figure 3.3.
Figure 3.3.
Regions in Jutland
used to distribute historical tile
drains.
No data on tile drainage area are available for the southern part of Jutland.
The geologically east-west division in Jutland extends south to the present
border to Germany. It was thus assumed that the drainage density in the west-
ern and eastern parts of South Jutland equalled that of West and East Jutland,
representing 24 and 9% of the agricultural land, respectively.
57
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0060.png
Within each region, the tile drains were distributed by associating drainage
with the areas with highest probability of being drained. For this purpose, the
drainage probability map (Møller et al., 2018) was used, which is developed
from statistical models describing the probability of an area being drained
based on selected explanatory variables. The final distribution of the drainage
into the three drainage categories is shown in Figure 3.4.
Figure 3.4.
Final distribution of
drains in the model representing
year 1900.
Abstraction
All abstractions (locations and amounts) are reported to the national ground-
water database JUPITER, from where data are extracted for the current situa-
tion. Around year 1900, waterworks started to become established in the
larger cities. At that time, city development was rapid, including a growing
number of waterworks. This project did not allow a detailed reconstruction of
the development of waterworks in the country and the amount of water ab-
stracted. It was therefore assumed that the water abstraction was 20% of the
current level and that the structure of the abstraction has not changed, i.e. the
city structure then and today is the same.
Discharge from point sources
Discharge from point sources such as wastewater treatment plants is in-
cluded as a source to the simulated stream discharge. For the current condi-
tions, the data are collected from national databases. No historical data are
58
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0061.png
available, and it was therefore assumed that discharge, as abstraction, is 20%
of the current amount and has a distribution identical to that of the present.
3.3
Results
The hydrological model was used to calculate the historical stream discharge
based on the two approaches for establishing a historical dataset on precipi-
tation: 1) correction of measured precipitation amounts based on estimates of
wind speed and shelter index and 2) the delta change approach.
3.3.1 Historical precipitation from corrected observations
Test of the historical precipitation calculated by correcting observed data was
achieved by comparing observed and simulated stream discharges for the pe-
riod 1917-1950. Within this period, data were available from 41 stations having
observations for two or more years. Basic statistics from the discharge stations
are found in Table 3.3, and their spatial location is shown in Figure 3.5.
Figure 3.5.
Location of stream
discharge stations and catch-
ments with data in the period
1917-1950.
59
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0062.png
Table 9.3.
Stream discharge stations with data from the period 1917-1950.
Discharge station no.
50000050
52000025
52000029
52000189
55000017
55000018
56000006
56000007
57000049
57000050
57000058
57000121
59000006
61000012
44000034
44000035
45000002
45000004
46000030
31000044
32000001
35000142
38000024
40000001
21000084
21000085
21000634
21000792
21000794
24000001
25000018
25000082
25000249
26000082
3000003
7000003
11000012
11000016
14000022
17000018
66000014
Mean
discharge (l/s)
103
130
442
483
660
1,604
81
779
383
3,728
5,068
387
852
181
220
133
4,421
2,726
555
6,261
3,542
255
6,920
2,971
15,456
2,370
876
275
2,591
538
1,316
13,431
97
1,042
1,451
895
2,667
1,226
2,297
1,344
300
First year
1943
1947
1948
1930
1945
1921
1922
1933
1919
1943
1935
1944
1918
1932
1926
1926
1931
1918
1919
1930
1917
1920
1933
1922
1917
1918
1931
1918
1918
1933
1948
1923
1949
1920
1917
1918
1918
1936
1926
1918
1922
Last year
1950
1950
1950
1948
1950
1950
1950
1950
1950
1944
1950
1950
1950
1950
1950
1950
1950
1950
1950
1944
1950
1950
1950
1950
1950
1950
1939
1939
1950
1950
1950
1950
1950
1950
1950
1950
1931
1950
1950
1949
1950
Observations
Years with data
8
4
3
15
6
29
27
16
31
2
16
7
33
19
20
19
20
33
31
3
34
24
18
29
34
33
7
22
33
18
3
26
2
31
34
33
14
15
25
26
29
60
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0063.png
Good agreement was found for several of the model domains, while a general
water balance error was found for other domains, i.e. the model simulated too
much or too little stream discharge, indicating that the historical precipitation
was over- or underestimated.
Several alternative approaches to calculate precipitation was tested, see chap-
ter 1 for a description of the alternative approaches to correct the precipitation.
Table 3.10 shows the total water balances calculated for each of the model do-
mains using the precipitation based on the assumptions expected to represent
the historical conditions best (best estimate). The water balance error for Fu-
nen, South Jutland and North Jutland was less than 10%, which is considered
acceptable given the uncertainty in estimating the historical period. On Zea-
land and Lolland-Falster, the water balance was 24 and 47% off, the simulated
values were higher than the observed values (negative water balance), indi-
cating that the estimated precipitation is overestimated. Furthermore, the sta-
tions on Zealand showed a clear trend as 11 out of 13 stations on Zealand
showed a negative water balance. In the model domain Mid Jutland, the trend
was opposite; there was a total water balance error of 17%, and eight out of
ten stations showed a positive water balance, indicating too little precipitation
under historical conditions. Finally, Bornholm also showed a high positive
water balance error of 34%.
Within the framework of the project, it was not possible to improve the match
between the observed and simulated water flow for all areas simultaneously
by applying a different scheme to correct the precipitation for the entire coun-
try. As the nitrate leaching in 1900 is calculated from the amount of water
recharging the subsurface and the nitrate concentrations estimated in chapter
5, an error in the water balance will invariably result in an error in the esti-
mated nitrate leaching.
Table 3.10.
Water balance error (%) for the seven model domains in the DK-model for the period 1917-1950 using “best esti-
mate” and adjusted precipitation.
Precipitation
scenario
Best estimate
Number of stations
-24
13
Zealand
Lolland-
Falster
-47
1
2
5
Funen
South
Jutland
-6
5
Mid
Jutland
17
10
North
Jutland
9
6
34
1
3
41
Bornholm
Country
3.3.2 Historical precipitation using the delta change approach
Using the delta change approach, precipitation is corrected to the period 1890-
1910 at a monthly time scale. No stream discharge data are available, and the
simulated discharge cannot be compared with the observed counterparts. It
is thus assumed that the approach used to correct precipitation to the period
1917-1950, which provided a small water balance error at national scale, can
also be used to correct precipitation to the period 1890-1910 with an acceptable
national water balance.
The mean simulated discharge to the sea in the period 1890-1910 is shown in
Figure 3.6 (middle) for fourth order streams, where also the present period,
1990-2010 (left) and – for comparison – the absolute difference between the
two time periods (right) are displayed. As seen, the delta change approach
preserves the overall national pattern with a higher discharge towards the
west. However, the pattern is not fully identical for the two periods. This is
61
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0064.png
due to differences in the local weather conditions. In the delta change ap-
proach, the present precipitation data are corrected using the same monthly
correction factor for the entire country. The relative variation in monthly pre-
cipitation is nevertheless not the same all over the country, and the effect of
the monthly correction will thus vary. Similarly, the calculated potential evap-
otranspiration may contribute to a different spatial pattern; here, a simple ap-
proach has been used based on daily minimum and maximum temperatures,
which may not capture the local variations correctly and to the same degree
as observed in the present potential evapotranspiration. Finally, the changes
in agricultural practices, i.e. cultivated areas and drainage density, will have
affected the water cycle and thereby also the stream discharge.
Figure 3.6.
Simulated discharge to the sea aggregated to fourth order catchments for the period 1900 (left) and the present
(centre). Right: Changes in discharge as mm/yr for the two periods.
Differences in the simulated historical and present stream discharge are
shown in Figure 3.6 (right); negative values indicate an increase in stream dis-
charge from 1900 until now. In the eastern part of the country (most parts of
Zealand, Lolland, Falster and Bornholm), there is either no or a modest change
the in stream discharge from 1900 until now, and in some areas a decrease is
found. This is a combination of the changes in precipitation calculated by the
delta change approach, where almost no increase is observed in the eastern
part of the country (Figure 1.12), and the reduced drainage density results in
a higher groundwater table and more evapotranspiration. Towards the west,
the increase in precipitation is higher, which is also reflected in the enhanced
stream flow.
A total discharge to the sea of 333 mm/yr is simulated for the present condi-
tions, while a discharge of 292 mm/yr is simulated for the historical period.
At national scale, the stream discharge is thus estimated to increase by 45
mm/yr, which corresponds to a 12% lower discharge than today. The per-
centage changes in stream discharge aggregated for fourth order catchments
are shown in Figure 3.7. A negative increase in discharge is found for the west-
ern part of Zealand and Lolland, i.e. a decrease in the discharge, which is gen-
erally below a 5% change. For some of the smallest catchments, a decrease in
stream discharge of more than 20% occurs. These small catchments are only
62
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0065.png
covered by a single precipitation value (one climate grid) and are thus very
sensitive to the local precipitation estimates. Expectedly, the calculated
change in discharge is therefore more uncertain for small catchments.
Historical changes in discharge were previously estimated by Jensen (ed.)
(2017) based on linear trend analysis of long time series with data from mini-
mum 80 years. The results are given in Figure 3.7 (right), showing the calcu-
lated changes between 1935 and 2015 in mm/yr and as percentages. In Jensen
(ed.) (2017), the statistical trend model was used to back-write discharge to
the year 1900. However, the analysis in chapter 1 of the development of pre-
cipitation indicates that the largest change in precipitation probably occurred
after 1950. Hence, it may be questioned if the discharge in year 1900 can be
back-written to 1900 using a linear model based on data from 1935-2015. For
this reason, the trend analysis for the period 1935-2015 has been used for com-
paring the results obtained by the delta change approach.
The simulated percentage changes in discharge for the main parts of Funen
range between 0 and 20%, which corresponds well with the trend analysis.
Close agreement is further seen for Mid and North Jutland, while the trend
analysis indicates an up to 33% increase in discharge in the southern part of
Jutland, where the simulated change is maximum 20%. Discrepancies be-
tween the two approaches are also seen for the western part of Zealand. Here,
the trend analysis estimates an increase between 18-28%, whereas a decrease
is simulated using the delta change-corrected precipitation data. No data are
available for Bornholm preventing comparison with the simulation results.
Figure 3.7.
Change in simulated stream discharge from 1900 to the present as percentage (left) and calculated changes in pre-
cipitation and stream discharge from trend analysis of time series (right) (from Jensen (ed.), 2017).
63
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
3.4
Conclusion
Applying the historical precipitation calculated from observed data corrected
for wind and shelter effects in the hydrological model resulted in only 3%
difference in the total observed and simulated discharges for the period 1917-
1950. The simulation similarly showed good agreement for several discharge
stations, but for Mid Jutland and the eastern part of Denmark the historical
discharge was under- and overestimated, respectively. Additionally, the wa-
ter balance errors between observed and simulated discharges for streams
within a region displayed noticeable variations. Despite the test of several al-
ternative approaches to correct the historical precipitation data, no approach
reproducing the spatial variation in stream discharge could be found. There-
fore, it was decided to apply a delta change approach by which a historical
time series of precipitation was established by adjusting the current precipi-
tation values by a national monthly correction factor. In this way, the same
factor, varying from month to month, was applied to the entire country. This
approach exploits the fact that the national water balance was low, while sim-
ultaneously acknowledging that the spatial variation could not be repro-
duced.
Using the delta change approach, the total discharge was found to increase
from an average value of 292 mm/yr for the period 1890-1910 to 333 mm/yr
for the present period (1990-2010). This change in discharge between the two
periods largely reflects the changes in precipitation, but for some areas it is
amplified due to the lower density of drains in the historical period. The cal-
culated change in discharge from 1900 to the present ranged between 0 and
20% for most of the country, which agrees with the trend analysis of long dis-
charge time-series conducted by Jensen (ed.) (2017). However, for South Jut-
land, the discharge development is underestimated as the observed data in-
dicates an increase in discharge just above 30%. For the western part of Zea-
land, the development in stream discharge is similarly underestimated as the
discharge observations indicate an increase of up to 28%, while the simulation
resulted in a decrease in discharge of approx. 5%. With the delta change
method it is thus assumed that the total discharge to the sea resembles the
conditions around year 1900, but local conditions cannot be reproduced.
Estimation of precipitation is crucial as this, together with evapotranspiration,
is the driving force for calculating the recharge to the groundwater. Uncer-
tainty regarding the estimated recharge will result in uncertainty as for the
groundwater pathways and, thus, the transport pathways of nitrate from the
root zone to the sea. Even more critical is it that the total mass of nitrate leach-
ing from the root zone is calculated from the root zone concentration esti-
mated in chapter 5 multiplied by the estimated recharge. Uncertainty regard-
ing the recharge will thus translate to the estimated total nitrogen load to the
subsurface in year 1900. From this follows that the estimates of nitrate leach-
ing, in accordance with the estimated stream discharges, cannot be used at
local scale.
3.5
References
Aslyng, H.C., 1980. Afvanding i jordbruget. København, DSR Forlag, Den
Kongelige Veterinær- og Landbohøjskole.
Havnø, K., Madsen, M.N. and Dørge, J., 1995. MIKE 1—a generalized river
modelling pack-age. In: Singh, V.P., (Ed.), Computer Models of Watershed
Hydrology, Water Resources Publications, 733-782.
64
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0067.png
Henriksen, H.J., Troldborg, L., Nyegaard, P., Sonnenborg T.O., Refsgaard, J.
C. and Madsen, B., 2003, Methodology for construction, calibration and vali-
dation of a national hydrological model for Denmark. Journal of Hydrology
280, 52-71. doi:10.1016/S0022-1694(03)00186-0
Højberg, A.L., Stisen, S., Olsen, M., Troldborg, L., Uglebjerg, T.B. and Jørgen-
sen, L.F., 2015a. DK-model2014 - Model opdatering og kalibrering, Danmarks
og Grønlands Geologiske Undersøgelse Rapport 2015/8, 117pp.
Højberg, A.L., Windolf, J., Børgesen, C.D., Troldborg, L., Tornbjerg, H., Bli-
cher-Mathiesen, G., Kronvang, B., Thodsen, H. and Ernstsen, V., 2015b. Nati-
onal kvælstofmodel, Oplandsmodel til belastning og virkemidler. Metode
rapport - Revideret udgave september 2015. GEUS, 111 s.
Højberg, A.L., Troldborg, L., Stisen, S., Christensen, B.B.S. and Henriksen,
H.J., 2013, Stakeholder driven update and improvement of a national water
resources model, Environmental Modelling & Software 40, 202-213.
doi:10.1016/j.envsoft.2012.09.010
Jensen, P.N. (Ed.), 2017. Estimation of Nitrogen Concentrations from root
zone to marine areas around the year 1900. Aarhus University, DCE – Danish
Centre for Environment and Energy, 126 pp. Scientific Report from DCE –
Danish Centre for Environment and Energy No. 241.
http://dce2.au.dk/pub/SR241.pdf
Møller, A.B., Børgesen, C.D., Bach, E.O., Iversen, B. and Moeslund, B., 2018.
Kortlægning af drænede arealer i Danmark. DCA rapport, 135, 131pp.
Yan J.J. and Smith, K.R., 1994, Simulation of Integrated Surface Water and
Ground Water Systems - Model Formulation. Water Resources Bulletin 30, 1-
12.
65
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
4
Point source emissions of nutrients from
urban areas in 1900
Author: Sarah Brudler
1
, Karsten Arnbjerg-Nielsen
1
, Camilla Bitsch
1
, Mikkel
Thelle
2
, Martin Rygaard
1
Ouality assurance: Peter Steen Mikkelsen
1
DTU Sustain, Department of Environmental and Resource Engineering, Tech-
nical University of Denmark
2
School of Culture and Society, Aarhus University
1
This chapter is based on the article: Brudler, S., Arnbjerg-Nielsen, K., Jeppesen,
E.B., Bitsch, C., Thelle, M. and Rygaard, M., 2020. Urban nutrient emissions in
Denmark in the year 1900.
Water 12,
789. doi:10.3390/w12030789
Abstract
Purpose:
The purpose of this chapter is to quantify nitrogen and phosphorous
emissions to water and soil from Danish towns around the year 1900.
Materials and methods:
Based on an extensive literature review, qualitative
information regarding discharge paths is translated into quantitative mass
flow charts. The resulting emissions to different environmental compartments
are then calculated using adjusted current measurements of nutrient contents
in human and animal faeces and industrial wastewater.
Results and discussion:
Total nutrient emissions are estimated to 4,261 ton
N/yr and 764 ton P/yr in 1900. The largest fraction was discharged to water
(2,531 ton N/yr and 462 ton P/yr), followed by landfills (811 ton N/yr and
143 ton P/yr) and agricultural soil (919 ton N/yr and 159 ton P/yr).
4.1
Introduction
Large parts of the Danish population migrated to towns in the second half of
the 19
th
century, and industries centred around towns were growing
(Statistics Denmark, 2000). By 1901, 39% of Denmark’s 2.45 million population
lived in towns (Matthiessen, 1985). Even though agriculture was not the main
source of income for the majority of the town population, animals were often
held for self-supply in the beginning of the 20
th
century (Mikkelsen, 2012,
2010; Statistics Denmark, 1969). Industries and factories developed around the
towns. With this growing pressure on towns, hygiene and sanitation gained
increasing attention in the 19
th
century, and efforts were put into limiting hu-
man contact with excrements and waste (Iversen, 2004).
However, this development varied significantly between towns. While under-
ground sewers were largely installed in Copenhagen by 1900, smaller towns
still had buckets for human waste, and wastewater was discharged in open
gutters along the roads. The use of artificial fertiliser was still very limited
around year 1900 (Statistics Denmark, 1968), and human and animal excre-
ments were a valuable resource that was used as fertiliser on farms. Emissions
from industries were largely unregulated, and wastewater treatment did not
exist (Engberg, 1999). Discharges around year 1900 of excrements and waste
66
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
from humans, animals and industries in Danish towns have not been quanti-
fied until now, and the magnitude of nutrient emissions to streams and ma-
rine waters is largely unknown. Here, a first estimate of total nitrogen (N) and
phosphorous (P) emissions from Danish towns around the year 1900 is pro-
vided. The assessment is based on national upscaling based on representative
towns for which detailed historical data have been collected and detailed es-
timates of nutrient mass flows are made.
4.2
Materials and methods
Individual assessment of each town lies outside the scope of this report. In-
stead, three model towns with representative characteristics were chosen and
analysed in more detail to derive typical conditions. Total emissions from
Danish towns with more than 5,000 inhabitants are then calculated by upscal-
ing the assessed characteristics of the model towns. The chosen cut-off is
based on the assumed demographic difference that smaller towns with less
than 5,000 inhabitants were more rural and less developed in terms of sanita-
tion than larger towns. Towns larger than 5,000 inhabitants covered approxi-
mately 80% of the urban population in 1900. 24 out of 29 towns larger than
5,000 inhabitants around year 1900 are coastal towns.
Copenhagen was assessed separately as it was distinctively different from all
other Danish towns as to size and level of industrialisation. Almost 500,000
people lived in Copenhagen around 1900, which accounts for 45% of the Dan-
ish town population (19% of the total population) (Matthiessen, 1985). Apart-
ments were scarce and overcrowded and industries were growing, which led
to faster development of organised waste disposal and underground sewer
systems than in the remaining part of Denmark (Iversen, 2004).
Besides Copenhagen, three model towns were selected to reflect the typical
population growth and the fraction of population working in agriculture
among Danish towns around year 1900. Towns with more than 5,000 inhabit-
ants showed widely varying characteristics regarding these two parameters,
population growth ranging between -6% and +773% between 1880 and 1901
(median; +58%) and the fraction of the town population working in agricul-
ture ranging between 1% to 18% in 1890 (median: 6%) (Dansk Center for
Byhistorie, n.d.; Matthiessen, 1985). Towns and settlements with less than
5,000 inhabitants had a markedly lower population growth (29%), and occu-
pation was dominated by agriculture and fishery (62%) (Dansk Center for
Byhistorie, n.d.; Matthiessen, 1985).
Besides proximity to calculated median values for both population growth
and occupational pattern (appendix 4.1), geographical distribution was con-
sidered, and based on this, Svendborg (Funen), Helsingør (Zealand) and
Randers (Jutland) were selected as model towns (Table 41). All three towns,
like most larger towns at the time, were connected to the railway, enabling
trade and industrial growth (Fertner, 2012). Furthermore, access to the sea
was an important factor for growth and industrialisation, which was the case
for most larger towns (Figure 4.1).
67
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0070.png
Table 4.1.
Median characteristics of Danish towns with more than 5,000 inhabitants, of three Danish model towns and of Co-
penhagen.
Population size
1900
(Matthiessen, 1985)
Median
Svendborg
Helsingør
Randers
Copenhagen
8,958
11,543
13,902
21,377
491,278
Population growth between
1880 and 1901
(Matthiessen, 1985)
+58%
+61%
+55%
+49%
+74%
Fraction of the population working in
agriculture or fishery in 1890
(Dansk Center for Byhistorie, n.d.)
6%
7%
8%
11%
1%
-
Funen
Zealand
Jutland
Zealand
Location
The inhabitants of the 29 towns with more than 5,000 inhabitants (Table 4.1)
cover 80% of the population living in towns (34% of the total Danish popula-
tion) (Matthiessen, 1985). Data on the numbers of cattle, pigs and sheep held
within the towns were gathered (Trap, 1906). For the three model towns, in-
formation regarding the infrastructure, regulations and processes to collect
and dispose of waste from humans, animals and industry was collected from
historical documents.
Slaughterhouses, dairies and tanneries existed in all three model towns and
Copenhagen. Other types of industry and factories varied between the towns
and are not analysed in detail in this report. Tanneries were assumed to
mainly emit toxic substances, especially chromium, and were therefore not
considered as a significant nutrient source. Current data on wastewater quan-
tity and quality from slaughterhouses and dairies were used to estimate emis-
sions around year 1900. In addition to emissions from larger towns, the emis-
sions from slaughterhouses in more rural areas of Denmark were also as-
sessed. Slaughterhouses were mainly located in smaller towns with less than
5,000 inhabitants and were therefore included as separate point sources in the
assessment (appendix 4.1). In general, food production resources were used
as efficiently as possible, and by-products were often reused. For instance, the
organic waste of breweries was fed to cows, and the skimmed milk was redis-
tributed from dairies to farms as feed for animals (Iversen, 2017; Statistics
Denmark, 1969). This is assumed to limit the nutrient emissions of industry
around year 1900. Finally, a distinction was made between coastal and inland
towns, where coastal towns were characterised by access to a major water
body and thereby easy access to discharge of sewage.
Based on the collected information, the flows of nutrients from humans, ani-
mals and industries were compiled in flow charts for coastal and inland towns
and for Copenhagen. Randers and Odense can be considered inland towns,
but their proximity to large water bodies led to their categorisation as coastal
towns. In the following, initial concentrations refer to the direct nutrient dis-
charge from humans and animals, i.e. what was emitted in excrements. The
initial concentrations of nitrogen and phosphorous from humans and animals
and in wastewater from industry were based on current values reported in
the literature and adjusted by the assumed change in nutrient intake since the
time around year 1900 (appendix 4.1). Based on the developed flow charts,
the resulting emissions to water, landfills and agricultural soil were calculated
for each town, including dairies and each slaughterhouse.
68
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0071.png
Figure 4.1.
Danish towns with
more than 5,000 inhabitants in
1900. Odense and Randers are
considered to have the character-
istics of coastal towns because of
their close proximity to major wa-
ter bodies.
4.3
Results and discussion
4.3.1 Latrines and dumpsites
Most Danish towns changed from simple pit latrines to collect human excre-
ments (Danish:
grubelatrine)
towards a bucket system in the 19
th
century (Dan-
ish:
tøndelatrine)
(Gædeken et al., 1894; Kongstad, 2016). In 1899, 44 Danish
towns (including all three model towns) had, at least partially, introduced a
bucket system, with only 29 still using pits (Carlsen, 1900). Collection of the
buckets was rarely organised and regulated. Buckets were often emptied at
small dumpsites within the towns (Danish:
mødding),
where also animal excre-
ments were disposed. The transport of excrements to the dumpsites often re-
sulted in significant spills (Steensberg, 1964), which were included as emissions
to soil (Figure 4.2). The dumpsites were emptied by the surrounding farmers
only one to two times per year (Gædeken et al., 1894). Only some towns, for
instance Randers, had organised bucket collection and disposal by private com-
panies that sold the buckets to surrounding farmers (Hyldegaard, 2002).
Water closets (WC) were not common around year 1900, with only 44 WCs re-
ported in Helsingør (Pedersen, n.d.) and approximately 100 in Randers. They
were connected to septic tanks that often overflowed (Hyldegaard, 2002). In Co-
penhagen, it was legalised to connect WCs to the subsurface sewer system in
1897, which led to a faster technology uptake, resulting in around 5,000 WCs in
1900 (Lützen, 1998). The majority of the Copenhagen population still relied on
69
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
buckets that were collected at night, transported to four dumpsites outside of
Copenhagen and sold to farmers (Iversen, 2017). The fraction of sold buckets
decreased steadily at the end of the century; thus, less than 10% of approxi-
mately 30,000 buckets were sold in 1900 (Gædeken et al., 1894).
4.3.2 Gutters and pipes
We assume that the preferred way of disposal was direct discharge to water
preventing the hygiene and odour problems caused by landfills. Wastewater
was mainly discharged to receiving water bodies in gutters, which replaced
ditches in the 19
th
century in order to reduce infiltration of wastewater
(Hyldegaard, 2002; Kongstad, 2016). The installation of pipes began at the end
of the 19
th
century in most Danish towns, including all three model towns
(Bro-Jørgensen, 1959; Hyldegaard, 2002; Kongstad, 2016). This was often done
in combination with the construction of water works and water distribution
pipes. Around year 1900, 62 water works had been constructed in larger
towns, including all the three model towns (Foreningen af Vandværker i
Danmark, 2012; Trap, 1906). The construction of sewer pipes often started in
the town centre and in connection with specific problematic industries, as for
instance reported for Esbjerg (Esbjerg Kommune, 2019). The update of this
technology varied widely in 1900 between complete absence of sewers to com-
plete coverage, for example in Roskilde (Fang, 2017; Kongstad, 2016). Pipes
and gutters were often connected to the nearest surface water body, for
instance to the River Gudenå in Randers (Jensen, 1989). Direct discharges to
water, especially from industries, were very common in the 19
th
century, as
documented, for example, for tanneries and slaughterhouses (Christensen,
1912; Iversen, 2017). If direct connection to surface water was not feasible, the
wastewater was discharged to dumpsites outside the town as was the case for
the town of Svendborg (Bro-Jørgensen, 1959).
Copenhagen had an extensive sewer system already around year 1900, into
which human waste and industrial wastewater were discharged (Eriksen, 2007;
Hanne Lindegaard, 2001). Even before it was legalised to discharge human ex-
crements to the sewers in 1897, it was estimated that 75% of the excrements was
discharged illegally by either disposing of excrements through kitchen pipes or
by throwing them on the streets (Hanne Lindegaard, 2001; Iversen, 2017). The
pipes and gutters discharged into canals, lakes and the harbour with a direct
connection to the open sea (Gædeken et al., 1894). The canals and lakes were
malodorous, and the sea between Køge and Copenhagen was so polluted that
it was brown at that time (Andersen, 2012; Hilden, 1973).
70
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0073.png
Figure 4.2.
Flow charts of nutrients in human, animal and industry waste in Danish towns in 1900. Losses are considered emitted
to soil on-site due to spill and infiltration (emissions to soil) or disposed of in landfills without further export to agricultural soil. For
example, town landfills would lose 20% nutrients to soil on site, and for Copenhagen landfills 90% of nutrients was lost on site.
4.3.3 Initial emissions
Initial emissions refer to nutrients excreted by humans, animals and industries.
Reported emissions of nutrients in human excrements are around 4 kg N/pers.
yr and 0.7 kg P/pers. yr (Hevesy et al., 1939; Holtze and Backlund, 2003; Rose
et al., 2015; Wrisberg et al., 2001). Nutrition was significantly less varied, and
the calorie intake was around 25% lower than today (Roser and Ritchie, 2019;
Staun, 2002), which also resulted in lower intake of nutrients (appendix 4.1) The
historical values were therefore adjusted, and nutrient emissions per person
was assumed to be 25% lower around year 1900 than today (Table 4.2).
71
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0074.png
The excretion of nutrients from animals depends on the ability to process nu-
trients and the feed. Protein-rich feed reduces the nitrogen demand and ex-
cretion, which such feed was not known or applied around year 1900
(Manitoba, 2015). We therefore also assume a 25% decrease in addition to us-
ing the minimum emissions reported for present-day conditions (Table 4.2)
(Hong et al., 2012).
As piped systems were not yet implemented, wastewater production at slaugh-
terhouses and dairies was assumed to be limited. Therefore, the lowest
wastewater production reported in the literature for present conditions com-
bined with the highest reported nutrient concentrations was used (Table 4.2)
(Rad and Lewis, 2013; Verheijen et al., 1996). Emissions from single slaughter-
houses were calculated by evenly distributing the mass of all slaughtered ani-
mals in 1900 over all 29 slaughterhouses, with the exception of Randers, Esbjerg
and Odense. These towns had larger slaughterhouses, twice the throughput was
therefore assumed (appendix 4.1) (Krak, 1950; Statistics Denmark, 1969). To de-
rive the emissions from dairies, the per person consumption of milk were calcu-
lated by dividing the total production by the total population (Matthiessen, 1985;
Statistics Denmark, 1969). As there were in total approximately 1,000 dairies in
Denmark, it was assumed that there was a dairy processing milk for all inhabit-
ants in every town (spreadsheet appendix) (Christensen, 2012).
Table 4.2.
Estimated emission of nutrients from humans, animals and industry in 1900 based on present-day values reduced
by 25% to account for differences in nutrition and animal feed.
Nitrogen
Human excrements
Animal excrements
Cattle and horses
Pigs
Sheep and goats
Industry
Dairies
Slaughterhouses
0.3 kg/pers./y.
27 kg/slaughterhouse/yr
0.1 kg/pers./yr
2 kg/slaughterhouse/yr
45.0 kg/animal/yr
6.8 kg/animal/yr
6.5 kg/animal/yr
7.5 kg/animal/yr
2.3 kg/animal/y.
1.3 kg/animal/yr
3.0 kg/pers./y.
Phosphorous
0.5 kg/pers./yr
4.3.4 Point source nutrient emissions 1900
The total nitrogen emissions from towns and slaughterhouses were 4,261
ton/yr, and the total phosphorous emissions were 764 ton/yr. Most of the
emissions was discharged to water (2,531 ton N/yr, and 462 ton P/yr.. Emis-
sions to landfills and soil (811 ton N/yr and 143 ton P/yr) and agricultural
soil (919 ton N/yr and 159 ton P/yr) are significant (Table 4.3). The estimated
emission of nutrients from humans, animals and industry around year 1900,
based on present-day values, was reduced by 25% to account for differences
in nutrition and animal feed. Slaughterhouses contributed insignificantly to
total emissions (Table 4.3).
72
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0075.png
Table 4.3.
Nutrient emissions to water, landfill or soil and agricultural soil from coastal towns, inland towns, Copenhagen and
slaughterhouses around year 1900. Emissions to landfills or soil include spills during transport and overflows.
Emission to
Emission at landfill or on site soil (not Emission to agricul-
tural soil [ton/yr]
P
54
21
68
0
143
N
600
280
40
0
919
Total
[ton/yr]
P
102
50
7
0
159
N
2,091
407
1,764
1
4,261
P
374
73
317
0
764
water [ton/yr] reaching agricultural soil) [ton/yr]
N
Coastal towns
Inland towns
Copenhagen
Slaughterhouses
SUM
1,182
13
1,336
1
2,531
P
218
2
242
0
462
N
309
114
388
0
811
In 2017, the estimated level of phosphorous in domestic wastewater was 0.7
kg P/pers./yr in Denmark (Arildsen and Vezzaro, 2019) This is mainly due
to the increased use of phosphorous in products like detergents, while around
year 1900 only excrements contributed to the emissions of phosphorous from
humans. While more than 80% of phosphorous is removed in wastewater
treatment plants today (Danish Nature Agency, 2014), a large fraction of hu-
man excrements was discharged directly to water around year 1900 (Figure
4.3). The resulting per capita emissions for discharges of phosphorous from
humans to water were estimated to 0.11 kg P/pers. yr today. For the time
around year 1900, it was found that, after losses (Figure 4.3,) discharges to
water were 0.05 kg P/pers./yr from inland towns and 0.24 kg P/pers./yr
from coastal towns.
Figure 4.3.
Total nitrogen and phosphorous emissions from humans and animals in towns with more than 5,000 inhabitants and
industry (slaughterhouses and dairies) to water, landfill (emissions to soil and not reaching agricultural soil) and agricultural soil
around year 1900.
Most nutrients were emitted in human waste, contributing almost twice as
much as animal excrements. The emissions from industry were one order of
magnitude lower than the emissions from humans or animals (Figure 4.3).
Copenhagen was the largest single point source of nitrogen and phosphorous
(53%), followed by coastal towns (47%). Less than 1% of the total nutrient
emissions stemmed from inland towns (Figure 4.4).
73
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0076.png
Figure 4.4.
Nutrient emissions to
water from coastal and inland
towns with more than 5,000 in-
habitants and Copenhagen (in-
cluding humans, animals and in-
dustry).
4.3.5 Main uncertainties
The flow paths of nutrients, the initial load of nutrients from human and ani-
mal excrements and the minimum population for inclusion of towns in the
assessment were identified as the main sources of uncertainty. For each
source, alternative scenarios and assumptions were tested and the resulting
nutrient emissions compared with the baseline scenario (Table 4.4, results
given in appendix 4.2).
Table 4.4.
Tested alternative scenarios for selected parameters.
Source of uncertainty
1. Flow paths of nutrients
Parameter
gen sold to farmers
1.b Fraction of waste disposed to pipes 10-20% of human excre- 50% decrease
ments
10% of industrial
wastewater
1.c Same as 1.b
man and animal excrements
3. Minimum population for inclu-
sion of towns in the assessment
and animal excrements
2.b Same as 2.a
ants
3.b Same as 3.a
Same as 3.a
Included as coastal towns
Same as 2.a
100% of present values
Included as inland towns
3.a Towns with less than 5,000 inhabit- Excluded
Same as 1.b
50% increase
50% of present values
2. Initial load of nutrients from hu- 2.a Initial load of nutrients from human 75% of present values
Baseline
Alternative
80%
1.a Fraction of buckets from Copenha- 10%
Changes in the initial load of nutrients in animal and human excrements was
the most significant factor affecting the resulting nutrient emissions. The excre-
tion of nutrients correlates directly to the nutrient intake, which is a highly un-
certain parameter as no measured values exists or the time around year 1900.
In the baseline scenario, a 25% reduction of the values reported for the present
conditions was assumed. Assuming no reduction of initial loads results in a 30%
increase in emissions. Assuming an even more significant reduction of 50% of
initial loads decreases the emissions by 30%. The inclusion of smaller towns
with less than 5,000 inhabitants leads to a maximum increase of 12% in the re-
sulting emissions to freshwater. Flow path changes did not cause significant
changes in the subsequent nutrient emissions to water (-4% to +4%). However,
the flow paths for solid and liquid waste, carrying different fractions of the ini-
tial nutrient loads, were not differentiated. Infiltration of urine possibly led to
higher emissions of nitrogen than of phosphorous to soil, but a detailed assess-
ment of the different paths lies outside the scope of this assessment.
74
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0077.png
Figure 4.5.
Relative change in
nutrient emissions to water com-
pared with the baseline of seven
alternative scenarios listed in ta-
ble 4.4.
The calculated emissions are based on the assumed average conditions in
Danish towns around year 1900. However, towns were developing very dif-
ferently, and the source and fate of emissions varied widely. It was not the
scope of our study to conduct a detailed mapping of nutrient emissions and
routing for each individual town, and our results must therefore be inter-
preted with caution for individual towns. The values are, as such, a qualified
indication of the magnitude of nutrient loads from point sources around year
1900 and a robust national estimate of emissions. A more detailed assessment
of each individual town and industry could be made based on specific mass
balances. This would improve the assessments of emissions to local surface
waters but is not expected to significantly affect the calculated national esti-
mate.
4.4
Conclusion
The disposal of excrements and wastewater differed widely between Danish
towns around year 1900. Hygiene and sanitation had gained increasing inter-
est, but sewer systems were still limited, and a large fraction of waste was
discharged in gutters. Discharges to receiving water bodies were not regu-
lated or treated, and a large fraction of waste was disposed of at landfills in
and around towns. The fate of nutrients differed between coastal and inland
towns, with more direct discharges to water along the coast and more dis-
charges to dumpsites and application on agricultural soil inland.
Towns were significant point sources around year 1900 with 4,261 ton N/yr
and 764 ton P/yr emitted in human and animal excrements and industrial
wastewater. These emissions are one order of magnitude lower than the esti-
mated emissions from agriculture. Our findings indicate that the majority of
the nutrients from point sources was discharged directly to receiving waters
(55%), but emissions to landfills (20%) and agricultural soil (25%) were signif-
icant as well. The contribution from slaughterhouses and dairies was found
to be negligible (<1%). 45% of the town population lived in Copenhagen in
1900, which led to large point source emissions (1,764 ton N/yr, 317 ton P/yr)
from. Most larger towns (>5,000 inhabitants) were located along the coast, and
the resulting total emissions (2,091 ton N/yr and 374 ton P/yr) were conse-
quently higher here than from inland towns (407 ton N/yr and 73 ton P/yr).
75
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
4.5
References
Andersen, U., 2012. Kampen om kloakkerne stopper aldrig [WWW
Document].
Ingeniøren.
URL
https://ing.dk/artikel/kampen-om-
kloakkerne-stopper-aldrig-135062 (accessed 2.21.19).
Arildsen, A.L. and Vezzaro, L., 2019. Revurdering af person ækvivalent for
fosfor - Opgørelse af fosforindholdet i dansk husholdningsspildevand i årene
fra 1990 til 2017. DTU Environment, Kongens Lyngby, Denmark.
Bro-Jørgensen, J.O., 1959. Svendborg købstads historie. Bind 2. Svendborg By,
Svendborg, Denmark.
Carlsen, J.N.G., 1900. Danmarks kultur ved aar 1900: som slutning af et
hundredaars udvikling. E. Bojesen, Copenhagen, Denmark.
Christensen, J., 2012. Fra andelsmejerier til Arla Foods 1882-2012 [WWW
Document].
URL
https://danmarkshistorien.dk/leksikon-og-
kilder/vis/materiale/fra-andelsmejerier-til-arla-foods-1882-2012/
Christensen, V., 1912. København I Kristian den Ottendes og Frederik den
Syvendes tid 1840-1857. G.E.C. Gads Forlag, København, Denmark.
Danish Nature Agency, 2014. Opdatering af nøgletal for miljøfarlige
forurenende stoffer i spildevand fra renseanlæg.
Dansk Center for Byhistorie, n.d. Danmarks Købstæder [WWW Document].
URL http://ddb.byhistorie.dk/koebstaeder/# indhold (accessed 2.18.19).
Engberg, J., 1999. Det Heles Vel. Forureningsbekæmpelsen i Danmark fra
loven om sundhedsvedtægter i 1850’erne til miljøloven 1974. Copenhagen,
Denmark.
Eriksen, J.M., 2007. Fra stinkende rendestene til computerstyrede kloakker:
Københavns kloakker gennem 150 år. Københavns Energi.
Esbjerg Kommune, 2019. Esbjerg Byhistoriske Arkiv [WWW Document]. URL
http://eba.esbjergkommune.dk/esbjergs-historie/tidslinje.aspx (accessed
2.20.19).
Fang, L., 2017. Roskilde - Begivenheder og deres årstal. Roskilde, Denmark.
Fertner, C., 2012. Digitalisering af det historiske jernbanenetværk. Jernbanen
6, 42-43.
Foreningen af Vandværker i Danmark, 2012. Vandsektorlovens betydning for
de private vandforsyninger.
Gædeken, C.G., Carøe, K., Hoff, E.M., Ulrik, A., Schierbeck, Ambt, C., Meyer,
J. and Rump, J., 1894. Renovationsvæsenet i København og danske
Provinsbyer. Særtryk af Tidsskrift for Sundhedspleje. København, Denmark.
Google, 2019. Google Maps [WWW Document]. Map data GeoBasis-DE7BKG.
URL maps.google.com (accessed 2.26.19).
76
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Lindegaard, H., 2001. Ud af røret. Planer, processer og paradokser omkring
det københavnske kloaksystem 1840-2001. Denmarks Tekniske Universitet.
Hevesy, G., Hahn, L. and Rebbe, O., 1939. Excretion of Phosphorus.
København, Denmark.
Hilden, H.P., 1973. Skrald, storby og miljø - En beretning om Københavns
kamp mod affaldet gennem 200 år. Renholdningsselskabet 1898, Copenhagen,
Denmark.
Holtze, A. and Backlund, A., 2003. Kompostering og eferkompostering af
humane restprodukter indehold i afvandet “sort” spildevand. Økologisk
byfornyelse og spildevandsrensning, Nr. 39. Copenhagen, Denmark.
Hong, B., Swaney, D.P., Mörth, C.M., Smedberg, E., Eriksson Hägg, H.,
Humborg, C., Howarth, R.W. and Bouraoui, F., 2012. Evaluating regional
variation of net anthropogenic nitrogen and phosphorus inputs
(NANI/NAPI), major drivers, nutrient retention pattern and management
implications in the multinational areas of Baltic Sea basin. Ecological
Modelling 227, 117-135. https://doi.org/10.1016/j.ecolmodel.2011.12.002
Hyldegaard, I.M., 2002. Randers. Fra Handelsplads til storkommune 1000-
1975.
Iversen, L., 2017. Kampen for liv - Fem danske forebyggelsessucceser. Statens
Institut for Folkesundhed, København, Denmark.
Iversen, U.O., 2004. Byens Plan og
Hygiehnebevægelsen og Byplanlægning i
Københavns Universitet.
Hygiejniske
København
Interesser.
1844-1865.
Jensen, P.E., 1989. Afløbsforhold og spildevandsplanlægning i Randers fra
1860-1930.
Kongstad, M., 2016. Skidt, smitteveje og sanitet. Kloakeringens betydning for
hygiejnen i Esbjerg ca. 1880-1904. Aarhus Universitet.
Krak, T., 1950. Fortegnelse over Danmarks ældste forretninger - Oprettet
mellem 1100 og 1911 som stadig var aktive i 1950.
Lützen, K., 1998. Byen Tæmmes. Kernefamilie, sociale reformer og velgø-
renhed i 1800-tallets København. Copenhagen, Denmark.
Manitoba, 2015. Properties of Manure.
Matthiessen, C.W., 1985. Danske byers folketal 1801-1981. Statistiske
Undersøgelser Nr. 42.
Mikkelsen, J., 2012. Købstadslandbruget og kampen om byernes naturlige
ressourcer. Landbohistorisk Tidsskrift 2, 9-44.
Mikkelsen, J., 2010. Danmarks Statistiks arkiv og landbohistorisk forskning.
Landbohistorisk Tidsskr. 2.
77
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0080.png
Pedersen, K., n.d. Helsingør Vandforsyning 1895-1995. Helsingør Kommunes
Tekniske Forvaltning.
Rad, S.J. and Lewis, M.J., 2013. Water utilisation, energy utilisation and waste
water management in the dairy industry: A review. International Journal of
Dairy Technology 67.
Rose, C., Parker, A., Jefferson, B. and Cartmell, E., 2015. The characterization
of feces and urine: A review of the literature to inform advanced treatment
technology. Critical. Review of Environmetntal Science and Technology 45,
1827-1879. https://doi.org/10.1080/10643389.2014.1000761
Roser, M. and Ritchie, H., 2019. Food per Person [WWW Document].
OurWorldInData.org.
Statistics Denmark, 2000. Befolkningen i 150 år. København, Denmark.
Statistics Denmark, 1969. Landbrugsstatistik 1900-1965. Bind II - Husdyrhold
og animalsk produktion samt foderforbrug. Statistiske Undersøgelser Nr. 25.
København, Denmark.
Statistics Denmark, 1968. Landbrugsstatistik 1900-1965. Bind I -
Landbrugsareal og høstudbytte samt gødningsforbrug. Statistiske
Undersøgelser Nr. 22. København, Denmark.
Staun, S.R., 2002. Familiens liv og hverdag år 1900: århundredeskiftet,
børneliv, mad og køkken, ægteskab, bolig, arbejde.
Steensberg, A., 1964. Dagligliv I Danmark I det nittende og tyvende
århundrede. Bind II. Nyt Nordisk Forlag.
Trap, J.P., 1906. Statistisk-topographisk Beskrivelse af Kongeriget Danmark.
3. Udgave 1898-1906. G.E.C. Gads Boghandel.
Verheijen, L.A.H.M., Wiersema, D., Pol, L.W.H. and Wit, J. De, 1996.
Management of Waste from Animal Product Processing.
Wrisberg, S., Eilersen, A.M., Nielsen, S.B., Clemmesen, K., Henze, M. and
Magid, J., 2001. Vurdering af muligheder og begrænsninger for recirkulering
af næringsstoffer fra by til land. Økologisk Byfornyelse og
Spildevandsrensning, Nr. 14. Copenhagen, Denmark.
78
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
5
Land use, agriculture and nitrate concen-
trations in root-zone percolates around
year 1900
Authors: Jørgen Eriksen
1
, Birger F. Pedersen
1
, Jørgen E. Olesen
1
and Bent T. Chris-
tensen
1
Quality assurance: Ingrid K. Thomsen
1
1
DCA,
Aarhus University, Department of Agroecology
Abstract
Purpose:
The purpose of this chapter is to provide estimates of the concentra-
tion of N in water leaving the root zone of land under agricultural use around
year 1900. At that time, agricultural management and land use differed sig-
nificantly from current practices.
Materials and methods:
The chapter relies on literature addressing agricul-
tural management around year 1900, on official agricultural statistics and on
N concentrations in root zone percolates measured in well-defined field ex-
periments with year 1900-relevant management. Estimates of concentrations
of nitrate-N in root zone percolates were established for eight categories of
land uses derived from detailed parish-level statistics recorded for land areas
under Danish and German administration in 1896/1900.
Results and discussion:
Parish level statistics for the area under Danish ad-
ministration included 34 land use categories (termed DA), while the area un-
der German administration included 51 categories (termed TY). These were
unified into 26 land uses (termed S) and finally into eight DK categories to
comply with model requirements. For each DK category, an N concentration
was ascribed to the root zone percolate: winter crops (DK-1, 18 mg N/l),
spring crops (DK-2, 13 mg N/l), grass (DK-3, 9 mg N/l), root crops (DK-4, 12
mg N/l), fallow (DK-5, 20 mg N/l), nature (DK-6, 1 mg N/l), forest (DK-7, 2
mg N/l) and other land use (DK-8, 0 mg N/l). These categories account for 8,
23, 36, 4, 6, 14, 7 and 3% of the area, respectively.
Conclusions:
Using the derived estimates, the area-weighted average N con-
centration for land in agricultural use (app. 78% of the land area) is 12 mg N/l,
while the value for the entire land area is about 9.6 mg N/l. This is in accord-
ance with previous preliminary estimates reported in Jensen (ed.) (2017).
The work in this chapter was published in Christensen et al. (2021).
5.1
Introduction
For the time around year 1900, measurements of N concentrations in streams
are extremely few (Westermann, 1898), and their representativeness is uncer-
tain. Consequently, the N load of coastal waters around year 1900 remains
currently unknown.
Bøgestrand et al. (2014a) deduced the N concentration in water reaching the
coastal area around year 1900 from current N concentrations in streams from
minimally disturbed catchments (Kronvang et al., 2015; Bøgestrand et al.,
79
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2014b) and national N balances for agriculture established for year 1900
Kyllingsbæk, 2008). However, Christensen et al. (2017) subsequently demon-
strated that this approach was not valid. One reason is the lack of correlation
between field N surplus and N leaching from the root zone (Blicher-
Mathiesen et al., 2014; Eriksen et al., 2015; Hansen et al., 2015). In an analysis
of 39 streams from Danish catchments with between 0 and 90% of the area in
agricultural use, Kronvang et al. (2015) found a good correlation between the
percentage of land in agriculture and the flow-weighted N concentration in
streams without incorporating the field N surplus. Further, current and year
1900 land use and agricultural management differ substantially. This affects
the interpretation of farm and field N balances, including the proportion of
the N surplus contributing to leaching.
In this chapter, we establish -for the time around year 1900- estimates of con-
centrations of nitrate-N in root zone percolates for eight categories of land
uses. The land use is derived from parish-level statistics recorded in
1896/1900 and N concentrations in root zone percolates measured in well-
defined field experiments with time around year 1900-relevant management.
5.2
Agriculture around year 1900
In the period 1861-1896, the area in agricultural use increased dramatically
and accounted in 1896 for ´nearly 3/4 of the area under Danish administra-
tion. For instance, heathlands and sand dunes declined by 656,000 ha during
1850-1907 (Mortensen, 1969). Large areas of land had been included in rota-
tional cropping in the decades preceding year 1900, and there was widespread
use of bare fallow where the vegetation-free land subject to frequent tillage
throughout a year. Tile drainage, first introduced in Denmark in the 1850s,
covered 26% of the agricultural area by year 1907. The activity peaked during
1860-1880 where 15,000 to 30,000 ha were drained annually (300,000 ha were
drained during the decade 1871-1881). For the agricultural area on the islands
Zealand and Funen, 45% was tile drained by 1907, mainly the more clayey
soils (Jensen, 1988; Olesen, 2009).
The agricultural and land use statistics reported in this section relates to the
area under Danish administration. From 1864 to 1920, the southern part of
Jutland (northern part of Schleswig) was under German administration. Data
based on parish statistics include areas under Danish and German administra-
tion around year 1900 (see section 5.2) and thus covers the current Danish ter-
ritory.
5.2.1 Plant production and nitrogen use
Crop production around year 1900 differed significantly from the current ag-
riculture for virtually all growth factors: inferior crop varieties, higher weed
pressure, lack of chemical crop protection and inferior plant nutrient supply,
including the absence of mineral N fertiliser. The main sources of N were solid
farmyard manure, liquid manure and N
2
fixation by legume crops.
The annual average (1900-1904) input of N with animal manure was estimated
to 21 kg N/ha when corrected for 15% loss of N during storage of feedstuffs
and stable feeding and 25% loss of N during storage (Danmarks Statistik,
1968). Subsequent losses of N in the field were not included. In 1896, the num-
ber of storage tanks for liquid manure was 28,000 (Iversen, 1944); manure
heaps with roof covers accounted for 16,500 in 1907. Statistics for 1895 show a
80
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
total number of 237,000 farms and smallholdings and 35,000 holdings without
land (Christensen, 1985). Thus, only a small proportion of the agricultural
holdings had proper storage facilities, leaving room for substantial losses of
manure-N from the site. Around year 1900, most of the manure was applied
during the period late summer, autumn and early winter, partly due to soil
tillage requirements and establishment of autumn-sown crops, the availabil-
ity of farm labour and the lack of manure storage capacity. Application of an-
imal manure during this period leads to poor N use efficiency with a substan-
tial leaching potential of mineral N present in the manure or derived from
mineralisation of organically bound N outside the growing season.
The grain yields for oats, barley, rye and wheat around year 1900 averaged
14, 18, 20 and 28 hkg/ha, respectively (Iversen, 1942; Danmarks Statistik,
1968). The grain yields are comparable to those achieved in the period 1894-
1904 in the Askov long-term experiments and currently obtained in plots kept
unmanured for more than 120 years (Christensen et al., 2019) and only slightly
lower than yields of cereal crops grown under unmanured conditions in on-
going organic farming experiments (Olesen et al., 2002). For hay produced on
rotational and permanent grassland (incl. meadows), the around-year 1900
yield was 24 and 27 hkg/ha, respectively (Danmarks Statistik, 1968). Even
though contemporary textbooks prescribed generous use of liquid manure to
meadows, permanent grasslands and grass-clover crops in rotation in late au-
tumn and again in the spring (Christensen, 1898), the yield level of grasslands
around year 1900 was below that obtained currently for rotational grass-clo-
ver grown under unmanured conditions (Christensen et al., 2019).
According to Christensen (1898), the typical crop rotation around year 1900
was spring cereals (mainly oats) undersown with grass-clover, three to five
years in grass-clover followed by one year in bare fallow and, finally, autumn
sown cereals (mainly cereal rye) and/or a root crop. On the more fertile soils,
one or more crops of spring-sown oats could follow until the soil became nu-
trient exhausted and a new grass-clover crop was established.
5.2.2 Animal production
Animal husbandry around year 1900 also differed significantly from current
Danish agriculture for most production factors: livestock composition, feed
quality and rate of feeding, grazing intensity and periods, and productivity
per animal unit. Converted into livestock units, the agricultural sector in-
cluded 2.6 million units in 1898 (Danmarks Statistik, 1969) with 54% cattle,
16% pigs, 15% horses, 8% poultry and 7% sheep and goats. Although the num-
ber of cattle around year 1900 and today is just slightly different, the produc-
tivity per livestock unit has increased 3- to 4-fold since 1900 (Kristensen et al.,
2015). Around year 1900, grass ingested in fresh condition was the dominant
source of ruminant forage (49%), while root crops, hay and cereal straw ac-
counted for 19%, 18% and 13%, respectively (Danmarks Statistik, 1968). Kris-
tensen et al. (2015) estimated that for cows, grazing accounted for more than
70% of the feed intake. In terms of digestible protein in homegrown forage
(the source of N ending up in animal manure), grass accounted for 67%, hay
for 24% and root crops for just 4.5%.
5.2.3 Farm structure
Compared with today, the distribution among different animal categories was
very different around year 1900. This is also true for the farm structure. Most
81
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
farms were small in terms of acreage and production volume. Thus, cattle
herds encompassing 1 to 14 cows accounted for 70% of all cows. When calcu-
lated in livestock units (1 unit = cow), the animal density in 1898 was 0.89
unit/ha on land under agricultural use; this ranged from 0.69 in West Jutland
to 1.14 unit/ha on Zealand (Danmarks Statistik, 1969). In 1900, farm sizes
were measured in hartkorn (Hkt.), a unit that combined land area, land use
and soil quality in providing an estimate of the production of individual
farms. Based on this unit, smallholdings (< 1 Hkt.) and smaller farms (1 to 8
Hkt.) accounted for 74% of the total agricultural crop production in 1895
(Christensen, 1985).
5.2.4 Land use and agriculture- based on parish level statistics
To establish an area-distributed account of land use and agriculture around
year 1900 covering the current Danish area, the parish unit appeared most
relevant. The boundaries of the church parishes have remained almost un-
changed since year 1900, and the parish thus probably represents the most
conservative land area unit. The church parish was also an important admin-
istrative unit with detailed information on land use and agriculture being rec-
orded for each parish every five to ten years. The present study relies on sta-
tistical information collected in 1896 for the area under Danish administration
and around year 1900 for the area under German administration, the latter
accounting for 1/11 of the current Danish territory. Matching parish bounda-
ries and parish land use statistics called for some minor adjustments (see Ap-
pendix 5.1). Thus, the present study relies on 1,766 individual parish units of
which 1,702 and 64 are from parishes under Danish and German administra-
tion, respectively.
For each parish, in 1896 the Danish administration allocated the area to 34
land use categories, here coded DA (Table 5.1). The land use categories en-
compassed not only the area in agricultural use but also accounted for the
total area within the parish. In 1900, the German administration allocated each
parish area to 51 land use categories designated by numbers (45 categories) or
by letters (six categories). These are coded TY (Table 5.2). The Danish and Ger-
man land use categories were somewhat different. Using the Danish catego-
ries as template, the two sets of categories were therefore merged into one
new set of categories, coded S and encompassing 26 categories (Table 5.3).
For each land use category, Table 5.3 also shows the area adapted to the water
catchment map anno 2015 (see Appendix 5.1 for details). Next, the 26 S-coded
land use categories were condensed into eight DK categories (Table 5.4). The
reduction of the number of categories was made to facilitate the establishment
of N concentrations in root zone percolates from major land uses based on
current year 1900-relevant data and facilitate their subsequent use in the DK-
model (National Nitrogen Model). Table 5.4 shows that the grass-covered area
and the area of spring-sown crops and nature accounted for 36, 23 and 14% of
the current Danish territory, respectively.
82
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0085.png
Table 5.1.
Land-use data collected in 1896 in parishes under Danish administration
(3,880,000 ha; Statens Statistiske Bureau & Danmarks Statistik, 1898).
Code
DA1
DA2
DA3
DA4
DA5
DA6
DA7
DA8
DA9
DA10
DA11
DA12
DA13
DA14
DA15
DA16
DA17
DA18
DA19
DA20
DA21
DA22
DA23
DA24
DA25
DA26
DA27
DA28
DA29
DA30
DA31
DA32
DA33
DA34
Total
Land use category
Wheat
Cereal rye
Barley
Oats
Mixed cereals (mature)
Buckwheat
Pulses
Spurrey (mature)
Caraway and oil-seed rape
Seed production (clover, grass, beets, lupines)
Potatoes
Sugarbeets and chicory
Carrots
Fodder beets
Green forage (mixed cereals, spurrey, lucerne)
Flax, hemp and tobacco
Garden crops
Black fallow (vegetation-free)
Black fallow ( green manure before ploughing)
Semi-black fallow (early summer-crop)
Cultivated grass for hay
Cultivated grass for grazing
Meadows
Fens and commons
Moors and peatland
Hedgerows and shelters
Gardens and plant nurseries
Forest area (planted)
Forest area (unplanted)
Heathland
Sand dunes and shifting sands
Swamp, foreshores, stone fields etc.
Roads, building sites and storage areas
Lakes, ponds, streams (outside sea territory)
% of total area
0.9
7.6
7.4
11.6
3.1
0.3
0.2
0.2
0.0
0.1
1.4
0.3
0.2
1.8
1.3
0.0
0.0
5.1
0.1
1.4
6.9
17.9
6.0
2.5
2.0
0.2
0.9
6.3
0.7
9.2
1.1
0.4
2.3
0.3
100.0
83
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0086.png
Table 5.2.
Land use data collected in 1900 in parishes under German administration.
Code
Land use category
% of total area
TY1
Winter wheat
1.6
TY2
Spring wheat
0.0
TY3
Winter rye
5.6
TY4
Spring rye
0.0
TY5
Winter barley
0.0
TY6
Spring barley
4.4
TY7
Oats
8.9
TY8
Mixed cereals (winter)
0.0
TY9
Mixed cereals (summer)
2.1
TY10
Buckwheat
0.7
TY11
Peas
0.1
TY12
Fava bean
0.0
TY13
Vetch
0.0
TY14
Mixed cereals
0.4
TY15
Mixed pulses
0.0
TY16
Other types
0.0
TY17
Potatoes
1.0
TY18
Sugar beets
0.0
TY19
Fodder beets
0.6
TY20
Carrots
0.1
TY21
Fodder radish
0.1
TY22
Swedes
1.3
TY23
Field herbs and caddish
0.0
TY24
Other types
0.0
TY25
Winter rape and radish
0.0
TY26
Leindotter (Camelina sativa)
0.0
TY27
Flax
0.0
TY28
Other types
0.0
TY29
Clover (for forage)
1.0
TY30
Lucerne
0.0
TY31
Seradel
0.0
TY32
Spurrey
0.0
TY33
Seed production (clover, grass-clover)
4.2
TY34
Maize
0.0
TY35
Vetch
0.0
TY36
Lupines (for forage)
0.0
TY37
Mixed legumes
0.2
TY38
Mixed vegetables (for forage)
0.0
TY39
Mustard
0.0
TY40
Lupines
0.0
TY41
Mixed vegetables
0.0
TY42
Mustard
0.0
TY43
Fallow
3.3
TY44
Cultivated grass
25.0
TY45
Gardens and fruit plantations
0.7
TYG
Meadows
10.7
TYH
Pastures
11.1
TYI
Forests
3.6
TYJ
Buildings and yards
0.7
TYK
Uncultivated land
5.8
TYL
Roads and lakes, ponds, streams
6.7
Total
100.0
84
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0087.png
Table 5.3.
DA and TY land use categories and distribution merged into 26 S-categories.
Code
S01
S02
S03
S04
S05
S06
S07
S08
S09
S10
S11
S12
S13
S14
S15
S16
S17
S18
S19
S20
S21
S22
S23
S24
S25
S26
Total
Land use category
Wheat
Cereal rye
Barley
Oats
Mixed cereals
Buckwheat
Pulses
Spurrey
Caraway and rape
Seed production
Potatoes
Sugar beets
Carrots
Fodder beets
Green forage
Flax, hemp and tobacco
Garden crops
Fallow
Cultivated grass
Meadows, fens and commons
Moors, peats and heathland
Forest
Roads and building sites
Lakes, ponds and streams
Buildings and yards
Roads and water areas
Area (ha)
41,428
315,998
301,057
479,084
129,831
14,530
11,600
7.641
525
25,028
56,743
13,583
6,535
76,910
51,277
293
37,549
267,843
1,053,714
407,621
587,555
295,580
88,336
13,086
2,967
17,441
4,303,762
% of total
area
1.0
7.3
7.0
11.1
3.0
0.3
0.3
0.2
0.0
0.6
1.3
0.3
0.2
1.8
1.2
0.0
0.9
6.2
24.5
9.5
13.7
6.9
2.1
0.3
0.1
0.4
100.0
DA code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17, 27
18, 19, 20
21, 22
23, 24
25, 30, 31, 32
26, 28, 29
33
34
No code
No code
TY code
1, 5
3
6
7
2, 4, 8, 9
10
11, 12, 13, 14, 15, 16
32
25, 39, 42
29, 30, 31, 33
17
18
20
19, 21,22
23,24, 35, 36,37, 38, 40
27, 28
45
43
44
G, H
K
I
No code
No code
J
L
5.3
Estimating the N concentration in root zone percolates
The N concentrations in water leaving the root zone remain unknown for the
specific land uses adopted around year 1900 due to lack of measurements.
Christensen et al. (2017) discussed factors affecting the N leaching around
year 1900 and provided a preliminary estimate of N concentrations in root
zone percolates in year 1900. Based on year 1900-relevant management, the N
concentrations in leachate from the root zone ranged between 5 and 15 mg
N/l for most of the land, with concentrations exceeding 20 mg N/l for land
under bare fallow, land subject to prolonged grazing periods and soil left bare
after ploughing of grasslands. The average concentration of N in water leav-
ing the root zone was set to 12 mg N/l from land subject to agricultural use.
To estimate N lost from different crop types (DK-1, DK-2, DK-4) and grass-
land (DK-3) around year 1900, results of recent studies with conditions similar
to those around year 1900 (in terms of farming practice, nutrient supply etc.)
were compared.
85
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
For the eight land use categories shown in Table 5.4, current measurements
from well-controlled agricultural experiments with year 1900-relevant man-
agement were employed. This allows us to link specific and well-known man-
agement with N concentrations measured in root zone percolates. For forest-
covered areas, we adopt measurements of N concentrations in soil solutions
from samples retrieved from the bottom of the root zone. The fate of N sup-
plied with residues of N
2
-fixing crop remains uncertain because organic N in
the residue must first be mineralised to become available to subsequent crops.
Mineralisation outside the growing season provides a substantial potential for
N leaching. For unfertilised grassland, terminated in the spring and then
seeded to spring barley, Eriksen et al. (2008) found an annual flow-weighted
nitrate concentration of 36 mg N/l in drainage collected during the leaching
period following the barley crop. Leaching of N from grazed grasslands can
be substantial, in particular grass-clover swards subject to long grazing peri-
ods. Around year 1900, grasslands were amended with liquid manure in the
early spring and after having delivered a first cut of hay, they were used for
grazing until late autumn (Christensen, 1898). Urination by grazing animals
creates locally high inputs of mobile N. For a four-year old grass-clover field,
subject to grazing from late April to late October, Hansen et al. (2012) esti-
mated that one-third of the area became affected by urination and that the N
concentration in percolate from this area was 23 mg N/l. When first exposed
to a grass cut in the spring and then subjected to grazing, the N concentration
was 19 mg N/l.
A study of different organic farming practices was initiated in 1997 at three
sites varying in climate and soil type (Jyndevad, sand; Foulum, loamy sand;
Flakkebjerg, sandy loam; see Olesen et al., 2000). One four-year crop rotation
included grass-clover ley, winter cereals, spring cereals and either a potato or
a grain legume crop. Animal manure was applied at an average annual rate
of 70 kg total-N/ha (Olesen et al., 2000). Nitrate leaching was measured using
porous ceramic suction cups situated at the bottom of the root zone
(Jyndevad, 60 cm; Foulum and Flakkebjerg, 100 cm; Askegaard et al., 2011).
The average flow-weighted nitrate-N concentration of the leachates was re-
markably constant across the different sites and the three rotation cycles in
the period 1997 to 2008. The nitrate-N concentration averaged 11.7 mg N/l,
with a variation from 7.4 to 14.9 mg N/l. Percolation was 637, 362 and 238 mm
at Jyndevad, Foulum and Flakkebjerg, respectively, and considerable differ-
ences thus occurred between sites as to the amount of N lost by leaching.
Eriksen et al. (1999) provide results from an organically managed rotation ad-
dressing cattle production. The total nutrient addition in liquid and bedding-
rich farmyard manure and from grazing cattle corresponded to 0.9 animal
units/ha. The six-course rotation was spring barley undersown with grass-
clover, two years with grass-clover, barley/pea, winter wheat and beetroots.
The grass-clover was subject to one cut early in the growth period and then
exposed to grazing. Except for the lack of bare fallow, the rotation and its
management, including nutrient load, corresponded well with a year 1900
scenario. The average flow-weighed nitrate-N concentrations measured by ce-
ramic cups extracting water leaving the root zone were 13 mg N/l for spring
sown crops (barley undersown with grass-clover and barley/pea mixture), 18
mg N/l for winter wheat, 9 mg N/l for grass-clover (first- and second-year
grass-clover) and 12 mg N/l for the root crop. These values are associated
with DK-codes 1 to 4 and inserted in Table 5.4.
86
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0089.png
Table 5.4.
DK-model codes merged from S codes: Land use, areas and nitrate-N concentrations in root zone percolates for
each DK-model category. Area adjusted to the total area of parish maps around year 1900.
DK-Model
Code
DK-1
DK-2
DK-3
DK-4
DK-5
DK-6
DK-7
DK-8
total
DK-Model: Land use and area
Land use category
Winter crops
Spring crops
Grass
Roots
Fallow
Nature
Forest
Other land use
1,000 ha
357
982
1,538
154
268
588
296
122
4,305
%
8.3
22.8
35.7
3.6
6.2
13.7
6.9
2.8
100.0
S land use codes
1,2
3,4,5,6,7,8,9,16,17
10,15,19,20
11,12,13,14
18
21
22
23,24,25,26
Nitrate-N concentration
mg N/l
18
13
9
12
20
1
2
0
Several studies show significant leaching losses of N from soil under bare fal-
low (DK-5). In a lysimeter experiment at Askov Experimental Station, the an-
nual average N leaching over a four-year period corresponded to 104 kg N/ha
for unmanured bare fallow (Thomsen et al., 1993). The average percolation
during the experimental period was 504 mm/yr, and the concentration of ni-
trate in the leachate was 21 mg N/l. This is in accordance with measurements
of N leaching in lysimeter experiments with undisturbed soil columns con-
ducted in 1870 at Rothamsted Experimental Station. This experiment showed
an average nitrate-N concentration of 19 mg N/l during the first seven years
under permanent fallow (Addiscott, 1988). Therefore, the nitrate-N concen-
tration was set to 20 mg/l for fallow (Table 5.4).
For non-forest areas with natural vegetation (DK-6), an N concentration of 1
mg N/l is applied as previously reported (see Jensen (ed.), 2017). This reflects
the assumption that in areas without net gain in standing vegetation biomass
and soil N storage, the leaching of N reflects deposition of N. The first system-
atic measurements of N in rainwater were made in the period 1921-1926 at
Askov, Blangstedgaard (near Odense), Spangsbjerg (near Esbjerg) and Hor-
num research stations and showed an annual deposition of N in ammonium
and nitrate of between 6 and 11 kg N/ha (Hansen, 1931). Measurements made
in the period 1880-1885 at the Royal Veterinary and Agricultural University’s
experimental field near Copenhagen showed an average annual deposition of
just under 14 kg N/ha (Tuxen, 1890).
For N lost from forest (DK-7), we rely on studies of differently sized forest
areas (Callesen et al., 1999) and forest of different age (Hansen et al., 2007).
The average concentration of N in soil solutions extracted from 75-100 cm soil
depth in forests with an area of < 10 ha, 10-50 ha and > 50 ha were 3.0, 2.0 and
1.3 mg N/l, respectively. For coniferous stands younger than 45 years grown
at a nutrient-poor location, the N concentration in 90 cm soil depth was typi-
cally between 0.5 and 1 mg N/l. The concentration in percolate from decidu-
ous forests at a more nutrient-rich locality varied around 4 mg N/l, generally
with the smallest values under younger stands. The concentrations of N in-
creased with increasing stand age for coniferous as well deciduous stands re-
siding on the more nutrient-rich soils. Based on this information, the value for
forest was set to 2 mg/l (Table 5.4).
87
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
For the category other land use (DK-8), we do not estimate a concentration.
Any losses of N related to this land use code categorise as loss from point
sources.
Appendix 5.1 describes how the data on N in root zone percolates are used in
the hydrological models.
5.4
References
Addiscott, T.M., 1988. Long-term leakage of nitrate from bare unmanured soil.
Soil Use and Management 4, 91-95.
Askegaard, M., Olesen, J.E., Rasmussen, I.A. and Kristensen, K., 2011. Nitrate
leaching from organic arable crop rotations is mostly determined by autumn
field management. Agriculture, Ecosystems & Environment 142, 149-160.
Blicher-Mathiesen, G., Andersen, H.E. and Larsen, S.E., 2014. Nitrogen field
balances and suction cup-measured N leaching in Danish catchments. Agri-
culture, Ecosystems & Environment 196, 69-75.
Bøgestrand, J., Windolf, J. and Kronvang, B. (2014a): Næringsstofbelastning
til vandområder omkring år 1900. Notat fra DCE – Nationalt Center for Miljø
og Energi, 15. december 2014.
Bøgestrand, J., Kronvang, B., Windolf, J. and Kjeldgaard A., 2014b. Bag-
grundsbelastning med total N og nitrat-N. Notat fra DCE- Nationalt Center
for Miljø og Energi, 16. december 2014.
Callesen I., Raulund-Rasmussen K., Gundersen P. and Stryhn H,. 1999. Ni-
trate concentrations in soil solutions below Danish forests. Forest Ecology and
Management 114: 71-82.
Christensen, C., 1898. Landbrugets kulturplanter. August Bangs Boghandels
Forlag, København.
Christensen, J., 1985. Landbostatistik. Håndbog i dansk landbohistorisk stati-
stik 1830-1900. Landbohistorisk Selskab, København.
Christensen B.T., Olesen J.E. and Eriksen J., 2017. Year 1900: Agriculture and
leaching of nitrogen from the root zone. In ‘Estimation of nitrogen concentra-
tions from root zone to marine areas around the year 1900’. Ed. P.N. Jensen.
Scientific Report from DCE No. 241, p. 37-46. Aarhus University, Danish Cen-
tre for Environment and Energy.
Christensen, B.T, Thomsen, I.K. and Eriksen, J., 2019. The Askov Long-Term
Experiments: 1894-2019 – A Unique Research Platform Turns 125 Years. DCA
Report No. 151, March 2019, DCA – Danish Centre for Food and Agriculture,
Aarhus University.
Christensen B.T., Petersen B.F., Olesen J.E. and Eriksen, J., 2021. Land-use and
agriculture in Denmark around year 1900 and the quest for EU Water Frame-
work Directive reference conditions in coastal waters. Ambio 50, 1882-1893.
88
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0091.png
Danmarks Statistik, 1968. Agricultural Statistics 1900-1965, Volume I: Agricul-
tural area and harvest and utilization of fertilizers. Danmarks Statistik, Kø-
benhavn.
Danmarks Statistik, 1969. Agricultural Statistics 1900-1965, Volume II: Live-
stock and livestock products and consumption of feeding stuffs. Danmarks
Statistik, København.
Engelbrecht, Th. H., 1907. Bodenanbau und Viehstand in Schleswig-Holstein
nach Ergebnissen der amtlichen Statistik. II, Anhang, Tabelle 1.
Eriksen J., Askegaard M. and Kristensen, K., 1999. Nitrate leaching in an or-
ganic dairy/crop rotation as affected by organic manure type, livestock den-
sity and crop. Soil Use and Management 15, 176-182.
Eriksen, J., Askegaard, M. and Søegaard, K. (2008): Residual effect and nitrate
leaching in grass-arable rotations: effect of grassland proportion, sward type
and fertilizer history. Soil Use and Management 24, 373-382.
Eriksen, J., Askegaard, M., Rasmussen, J. and Søegaard, K., 2015. Nitrate
leaching and residual effect in dairy crop rotations with grass-clover leys as
influenced by sward age, grazing, cutting and fertilizer regimes. Agriculture,
Ecosystems & Environment 212, 75-84.
Hansen, F., 1931. Undersøgelser af regnvand. Tidsskrift for Planteavl 37, 123-150.
Hansen K., Rosenqvist L., Vesterdal L. and Gundersen P., 2007. Nitrate leach-
ing from three afforestation chronosequences on former arable land in Den-
mark. Global Change Biology 13, 1250-1264.
Hansen, E. M., Eriksen, J., Søegaard, K. and Kristensen, K., 2012. Effects of
grazing strategy on limiting nitrate leaching in grazed grass-clover pastures
on coarse sandy soil. Soil Use and Management 28, 478-487.
Hansen, E.M., Munkholm, L.J., Olesen, J.E. and Melander, B., 2015. Nitrate
leaching, yields and carbon sequestration after non-inversion tillage, catch
crops, and straw retention. Journal of Environmental Quality 44, 868-881.
Iversen, K., 1942. Gødskningens betydning for kornavlen. Side 178-202 i red.
A. Ranløv: Korn. Alfred Jørgensens Forlag, København.
Iversen, K., 1944. Staldgødning og ajle. Side 281-320 I red. K.A. Bondorff and
J. Petersen-Dalum: Den Ny Landmandsbog, Bind I, Planteavl. Westermann,
Egmont H. Petersen, Kgl.
Hof-Bogtrykkeri, København.
Jensen, S.P., 1988, Statistisk oversigt. Side 250-265 i red. C. Bjørn: Det danske
landbrugs historie III – 1810-1914. Landbohistorisk Selskab.
Jensen, P.N. (Ed.), 2017. Estimation of Nitrogen Concentrations from root
zone to marine areas around the year 1900. Aarhus University, DCE – Danish
Centre for Environment and Energy, 126 pp. Scientific Report from DCE –
Danish Centre for Environment and Energy No. 241.
http://dce2.au.dk/pub/SR241.pdf
89
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Kristensen, T., Aaes, O. and Weisbjerg, M.R., 2015. Production and environ-
mental impact of dairy cattle production in Denmark 1900-2010. Livestock
Science 178, 306-312.
Kronvang, B., Windolf, J., Larsen, S.E. and Bøgestrand, J., 2015, Background
concentrations and loadings of nitrogen in Danish surface waters. Acta Agri-
culturae Scandinavica B65, 155-163.
Kyllingsbæk, A., 2008, Landbrugets husholdning med næringsstoffer 1900-2005.
Kvælstof, fosfor, kalium. Intern Rapport, DJF Markbrug nr. 18, august 2008.
Mortensen, E., 1969, De viste vejen. Landhusholdningsselskabets Forlag, Kø-
benhavn.
Olesen, J.E., Askegaard, M. and Rasmussen, I.A., 2000, Design of an organic
farming crop rotation experiment. Acta Agriculturae Scandinavica B50, 13-21.
Olesen, J.E., Rasmussen, I.A., Askegaard, M. and Kristensen, K., 2002, Whole-
rotation dry matter and nitrogen grain yields from the first course of an or-
ganic farming crop rotation experiment. Journal of Agricultural Science, Cam-
bridge 139, 361-370.
Olesen, S.E., 2009. Kortlægning af potentielt dræningsbehov på landbrugsare-
aler opdelt efter landskabselement, geologi, jordklasse, geologisk region samt
høj/lavbund. DJF Intern Rapport Markbrug nr. 21, marts 2009, Det Jordbrugs-
videnskabelige Fakultet, Aarhus Universitet, Tjele.
Statens Statistiske Bureau & Danmarks Statistik, 1898. Arealets Benyttelse i
Danmark den 15. juli 1896 (Statistisk Tabelværk Rk. 5 Litra C Nr 1). Køben-
havn, Bianco Lunos Hof-Trykkeri (F. Dreyer).
Thomsen, I.K., Hansen, J.F., Kjellerup, V. and Christensen, B.T., 1993. Effects of
cropping system and rates of nitrogen in animal slurry and mineral fertilizer on
nitrate leaching from a sandy loam. Soil Use and Management 9, 53-58.
Tuxen, C.F.A., 1890. Undersøgelser over regnens betydning her i landet som
kvælstofkilde for kulturplanterne. Tidsskrift for Landøkonomi 9, 325-350.
Westermann, T., 1898, Om indholdet af plantenæring I vandet fra vore vand-
løb. Tidsskrift for Landbrugets Planteavl 4, 157-165.
90
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
6
The National Nitrogen Model and other Ni-
trogen Sources
Hans Thodsen
1
, Henrik Tornbjerg
1
, Anker Lajer Højberg
2
and Lars Troldborg
2
Quality assurance: Dennis Trolle
1
Aarhus University, Department of Ecoscience
2
GEUS, Geological Survey of Denmark and Greenland´
1
DCE,
Abstract
In this chapter, the NNM model structure is described as well as the modifi-
cations of the model that are performed to represent the year 1900 conditions.
Further additional nitrogen sources, besides leaching from soils and point
sources, are evaluated.
6.1
The national nitrogen model
The national nitrogen model (NNM) was developed in a collaboration be-
tween GEUS and the departments of Ecoscience and Agroecology at Aarhus
University. The model development is described in detail in Højberg et al.
(2015), while only a brief description is provided here. The model is con-
structed by coupling three existing modelling systems describing nitrogen
transport and -reduction in the root zone; groundwater; and surface waters
(Figure 6.1):
1.
Root zone. N-leaching from the root zone is estimated by the NLES4
model for agricultural areas (Kristensen et al., 2008). For non-agricul-
tural areas, land use specific values are used (Højberg et al., 2015).
NLES is a statistical model developed on the basis of data from field
trials where nitrate leaching has been measured one metre below the
surface. Root zone leaching is calculated at the same 500 m grid used
in the hydrological model (see chapter 3).
Groundwater. Flow and transport from the root zone to the surface
water system is simulated by the DK-model (see chapter 3). The
transport is simulated using the particle tracking approach. By this
approach, particles are tracked through the groundwater system from
the root zone until it is removed by a sink, such as streams, wells,
drains or the sea.
Surface water. Statistical models are applied to describe retention and
reduction of nitrate in streams, wetlands and lakes.
2.
3.
91
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0094.png
Figure 6.1.
Sketch illustrating the model system combined in the national nitrogen model (NNM).
Figure 6.2.
Schematic illustration of NNM. Orange is nitrogen sources. Blue is nitrogen sinks.
The conceptual structure of the NNM is illustrated in Figure 6.2. Transport
and retention/removal of nitrate is calculated at sub-catchment scale (ID15
catchments) defined by topographic catchments with a mean area of approx.
15 km
2
(for ID15 map see Figure 6.5). Conceptually, each ID15 catchment is
divided into two compartments: an “Internal” part and a “Main” part. The
“Main” part accounts for processes in the main river reach that flows across
the catchment (Figure 6.2, left-hand side), which is included in the hydrolog-
ical model. The Internal part accounts for tributaries to the main river and
associated surface water features. Within each ID15 catchment, a mass balance
is calculated from the different nitrogen sources entering the catchment and
sinks accounting for nitrogen retention. Diffuse ID15 nitrogen load is first
added to the internal compartment, which also receives nitrogen from point
92
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
sources, atmospheric deposition and organic nitrogen. The total nitrogen
amount in the system is then subject to retention in accordance with the re-
ductive capability of the internal surface water system. Excess nitrogen is
routed to the Main part, which also receives nitrogen from upstream ID15
catchments, before reduction in the main river and associated larger lakes is
accounted for (Figure 6.2). The order and content of the different steps in the
model are:
1. Root zone leaching: The diffuse source of nitrogen leaching from the root
zone, primarily in the form of nitrate
2. Reduction in groundwater: During transport in the subsurface, nitrate
may be subject to removal, with reduction of nitrate to free nitrogen gas
by denitrification being the most important process. The redox conditions
of the subsurface in Denmark are generally characterised by an upper
oxic part and a lower reduced part that is separated by a redox interface.
Estimating the nitrate reduction in the subsurface is thus accomplished in
the NNM by establishing whether particles are crossing the redox inter-
face. Particles transported by tile drains will usually not have crossed the
redox interface. Nitrate entering the lower reduced environment in the
subsurface is thus assumed to be removed by denitrification.
3.
Reduction in groundwater-surface water interactions (Nret Near surface
processes in Figure 6.2): The nitrate reduction processes in the groundwa-
ter-surface water interface have not been included directly in the model
but are accounted for by specifying reduction proportionally to the num-
ber of small streams/ditches, with a higher reduction potential in sandy
areas. In the following, this part is called “the conceptual reducer” (e.g.
Table 7.10).
4. Addition of other nitrogen sources: This accounts for point sources that
are discharged directly to streams, atmospheric nitrogen deposition on
lake surfaces and most importantly the organic nitrogen fraction originat-
ing from the landscape and aquatic environment.
5. Internal surface water retention in ID15: The internal retention is calcu-
lated based on the lengths and sizes of the tributaries as well as the num-
ber and area of small lakes and wetlands in the ID15 sub catchment. The
actual retention is calculated in a lumped process, i.e. the sources in pt. 4
is added to the internal diffuse load (pt. 1 minus pt. 2 and pt. 3), after
which the calculated internal retention is subtracted, considering that the
nitrogen mass cannot be negative. The approach to calculate the internal
reduction capacity is detailed in Højberg et al. (2015).
6. Addition of N from upstream: The main rivers receive nitrogen from the
internal part of the ID15 catchment and from associated upstream ID15
catchments.
7. Retention in main rivers and lakes: This includes the retention in the
main river and associated larger lakes.
For NNM results from the year 1900 period and the present period see Table
7.10.
6.1.1 Modifications of the national nitrogen model (NNM) to represent
year 1900 conditions
The NNM was developed and calibrated to the present conditions (1990-2011)
(Højberg et al., 2015). The present project has not included an update or recal-
ibration of the model to new data, and the model structure and parameters
93
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
have thus been reused from Højberg et al. (2015), with only a few modifica-
tions. These modifications are related to the representations of the small
streams and the organic nitrogen model as described in section 6.4 and chap-
ter 7. Furthermore, a new module to describe the effect of natural wetlands
has been developed (chapter 7) and included in the NNM for both the present
and the year 1900 period.
N leaching from the root zone
The nitrate leaching model NLES4 (Kristensen et al., 2008) has been used to
estimate the present period nitrate leaching. The model was set up to reflect
2011 agricultural practices and was run with climate data from the period
1990 to 2011, i.e. using the same agricultural practice each year but with the
actual climate. Afterwards, the model results were averaged across the period
1990-2011 to avoid the impact of year-to-year variations on the climate
(Troldborg et al., 2016). Output from the model is a “climate-normalised” ni-
trate leaching in kg N/ha/yr. The yearly data have been disaggregated to
monthly values, forming the temporal basis of the NNM, by calculating
monthly nitrate leaching fractions based on daily nitrate leaching time series
from 1990-2011 included in the original NNM (Højberg et al., 2015). This is
done at the 500 m grid, geographical level.
NLES4 is a statistical model and can thus not be applied directly to estimate
historical conditions. Instead, the year 1900 soil water concentrations esti-
mated in chapter 5 have been multiplied with the net precipitation for year
1900 calculated by the national hydrological model, chapter 3. To similarly
utilise the year-to-year climatic variation for the historical period, average
monthly net precipitation was calculated during the period 1890-1910, prior
to calculating the nitrate leaching in kg N/ha/month. The 1890-1910 net pre-
cipitation is calculated from the “delta change” 1990-2010 precipitation de-
scribed in chapter 1.
In both periods, a climate-normalised monthly nitrate leaching has thus been
used as input for the NNM, while transport in the subsurface reflects the ac-
tual climatic variability in the two periods.
Groundwater reduction
Reduction of groundwater is calculated by the same approach for the two pe-
riods, where particle tracking is used to describe the nitrate transport from the
root zone to the surface water system, keeping track of particles crossing the
redox interface. Particle tracking is based on groundwater flow calculations
from 2000 to 2017 to represent the current climate, while flow calculations
from 1890 to 1910 have been used for the historical period.
The reducing capacity of sediments is slowly consumed, resulting in a vertical
movement of the redox interface. Observations of this movement have been
reported by Postma et al. (1991), who found a vertical migration rate of 0.34
cm/yr in a sandy aquifer, while an even lower migration of 0.04cm/yr has
been found in till areas (Robertson et al., 1996). In the 120 years since 1900, the
redox interface can thus be expected to have moved vertically between 4 and
34 cm, which is much less than the accuracy by which the location of the redox
interface can be determined for the present conditions. The same national map
of the redox interface developed by Ernstsen and von Platen (2014) has thus
been used in both periods. This map was also applied in the development of
the nitrogen model.
94
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0097.png
6.2
Point sources
The point sources applied for the present period are the inland point sources
presented in Miljøstyrelsen (2019). Point sources applied for the year 1900
period are described in detail in chapter 4.
6.3
Atmospheric nitrogen deposition year 1900
Atmospheric nitrogen deposition is a nitrogen input to lake surfaces in the
NNM (Højberg et al., 2015; Jensen (ed.) 2017). Three different model estimates
of the nitrogen deposition for year 1900 were examined in Jensen (ed.) (2017).
The three modelling approaches gave nearly similar results and estimated
that the nitrogen deposition in 1900 was about 30% of that in 2000. The nitro-
gen deposition applied for year 1900 in the NNM is based on the 28% of the
EMEP depositions calculated for year 2000, which represents the present time.
A deposition in the time around year 1900 of 28% of that in 2000 was estimated
based on the difference found using the Danish Eulerian Hemispheric Model
(DEHM) for freshwater surfaces. EMEP and the MATCH model show indexes
close to 30% for the year 1900 when year 2000 =1 (Jensen (ed.), 2017).
For the present time, gridded deposition time series from EMEP were applied
(Højberg et al., 2015).
Nitrogen deposition on land surfaces is included as part of soil leaching.
6.4
Organic nitrogen concentrations
The organic nitrogen (OrgN) fraction found in rivers consists of both particu-
late and dissolved organic nitrogen. It originates from both landscape sources
and surface water sources, such as rivers and lakes. Landscape sources can be
relatively fresh organic matter (e.g. fallen leaves and grasses) reaching the
freshwater system from the surface. Older organic soil particles might reach
the surface water through bank erosion, surface erosion or dissolved organic
matter through tile drains or leaching. Sources within the freshwater system
originate mainly from the primary production within lakes and rivers but de-
pend to some degree on landscape sources of nitrogen, for example aquatic
vegetation uptake of dissolved inorganic nitrogen originating from the land-
scape.
The organic nitrogen fraction found in rivers does not originate directly from
the nitrogen applied to cropped fields and the leaching of this nitrogen. There-
fore, the NNM needs a module specifying the amount of organic nitrogen
from landscape sources transported in rivers.
The organic nitrogen (OrgN) model included in the original NNM was im-
proved as a part of this study. As OrgN is not measured directly, monthly
flow-weighted OrgN concentrations were calculated from measured values
of total nitrogen (TN) and fractions of DIN as: TN – NH
x
–NO
x
= OrgN (Figure
6.3). Monitoring stations with at least 96 monthly (eight years) loads from the
period 1990-2017 were used. Monitoring stations downstream larger lakes in-
fluencing the OrgN concentration were excluded. For areas upstream moni-
toring stations (for river catchments with more than one monitoring station,
only the most downstream station was used) the average monthly flow-
weighted OrgN concentration at the monitoring station was used in the
NNM.
95
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0098.png
For ungauged areas (ID15 catchments see section 7.1), monthly flow-weighted
averages of monitoring station concentrations in 12 regions were used (Figure
6.3) for OrgN concentrations, see Figure 6.5.
Figure 6.3.
12 regions (original
region 7 is split into four subre-
gions). Dots are monitoring sta-
tions with mean organic nitrogen
concentrations.
After these concentrations were calculated and applied in the NMN model, it
was established that OrgN concentrations for the period 2011-2015 and for
some stations also for 2009-2010 were underestimated due to the application
of an erroneous laboratory analysis method. Therefore, the presented OrgN
concentrations are somewhat underestimated (Larsen et al., 2021a, 2021b). It
was beyond the scope of this project to rerun the entire NNM with recalcu-
lated OrgN concentrations.
The regional average monthly flow-weighted OrgN concentrations used for the
present period are shown in Figure 6.4. Some geographical differences were
found, for example between the low OrgN concentrations in western Jutland
(region 3) and the high concentrations in central and northern Zealand (region
7.3 and 8).
96
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0099.png
Figure 6.4
Average monthly
OrgN concentrations (1990-2017)
in 12 regions (original region 7 is
split into four subregions, see
Figure 6.3).
ID15-specific mean annual OrgN concentrations used in the NNM for the pre-
sent period are shown in Figure 6.5.
For the period around the year 1900, a few studies estimate the OrgN concen-
trations and compare the results with modern values. Goolsby and Battaglin
(2001) found that total OrgN concentrations in the lower Illinois River were
21% lower during the period 1897-1902 than between 1980-1998. They found
that the difference was 15% for the Upper Mississippi River (US) near Grafton
(below Illinois R. and above Missouri R.) between the periods 1899-1900 and
1980-1998. For the lower Missouri river (US), OrgN concentrations fell from
1.8 mg/l in 1899-1900 to 1.1 mg/l in 1979-1981 and 1995-2007, this reduction
is, though, primarily due to impoundment of the river (Blevins et al., 2014).
Green et al. (2004) estimated that in the temperate climate zone (global scale),
the OrgN concentrations in pre-industrial times were half of those of the
1990s. However, in a Danish context, the period around year 1900 cannot be
perceived as pre-industrial, and the change is, therefore, likely smaller. Based
on these earlier studies and the Danish context, a 20% OrgN concentration
reduction around the year 1900 period compared with the present period is
assumed.
97
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0100.png
Figure 6.5.
Areas (ID15 catchments) with similar mean annual OrgN concentrations (present period).
6.5
References
Blevins, D.W., Wilkison, D.H. and Niesen, S.L., 2014. Pre- and post-impound-
ment nitrogen in the lower Missouri River. Hydrological Processes 28, 2535-
2549. DOI: 10.1002/hyp.9797.
Ernstsen, V. og von Platen, F. 2014. Opdatering af det nationale redoxkort fra
2006. Danmarks og Grønlands Geologiske Undersøgelse, rapport 2014/20.
Goolsby, D.A. and Battaglin, W.A., 2001. Long-term changes in consentrations
and flux fo nitrogen in the Mississippi River Basin, USA. Hydrological Pro-
cesses 15, 1209-1226. doi:10.1002/hyp.210.
Green, P.A., Vörösmarty, C.J., Meybeck, M., Galloway, J.N., Peterson, B.J. and
Boyer, E.W., 2004. Pre-industrial and contemporary fluxes of nitrogen
through rivers: A global assessment based on typology. Biogeochemistry 68,
71-105. doi:10.1023/B:BIOG.0000025742.82155.92.
98
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0101.png
Højberg, A.L., Windolf, J., Børgesen, C.D., Troldborg, L., Tornbjerg, H., Bli-
cher-Mathiesen, G., Kronvang, B., Thodsen, H. and Erntsen, V., 2015. National
kvælstofmodel. Oplandsmodel til belastning og virkemidler. De Nationale
Geologiske Undersøgelser for Danmark og Grønland – GEUS, 111pp.
Jensen, P.N. (Ed.), 2017. Estimation of Nitrogen Concentrations from root
zone to marine areas around the year 1900. Aarhus University, DCE – Danish
Centre for Environment and Energy, 126 pp. Scientific Report from DCE –
Danish Centre for Environment and Energy No. 241.
http://dce2.au.dk/pub/SR241.pdf
Kristensen, K., Waagepetersen, J., Børgesen, C.D., Vinther, F.P., Grant, R., Bli-
cher-Mathiesen, G., 2008. Reestimation and further development in the model
N-LES to N-LES4. Det Jordbrugsvidenskabelige Fakultet, Aarhus Universitet
og Danmarks Miljøundersøgelser, Aarhus Universitet. DJF rapport 139.
Larsen, S.E., Tornbjerg, H., Thodsen, H., Kronvang, B. and Blicher-Mathiesen,
G. 2021a. Analyse af organisk kvælstof koncentrationer i vandløb i to perioder
med henblik på at udvikle en korrektionsformel for perioden 2009-2014. Aar-
hus Universitet, DCE – Nationalt Center for Miljø og Energi, 115 s. – Fagligt
notat nr. 2021|29 https://dce.au.dk/fileadmin/dce.au.dk/Udgivelser/No-
tater_2021/N2021_29.pdf
Larsen, S.E., Tornbjerg, H., Thodsen, H., Kronvang, B. and Blicher-Mathiesen,
G. 2021b. Analyse af organisk kvælstof koncentrationer i vandløb med hen-
blik på at udvikle en korrektionsformel for 2015. Aarhus Universitet, DCE –
Nationalt Center for Miljø og Energi, 19 s. – Fagligt notat nr. 2021|39
https://dce.au.dk/fileadmin/dce.au.dk/Udgivelser/Nota-
ter_2021/N2021_39.pdf
Miljøstyrelsen, 2019. Punktkilder 2017. Miljøstyrelsen. 88pp.
Punktkilder 2017
(mst.dk)
Postma, D., Boesen, C., Kristiansen and H., Larsen, F., 1991. Nitrate reduction
in an unconfined Sandy aquifer – water chemistry, reduction processes, and
geochemical
modeling.
Water
Resour.
Res.
27,
2027–2045.
http://dx.doi.org/10.1029/91WR00989.
Robertson, W.D., Russell, B.M. and Cherry, J.A., 1996. Attenuation of nitrate
in aquitard sediments of southern Ontario. J. Hydrol. 180, 267–281.
http://dx.doi.org/10.1016/0022-1694(95)02885-4
Troldborg, L., Børgesen, C.D., Thodsen, H. and v.d. Keur, P., 2016. National
Kvælstofmodel - Kvælstofpåvirkning af Grundvand. De nationale Geologiske
Undersøgelser for Danmark og Grøndland
- GEUS. 78pp.
http://www.geus.dk/DK/water-soil/water-cycle/Sider/national_kvael-
stofmodel-dk.aspx
99
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
7
Modelling nitrogen retention in surface
waters
Hans Thodsen
1
, Henrik Tornbjerg
1
, Carl Christian Hoffmann
1
, Anker Lajer Højberg
2
Quality assurance: Dennis Trolle
1
Aarhus University, Department of Ecoscience
2
GEUS, Geological Survey of Denmark and Greenland
1
DCE,
Abstract
Purpose:
The purpose of this chapter is to analyse the difference in landscape
nitrogen (N) retention between the current period and the period around the
year 1900. This was done by analysing the major elements of N retention dur-
ing both periods.
Materials and methods:
The National Nitrogen model (NNM) is used for cal-
culating N retention (Nret) in surface freshwater (Højberg et al., 2015a;
Højberg et al., 2020). The NNM has routines calculating Nret in wetlands, con-
structed wetlands, small watercourses (streams and ditches), small lakes,
larger watercourses and larger lakes. For a more detailed description of how
historic climate/weather data is used see chapter 1, 2 and 3. As for climate,
the present period is represented by the period 1990-2010.
For all surface water environments, except wetlands, a present period map is
used as the basis to derive the year 1900 period conditions. For example, the
20% increased length of year 1900 larger rivers is distributed through manip-
ulating the length of each present period river stretch. For wetlands, the par-
ish land use survey from 1896/1900 (see section 5.1) was used for the time
around year 1900, and a land use map from 2016 was used for the present
period (Levin et al., 2017).
As there has been some development in the numbers, functions and areas of
aquatic environments during the two periods, for example due to restoration
or drainage of larger lakes during the period around year 1900 to 2017, a
standard year was chosen for both periods. The year 1900 was chosen for the
year 1900 period and 2017 for the present period. Thus, a lake restored in, for
example, 2008 and present in 2017 is included in the model for the entire pre-
sent period. This approach was chosen to make the year 1900 period estimates
as comparable to the present as possible.
Results and discussion:
Mean monthly nitrogen retentions were calculated
for each aquatic environment at the ID15 geographical level for each of the
two periods. On this basis, mean annual absolute and percentage changes be-
tween the present period and the period around year 1900 were calculated.
Conclusions:
The absolute Nret amount was shown to be higher in the pre-
sent period (28,000 ton N) than around the year 1900 period (26,000 ton N)
due to a larger load. However, the relative Nret was larger during the year
1900 period, where 43% of the load to freshwater was estimated to be removed
compared with 33% for the present period.
100
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0103.png
7.1
Introduction
The aim is to calculate mean monthly Nret for all environments represented
in the NNM for the period around year 1900 and the present period and to
present the differences between the two periods.
7.2
Materials and methods
NNM includes a surface water nitrogen retention (Nret) part consisting of six
separate modules describing Nret in six different aquatic environments. The
modules are divided into two groups, one (Internal/Tributary) handling
small-scale retention at tributary level and one (Main Watercourses) handling
Nret at the larger main river level (Figure 7.1). The Internal Nret group han-
dles Nret in streams, small lakes, constructed wetlands and natural wetlands.
The larger-scale group handles Nret in rivers and larger lakes and receives
water and nitrogen from upstream sub-catchments. Compared with other
North European countries, there is little modern hydropower in Denmark and
rivers are almost not used for navigation, and hence there are no inland canals
for navigational use. Therefore, almost no flow-regulating structures are
needed for these activities. Most weirs are historical relics from closed old wa-
ter mills, and mill ponds were generally small with a short water residence
time. The Danish database for river restoration activities lists “replacing old
weirs with riffles” as a clear no. 1 activity during the 1980s and 1990s, and as
a result the number of mill ponds has decreased since 1900 (Iversen and An-
dersen, 1997). The effect of this process on nitrogen retention is thought to be
small but almost certainly would have meant a slightly higher retention rate
in year 1900 (see Jensen (ed.) (2017) for more information).
All modules can be altered to represent year 1900 conditions.
Figure 7.1.
Schematic flow diagram of the National Nitrogen Model, N source (orange) and N retention (blue).
Because of the methodology applied in assessing the load to the sea in year
1900 (see chapter 8), neither the total mass balance nor the change in nitrogen
101
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
retention in each environment presented in this chapter is completely compa-
rable to the calculated load for year 1900 (chapter 8). However, the changes in
nitrogen retention in each environment are described below to provide some
indication of the changes between year 1900 and the present.
7.3
Nitrogen retention in natural wetlands
Nitrogen retention in natural (non-constructed) wetlands was not included in
the original NNM (Højberg et al., 2015a). A new model was developed and
included in the NNM, calculating Nret for each ID15 with a wetland area.
Wetland Nret is calculated based on the internal load generated in each sepa-
rate ID15, which consequently does not receive any load from upstream ID15
catchments. This will be the case for most wetlands.
The wetland retention model depends on the wetland area of each ID15 catch-
ment and a monthly outflow concentration from the wetland. The load to the
wetland is an area-weighted fraction of the load generated within the ID15.
Through a national scale calibration procedure, the wetland area multiplied
by 3 is estimated as the source area to the wetland. The wetland source area
cannot be larger than 75% of the ID15 area.
Wetland source area load = ID15 load × (wetland area x 3) / ID15 area)
Wetland outflow load = wetland outflow N concentration x wetland source
area (+wetland area) runoff.
Wetland retention = Wetland source area load - Wetland outflow load (if Wet-
land outflow load > Wetland source area load then is set to Wetland outflow
load = Wetland source area load).
The monthly wetland outflow N concentrations were calculated based on
measurements from three natural wetlands in Denmark – one in East and two
in West Denmark. From these measurements, two sets of wetland outflow
concentrations were generated (Table 7.1), where one set was applied to wet-
lands in western Denmark and the other to wetlands in eastern Denmark. It
is assumed that the two sets of outflow concentrations are more suitable for
East and West Denmark than an average would be for the entire country. The
same monthly outflow concentrations are used for year 1900 and the present
period, respectively.
The wetland data from eastern Denmark originate from a riparian wetland
along River Stevns. The land use in the upland catchment to this wetland is
dominated by agriculture. Nitrogen discharge data were generated from three
outflow stations situated approximately 2 m from the river. The data include
concentrations measured at several depths below terrain (15-600 cm).
The wetland data from western Denmark originate from two riparian wet-
lands that are both situated along River Gjern. The land use in the upland
catchments to the riparian wetlands was dominated by agriculture. Data were
generated based on several outflow stations and included nitrogen concentra-
tions from several depths (0-500 cm) (Hoffmann et al., 1993; Hoffmann, 1998;
Hoffmann et al., 2000).
Wetland areas for each ID15 catchment from both the period around year 1900
and the present period were applied in the NNM.
102
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0105.png
Table 7.1.
Wetland outflow total nitrogen concentrations. Based on (Hoffmann et at. 1993,
Hoffmann 1998, Hoffmannet al. 2000).
Month
January
February
March
April
May
June
July
August
September
October
November
December
Eastern Denmark (mg/l)
0.75
0.60
0.48
0.99
1.55
1.46
1.92
2.47
1.72
0.99
0.73
0.76
Western Denmark (mg/l)
1.64
1.59
1.17
1.29
1.48
0.97
0.61
0.58
0.72
0.99
0.81
0.93
Data on wetland areas for 1896/1900 were collected from the land use assess-
ment made for each parish (n = 1766) and are the same data used in the assess-
ment of crop distributions (section 5.2.4). The land use categories used in esti-
mating the wetland area were meadow and peat bog. Comparable statistical
land use assessments were found for the southern part of Jutland that in 1900
was part of Germany where the category meadow was used (section 5.2.4). The
national wetland area for 1896/1900 (used unaltered as wetland area in 1900) is
reported to be 3,483 km
2
. The parish land use survey is a statistical survey and
not a map survey; hence, the location of each wetland within a parish is not
known. Therefore, the wetlands were treated as being equally distributed
within each parish and transferred to ID15 catchments with a parish-specific
area weight. The spatial distribution is seen in Figure 7.2.
Figure 7.2.
Fraction of wetland
area in each ID15 catchment
year 1900.
103
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0106.png
For the present period, the 2016 Basemap (Levin et al., 2017) was applied using
the category “Nature open wet” (grid code 802). The total wetland area is 1,132
km
2
. Thus, the wetland area today is estimated to be 67% smaller than that
around year 1900. ID15 catchment wetland areas are calculated directly by over-
laying with the Basemap. The spatial distribution is shown in Figure 7.3.
Figure 7.3.
Fraction of wetland
area in each ID15 catchment
year 2016.
The methods for estimating ID15 wetland areas are not the same as there is
no digitised map from year 1900. This introduces small differences in the Nret
estimations between the two periods.
Table 7.2
Nitrogen retention and change in national wetlands between the present and
year 1900.
Wetlands nitrogen retention
(ton/yr)
8,100
3,300
-60 (147)
Change in wetlands nitrogen retention
– from 1900 to present %
– (from present to 1900 %)
Period
Year 1900
Present
As shown in Table 7.2, the area-specific retention is about 29 kg N/ha in the
present period and 23 kg N/ha around year 1900.
104
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0107.png
Nitrogen retention is modelled to have been 147% larger around the year 1900
than at present. This is due to the substantially larger area of wetlands around
at that time, which was approximately three times larger than at present. The
fact that the present increased nitrogen load to the remaining wetlands also
increases the retention in these cannot counterbalance the loss in wetland area
from year 1900 to the present.
7.4
Nitrogen retention in constructed wetlands
Constructed wetlands, aiming at maximising nitrogen retention, is a recent in-
vention and was not in use around year 1900. Therefore, this module is turned
off for this period and only active for the present period. For the present period,
the NNM was updated with new wetlands constructed no later than 2017.
The Nret model for constructed wetlands is based on measurements from a
0.8 ha constructed wetland monitored between 1990 and 2005 (Fuglsang, 2006;
Hoffmann et al., 2006). The Nret model has a summer (May - October) and a
winter (November – April) stage as monitoring data show higher area-specific
Nret rates during summer than during winter. Hoffmann et al. (2006) found
lower area Nret rates for constructed wetlands in sandy areas (120 kg
N/ha/yr. than in loamy/clayey areas (190 kg N/ha/yr). Therefore, there are
separate versions for both soil types. However, the area-specific retention is
markedly higher for constructed than for natural wetlands, which was calcu-
lated to 29 kg N/ha in the present (section 7.3). The constructed wetland Nret
models are simple power functions yielding monthly area-specific Nret rates
(kg N/ha/month) as a function of wetland runoff (mm/month). As con-
structed wetlands need to meet minimum estimated Nret rates (kg N/ha) to
receive subsidies for construction, the load to the constructed wetland must
be large enough to ensure such Nret rates. Therefore, there is a fairly constant
ratio between the area of the constructed wetland and the upstream source
area. In this way the runoff entering the constructed wetland can be estimated
knowing the size of the constructed wetland (Højberg et al., 2015b). Modelled
nitrogen retention in constructed wetlands is given in Table 7.3.
Table 7.3.
Nitrogen retention in constructed wetlands.
-
Year 1900
Present
Nitrogen retention in constructed wetlands (ton/yr)
0
265
The location of constructed wetlands is depicted in Figure 7.4.
After the modelling process was completed, a modelling mistake was discov-
ered. The constructed wetland area used for the present period should have
been that of 2017. Instead, the actual area of constructed wetlands of each year
between 1990 and 2010 was used, thereby including too few constructed wet-
lands for too short a period, leading to underestimated retention. On the other
hand, during the latest work with the NNM (Højberg et al., 2021), it was dis-
covered that the maps representing the constructed wetlands included the
whole project area and not only the constructed wetlands area. Besides this,
the project area often included areas with natural wetlands and lakes that are
also included in other modules of the NNM. This led to an overestimation of
the retention. In Højberg et al. (2021), the average N retention between 1990
and 2010 is reported to be about 100 ton N/yr (0.1x10
3
ton N/yr), which is
105
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0108.png
lower than it would be for 2017. Therefore, and because the retention in con-
structed wetlands is of minor importance in all cases, it was decided to keep
265 ton N/yr as the retention in constructed wetlands.
Figure 7.4.
Location of con-
structed wetlands (sizes are ex-
aggerated for the sake of visibil-
ity).
7.5
Nitrogen retention in small watercourses
For calculating Nret in both small and larger watercourses, a residence time
approach suggested by Seitzinger et al. (2002) is applied in the NNM (see sec-
tion 7.2). To represent both summer and winter conditions (in-stream vegeta-
tion growing and non-vegetation seasons) and different watercourse size cat-
egories, seasonally different depths and flow velocities are applied. Besides
this, a scaling factor considering that a drop of water will not run through the
entire lengths of all watercourses (all tributaries) in a subbasin is introduced.
The scaling factor is calculated based on ID15 catchment watercourse density
(km/km
2
) (Højberg et al., 2015b).
In Jensen (ed.) (2017), the total length of small watercourses (<2.5 m width),
(streams/ditches) was estimated to have been 20% longer for the year 1900
period. This was mainly due to the existence of more open ditches that were
later converted into tile drains or irrigation canals, which are not in use in the
present period, and because many small watercourses have been straight-
ened, from a natural meandering to a straight planform, since 1900. However,
the process of straightening small watercourses is believed to have started be-
fore 1900 but continuing well into the 20
th
century. This increases the total
length from 46,610 km at present to 55,930 km in year 1900. The national wa-
tercourse GIS layer used in the original NNM was cleaned for watercourses
located in marine areas and other errors (Højberg et al., 2015a). Besides this,
in earlier work with the NNM some forested areas were observed to have a
106
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0109.png
very high density of streams that in reality were dry most of the year, thus
creating unrealistically high retention values. Therefore, small watercourses
in forest areas were omitted from the analysis.
Table 7.4.
National small watercourse nitrogen retention and change between the present
and year 1900.
Period
Small watercourse nitrogen
retention (ton/yr)
Change in small watercourse
nitrogen retention
– from 1900 to present %
– (from present to 1900 %)
Year 1900
Present
7,500
10,400
39 (-28)
Nitrogen retention in smaller watercourses is shown to have been about 28%
lower in the year 1900 period than at present. This is due to the lower load but
despite the estimated 20% longer small watercourses around year 1900. The
longer watercourses will have longer residence times, yielding higher nitro-
gen retention percentages.
7.6
Nitrogen retention in larger watercourses
For a description of the modelling approach see section 7.2 and Højberg et al.
(2015a,b).
In Jensen (ed.) 2017, the total length of larger watercourses (>2.5 m width) was
estimated to have been 20% longer for the year 1900 period. This was mainly
because many larger watercourses have been straightened from a natural me-
andering to a straight planform since 1900. Because of the straightening, larger
watercourses are more technologically demanding and have benefitted more
from the invention of heavy earth moving machinery than small straightened
streams. The straightening process took place later for large than for small
streams. The process of straightening meandering watercourses accounts for
the entire 20% decrease in the length of large watercourses, increasing total
length from 12,480 km at present to 14,970 km in year 1900.
Table 7.5.
National large watercourse nitrogen retention and change between the present
and year 1900.
Period
Larger watercourse nitrogen Change in larger watercourse nitrogen
retention (ton/yr)
retention
– from 1900 to present %
– (from present to 1900 %)
Year 1900
Present
2,700
4,400
61 (-38)
Nitrogen retention is modelled to have been 38% lower around year 1900 than
at present (Table 7.5). This is due to the higher load in the present period
(which results in higher retention) and despite the shorter length and shorter
residence time of the larger watercourses in the present period (which results
in lower retention).
107
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0110.png
7.7
Nitrogen retention in small lakes
Small lakes are defined as smaller lakes with an outflow. Smaller lakes are
assumed located on tributaries and not receiving water or nitrogen from up-
stream ID15 catchments. As small lakes are located inside the ID15 catchments
and as there often is >1 small lake in an ID15, the specific load of water and
nitrogen to each of these lakes is unknown (un-modelled), as is the catchment
area of each small lake. The mean water depth is also unknown. Therefore,
the retention in the >27,000 small lakes (Højberg et al., 2015b) cannot be esti-
mated using a water residence time approach as for larger lakes (section 7.8),
and which is often recommended as the best approach by, for instance,
Saunders and Kalff (2001).
Nitrogen retention in small lakes is believed to have been different (in
amount) around year 1900 than during the present period due to two factors.
The first factor is that many small lakes and ponds have been filled or drained,
and the total area is thus smaller today. The same estimation as that used by
Jensen (ed.) (2017) is applied, increasing the national area of small lakes by
100% for year 1900 (170 km
2
) compared with the present area (85 km
2
). The
second factor is that the nitrogen load to small lakes would also have been
different. The load estimate (for description see Højberg et al., 2015a, b) is
changed with the same proportion as the total N leaching at the national scale,
which is modelled to have been 13% lower in the year 1900 period than in the
present period (see Table 7.10). Assuming that Nret follows the load to small
lakes, the area retention rates decrease by 13% for the year 1900 period as well
(Table 7.6).
Nitrogen retention values in Table 7.6 are derived from mass balances for 13
Danish lakes, with an average Nret of 320 kg/ha (lake surface area) per year.
Soil type and land use are calculated from a buffer around each lake, with an
area 5-10 times the lake area (Højberg et al. 2015b).
Table 7.6.
Small lake nitrogen retention rates (kg/ha lake surface area) used for the year
1900 and the present period.
Soil type
Land use
(present)
Clay
Clay
Clay
Sand
Sand
Sand
>60% agriculture
30-60% agriculture
<30% agriculture
>60% agriculture
30-60% agriculture
<30% agriculture
1900 nitrogen retention Present nitrogen reten-
rate
kg/ha
348
148
70
261
109
52
400
170
80
300
125
60
tion rate kg/ha
Table 7.7.
National small lake nitrogen retention and change between the present and
year 1900.
Period
Small lakes nitrogen retention Change in small lakes nitrogen retention
(ton/yr)
Year 1900
Present
2400
1500
-36 (56)
– from 1900 to present %
– (from present to 1900 %)
The nitrogen retention in small lakes is shown to have been 56% larger around
year 1900 than at present. This is due to the larger area of the lakes (estimated
108
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
to be 100% larger than at present) during the earlier period. The retention in
year 1900 is larger despite a smaller load to the lakes in this period compared
with the present.
7.8
Nitrogen retention in larger lakes
Larger lakes are defined in the same way as in the NNM (Højberg et al., 2015b;
Windolf et al., 1996)). The model used to calculate Nret in larger lakes is the
same as in the original NNM. All lakes are connected to the river network and
have an outflow. Monthly lake nitrogen retention is simulated by adding wa-
ter and nitrogen load to each lake at each time step (month) and multiplying
this by a dynamic monthly retention factor (Eq.1 and Eq.2).
��������1
����
(
)
= (1
− ��������
) ×
����
( )
+
����
( )
/
���������������� ������������������������
+
����
( )
N
lake(t+1)
is lake N concentration in month t+1.
FN
ret
is the relative monthly N retention in month t.
N
lake(t)
is lake N concentration in month t.
N
load(t)
is the N load to a lake in month t.
Q
load(t)
is the water load to a lake in month t.
FN is calculated either by Eq2 or Eq3 depending on residence times (T
w
) < 1
year (Eq2) and one for T
w
> 1 year (Eq3). Eq2 from Windolf et al. (1996).
��������2
��������
=
����
×
����
α
is 0.455 (± 0.074 C.L.).
θ
is 1.087 (± 0.014 C.L.).
T is monthly average water temperature (°C) for Danish lakes calculated from
mean monthly air temperature (T
air_t
) in the current month and the previous
month (T
air_t-1
), C.L. is confidence limits.
T = 1.517 + 0.3034 × (T
air_t
) + 0.1909 × (T
air_t-1
) + 0.6347 × (T
air_t
) × Sin ( ×
month(1-12) / 13)
��������3
��������
= 0.01 ×
����
×
����
− ����
_
K is 6.117
M
05
is a month-specific value (Jan=1, Feb=2…Dec=1) (1,2,3,4,5,6,6,5,4,3,2,1)
This model will automatically respond to changed flows, nitrogen loads and
air temperature.
In the period from 1900 to the present, lakes have been drained, and new lakes
have been created. Some have even been both drained and restored during
this period. A full overview of which lakes have disappeared or been created
does not exist; nor are GIS layers available showing the surface areas of many
of these lakes. The NNM includes a functionality applying starting and end
years of the Nret calculation. For lakes with known (by the authors) history,
this information is included. Thus, lakes known to exist only in the year 1900
or at present (2017) are only included during this period. All other lakes are
included for both periods. The bulk of the lake draining in Denmark hap-
pened before 1900 (Figure 7.5). More than 2,700 ha of lakes have been restored
since 1990 (Thodsen, unpublished data). Therefore, the lake area is larger in
109
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0112.png
the present period than it was around year 1900. In total 569 lakes (366 km
2
)
are included for the year 1900 and 611 (422 km
2
) for 2017 (present).
Figure 7.5.
Lake area drained
(Søudtørring) and reclaimed ma-
rine area (inddæmning) in Den-
mark during different eras (Hof-
meister ed., 2004).
Table 7.8.
National large lake nitrogen retention and change between the present and
year 1900.
Period
Larger lakes nitrogen
retention (ton/yr)
Year 1900
Present
5,000
8,200
64 (-39)
Change in larger lakes nitrogen retention
– from 1900 to present %
– (from present to 1900 %)
Nitrogen retention in larger lakes is modelled to have been about 39% smaller
in year 1900 compared with the present period (64% larger in the present com-
pared with year 1900). This is primarily due to an increase in the nitrogen load
to the lakes and, to a lesser extent, due to a slight increase in the number and
surface area of lakes because of lake (re)construction in the later part of the
period since 1900. Because of the generally lower runoff in the year 1900 pe-
riod and consequently larger lake water retention time, the N retention per-
cent is calculated to have been higher (Eq.1-3).
7.9
Total surface water nitrogen retention around year 1900
compared with the present period
The total amount of surface water nitrogen retention around year 1900 and
the present period is calculated in ton per year (Table 7.9). The total surface
water retention have been 8% lower around year 1900 than for the present
period. This is because of a higher load of nitrogen in the present period com-
pared with the time around year 1900, which means that the absolute amount
of nitrogen removed is higher in the present period.
110
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0113.png
Table 7.9.
Estimated (NNM) changes in national surface water nitrogen retention between
year 1900 and the present period.
Period
Surface water nitrogen
retention (ton/yr)
Year 1900
Present
26,000
28,000
9 (-8)
Change in surface water nitrogen retention
– from 1900 to present %
– (from present to 1900 %)
However, when calculating the relative (percent) nitrogen retention of the
load entering streams and rivers around the year 1900, 43% is removed in the
surface water system compared with 33% in the present period.
7.10 National Nitrogen Model – nitrogen mass balance
NNM nitrogen balances for the present period and the period around 1900
were calculated. This was done to ensure that the balance is closed and to
provide an overview of the size of each source and sink (Table 7.10). The N
balance is calculated at national scale. The overall nitrogen mass balances of
the two NNMs are shown to be very close to zero (Table 7.10 – bottom row),
meaning that all nitrogen is routed through the models. The only parts of the
NNM storing nitrogen between time steps are the groundwater routing mod-
ule and the larger lakes module.
It is important to notice that the loads to the sea given in Table 7.10 are not
directly comparable to the real loads given in chapter 8 (see section 7.2 for
detailed explanation).
The national root zone leaching is found to be 15% higher in the present pe-
riod than around year 1900 (Table 7.10). The total national land-based nitro-
gen sources, of which the root zone leaching is by far the largest source, are
found to be 18% higher in the present period than during the year 1900 period
(Table 7.10). The groundwater retention is modelled to be only 3% larger in
the present period despite a 15% larger root zone leaching, resulting in a lower
groundwater retention percentage of 58% compared with 64% around year
1900. This is primarily due to the smaller tile-drained area and the lower pre-
cipitation in 1900. Both circumstances promote a larger fraction of the rainfall
recharging to the groundwater and thereby higher groundwater nitrogen re-
tention. For surface water retention, the largest sink around year 1900 is mod-
elled to be natural wetlands, but due to a large reduction in the wetland area
the largest retention in the present period is modelled to be that in smaller
watercourses. The total surface water retention is 4% higher in the present
period than around 1900, mainly because of the larger load. However, the to-
tal source to surface waters is modelled to be 41% higher, resulting in a larger
surface water retention percentage around year 1900, 43%, compared with
33% in the present period. The land-based nitrogen load to the sea is modelled
to be 65% larger in the present period than around 1900, corresponding to an
about 22,000 ton N increase (Table 7.10). The larger load in the present period
is caused partly by an 18% larger total source and a lower retention percent-
age in both groundwater and surface waters, however.
The year 1900 point sources are in Table 7.10 shown to be 679 ton N/yr. In
chapter 4, the inland point sources from towns with >5,000 inhabitants are
given as 407 ton N/yr. The difference originates from additional point sources
estimated for smaller towns.
111
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0114.png
Table 7.10.
National Nitrogen Model, nitrogen balance for the present and the year 1900. The nitrogen amount removed
through the conceptual reducer has added to the groundwater retention component (for “conceptual reducer” see, Højberg et al.
(2015a,b) and section 6.1)). The presented model output values are rounded to the nearest ton of nitrogen to show the model
nitrogen balance. However, the uncertainty of the individual values is much larger than this.
Present period
Source/sinks
Source
Root zone leaching
Groundwater & tile drains
Point source
(Not including point source loads directly to the sea)
Organic nitrogen
Atmospheric deposition (deposition on lake surfaces)
Total sources
Total sources to surface waters
Sinks/Retention
Groundwater
Groundwater Nret %
Smaller lakes
Smaller watercourses
Constructed wetlands
Natural wetlands
Larger watercourses
Larger lakes
Total surface water sinks
Total sinks
Surface water Nret %
Total Nret %
Load to the sea
Sources – sinks
Modelled load to the sea
N balance error
Ton nitrogen
33
22
% of load to the sea
0.06%***
0.07%***
* Nret groundwater % is percent of leaching removed by retention in groundwater.
** Total surface water Nret % is percent total surface water N source removed as total surface water N retention.
*** N balance error percentage of load to sea is the percent difference between the N load to sea calculated through the mass
balance and the “modeled load to sea” calculated by summing the NNM outlets to the sea.
56,316
56,283
34,048
34,026
65% (-40%)
65% (-40%)
96,995
58%*
1,527
10,441
265
3,268
4,382
8,201
28,084
125,079
33%**
69%
94,157
64%*
2,376
7,523
0
8,087
2,715
5,013
25,712
119,869
43%**
78%
-36% (56%)
39% (-28%)
-
-60% (147%)
61% (-38%)
64% (-39%)
9% (-8%)
4% (-4%)
3% (-3%)
9,559
527
181,395
84,440
6,555
251
153,917
59,760
46% (-31%)
110% (-52%)
18% (-15%)
41% (-29%)
167,702
70,707
3,607
146,432
52,275
679
15% (-13%)
35% (-26%)
431% (-81%)
ton N/yr
1900 period
ton N/yr
Percent change
– from 1900 to present
– (from present to 1900)
7.11 References
Fuglsang, A. 2006. Kvælstofomsætning i våd eng - Forsøgsprojekt langs Storå
på Fyn 1990 – 2005. Wetland Consult & Fyns Amt. 18 pp.
Hoffmann, C.C., Dahl, M., Kamp-Nielsen, L. and Stryhn, H., 1993. Vand- og stof-
balance i en natureng. Miljøprojekt nr. 231, 150pp, Miljøstyrelsen. (in Danish).
112
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0115.png
Hoffmann, C.C., 1998. Nutrient retention in wet meadows and fens. PhD the-
sis, University of Copenhagen, Freshwater Biological Institute, and National
Environ-mental Research Institute, Department of streams and Riparian Ar-
eas, 134pp.
Hoffmann, C.C., Rysgaard, S. and Berg, P., 2000. Denitrification rates pre-
dicted by nitrogen-15 labeled nitrate microcosm studies, in situ measure-
ments, and modeling. Journal of Environmental Quality 29(6), 2020-2028.
Hoffmann, C.C., Berg, P., Dahl, M., Larsen, S.E., Andersen, H.E. and Ander-
sen, B., 2006. Groundwater flow and transport of nutrients through a riparian
meadow – Field data and modelling. Journal of Hydrology 331, 315-335.
Hofmeister, E. (ed.), 2004. De ferske vandes kulturhistorie i Danmark. Aqua,
Ferskvands Akvarium. Danmark. 40pp.
Højberg, A.L., Windolf, J., Børgesen, C.D., Troldborg, L., Tornbjerg, H., Bli-
cher-Mathiesen, G., Kronvang, B., Thodsen, H. and Erntsen, V., 2015a. Natio-
nal kvælstofmodel. Oplandsmodel til belastning og virkemidler. De Natio-
nale Geologiske Undersøgelser for Danmark og Grønland – GEUS, 111pp.
Højberg, A.L., Windolf, J., Børgesen, C.D., Troldborg, L., Tornbjerg, H., Bli-
cher-Mathiesen, G., Kronvang, B., Thodsen, H. and Erntsen V., 2015b. Natio-
nal kvælstofmodel. Oplandsmodel til belastning og virkemidler, bilag. De Na-
tionale Geologiske Undersøgelser for Danmark og Grønland – GEUS, 402pp.
Højberg, A.L., Thodsen, H., Børgesen, C.D., Tornbjerg, H., Nordstrøm, B.O.,
Troldborg, L., Hoffmann, C.C., Kjeldgaard, A., Holm, H., Audet, j., Ellermann,
T., Christensen, J.H., Bach, E.O. and Pedersen, B.F., 2021. National kvælstof-
model – version 2020, Metode rapport. De Nationale Geologiske Undersøgel-
ser
for
Danmark
og
Grønland.
GEUS
Specialrapport.
https://www.geus.dk/Media/637576521860083405/NKM2020_Rap-
port_18maj2021_web.pdf
Iversen, T.M. and Andersen, S.P., 1997. Danish experiences on river restora-
tion – Trends and Statistics. In: Hansen H.O. and Madsen B.L. (eds), River
Restoration ‘96 – Plenary Lectures: International Conference Arranged by the
European Centre for River Restoration. National Environmental Research In-
stitute, 151pp.
Jensen, P.N. (Ed.), 2017. Estimation of Nitrogen Concentrations from root
zone to marine areas around the year 1900. Aarhus University, DCE – Danish
Centre for Environment and Energy, 126 pp. Scientific Report from DCE –
Danish Centre for Environment and Energy No. 241.
http://dce2.au.dk/pub/SR241.pdf
Levin, G., Iosub, C.I. and Rudbeck Jepsen, M., 2017. Basemap02. In: Technical
report from DCE, Aarhus University, DCE - Danish Centre for Environment
and Energy, 68pp.
Saunders, D.L. and Kalff,J., 2001. Nitrogen retention in wetlands, lakes and
rivers. Hydrobiologia 443, 205-212. doi:10.1023/A:1017506914063.
113
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Seitzinger, S.P., Styles, R.V., Boyer, E.W., Alexander, R.B., Billen, G., Howarth,
R.W., Mayer, B. and Van Breemen N., 2002. Nitrogen retention in rivers:
Model development and application to watersheds in the northeastern U.S.A.
Biogeochemistry 57-58, 199-237. doi:10.1023/A:1015745629794.
Windolf, J., Jeppesen, E., Jensen, J.P. and Kristensen, P., 1996. Modelling of
seasonal variation in nitrogen retention and in-lake concentration. A four-year
mass balance study in 16 shallow Danish lakes. Biogeochemistry 33, 25-44.
114
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
8
Calculating the nitrogen load to the sea
Author: Anker Lajer Højberg
1
, Hans Thodsen
2
, Henrik Tornbjerg
2
, Lars Troldborg
1
Quality assurance: Dennis Trolle
2
1
GEUS,
1
DCE,
Geological Survey of Denmark and Greenland
Aarhus University, Department of Ecoscience
Abstract
Purpose:
The purpose of this chapter is to calculate the nitrogen and freshwa-
ter load to the sea for the period around year 1900. Additionally, the mean
annual year 1900 riverine nitrogen concentration is calculated.
Materials and methods:
The calculation of the year 1900 nitrogen and fresh-
water load to the sea is done using a delta change approach, with the official
load estimate from the period 2008–2019 as the baseline (Thodsen et al., 2021).
Regional, monthly delta change factors are calculated from the results pre-
sented in chapters 7 (Table 7.10) and 3, point sources transported directly to
the sea are added from chapter 4.
Results and discussion:
The year 1900 freshwater load to the sea is modelled
to be 297 mm/yr compared with 335 mm/yr in the present period (2008-2019),
and it was thus 11% lower in the period around year 1900. The nitrogen load
is modelled to be approximately 36,000 ton N/yr around year 1900, which is
about 40% less than during the present period, 59,000 ton N/yr. The mean
annual river nitrogen concentration is modelled to be around 2.8 mg N/l
around year 1900 compared with 4.1 mg N/l in the present period.
Conclusion:
The freshwater load to the sea was lower around year 1900 than
during the present period. The nitrogen load to the sea and the nitrogen con-
centration are modelled to be about 40% and 32% lower during the year 1900
period than during the present period, respectively.
8.1
Introduction
The landscape has changed significantly during the last 120 years, with the
present having a higher drainage density, fewer and shorter streams as well
as fewer small lakes and especially wetland areas. The natural removal of ni-
trate and total nitrogen during transport from agricultural fields to the sea has
thus been reduced. This was also found by Andersson and Arheimer (2003)
in their analysis of a Swedish catchment. The present chapter describes the
approach taken to assess the nitrogen load around year 1900, taking the
changes in climate, landscape, agricultural practise, point source loads and
their spatial variations into account.
8.2
Material and methods
The national nitrogen model (NNM) (Højberg et al., 2015) is applied to esti-
mate a spatially distributed nitrogen load to the sea for the entire country (not
including point sources discharged directly to the sea). No data on the nitro-
gen load to the sea exist for the period around year 1900, which prevents a
direct evaluation of the accuracy of the estimates provided by the model. The
115
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0118.png
approach taken in the study has therefore been to use the model to estimate
the relative difference in the load, invoking a common assumption that a
model is better at estimating the relative change between two contrasting con-
ditions rather than providing an absolute estimate. The nitrogen model has
thus been used to estimate the relative difference in the nitrogen load for the
current and historical conditions. This relative change is then forced upon the
official national load estimate for Denmark (point sources applied directly to
the sea excluded) (Thodsen et al., 2021). Thus, the National Nitrogen Model
(NNM) was run for both the present period and for the period around year
1900 (Højberg et al., 2015). Subsequently, the relative difference between the
two NNM runs was used in calculating the total nitrogen load to Danish
coastal waters in the period around year 1900 with the average load (Point
sources directly to the sea excluded) between 2008 and 2019 as the baseline
load (Figure 8.1) (Thodsen et al., 2021). Point sources directly to the sea are
added subsequently to calculate the total load to the sea. This approach is par-
allel to the delta change approach often used in hydrological climate change
studies where climate model-generated climate/weather estimates are not di-
rectly comparable to observed climate values. However, climate model data
from different periods are mutually comparable and the relative difference
can be used in changing observed climate time series to represent the future
climate (e.g. van Roosmalen et al., 2010).
Figure 8.1.
Concept for calculat-
ing the nitrogen load to the sea
year 1900. NNM is the National
Nitrogen Model. The official load
to the sea as calculated in Thod-
sen et al. (2021).
The comparison between the two NNMs was carried out at a mean monthly
scale. The NNM
present
was run on climate/weather data for the 21-year period
1990–2010 (both years included). The NNM
1900
was run with delta-changed
precipitation input baselined from the period 1990-2010, representing the cli-
mate around year 1900 (se chapter 1 for background and method regarding
the delta change method).
The NNM
1900
mimics the year 1900 as closely as possible, while the NNM
present
mimics the year 2017 as closely as possible but with year 2011 agricultural
leaching. 2011 is used as it is the most recent year modelled with the 2015
version of the NNM, on which the present period NNM is based (Højberg et
al., 2015).
It was chosen to model a period of 21 years, for both NNMs, to base the anal-
ysis on long-term climate/weather series rather than single-year climate se-
ries that tend to be unusual in some way for some parameters. In this way, the
load estimates are based on a robust climate foundation.
The NNM results are aggregated at the fourth order coastal catchment level
(Figure 8.2).
8.2.1 Present period
The NNM
present
was run for the period 1990-2010. The target year (year of the
general conditions of the NNM) is 2017, except for the nitrogen leaching from
agricultural fields which is 2011 as it utilises the data set also used in Højberg
et al. (2015).
116
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0119.png
Figure 8.2.
Fourth order catch-
ments (n=315).
8.2.2 Year 1900
The NNM
1900
was run for the period 1890-1910 but based on delta-changed
climate data for the period 1990-2010. The target year is 1900.
8.2.3 The national nitrogen load estimate
The baseline for the year 1900 nitrogen load estimation is the national annual
nitrogen load estimation for 2008-2019 (Windolf et al., 2011; Thodsen et al.,
2021). Originally, the national nitrogen load estimate has a monthly time step
and is based on the fourth order coastal catchment scale (Figure 8.2) (Thodsen
et al., 2021).
The national nitrogen load estimate is reported annually and is based on both
measured water discharge and measured total nitrogen concentrations as well
as modelled values of water discharge and empirically modelled total nitro-
gen concentrations (Windolf et al., 2011; Thodsen et al., 2021).
8.2.4 Calculating delta change factors
Delta change factors for runoff and nitrogen loads need to be calculated to
transfer the differences found between the NNM
present
and NNM
1900
to the na-
tional nitrogen load estimate. It was found that utilising the individual fourth
order catchment scale for calculating delta change factors introduced too
much uncertainty in some usually small fourth order catchments. Therefore,
it was decided to calculate area-weighted mean delta change factors for both
runoff and nitrogen loads on a subset of fourth order catchments in the nine
regions, shown in Figure 8.3, and use these nine delta change factors on all
fourth order catchments within a region.
117
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0120.png
Fourth order catchments on which the regional delta change factor calculation
was based were chosen in the following way:
Fourth order catchments with the 20% highest and lowest values were ex-
cluded.
a.
b.
Minimum 50% of the region area is covered.
Minimum three fourth order catchments are included in a region.
All fourth order catchments >300k m
2
are included.
c.
Figure 8.3.
Nine delta change re-
gions and fourth order coastal
catchments.
8.3
Results
8.3.1 NNM
1900
and NNM
present
Percent differences in runoff, nitrogen load and nitrogen concentration at na-
tional and mean annual level have been calculated based on the regional
monthly delta change factors. The regional monthly delta change factors for
runoff (Figure 8.4) and for nitrogen loads (Figure 8.5) are presented below.
The change in single regions is less certain than at the national scale, both
because of the smaller scale and because of the national-scale delta change
procedure used for year 1900 precipitation.
118
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0121.png
Figure 8.4.
Monthly delta change factors for regional runoff (Q) between the present period (baseline) and the around year
1900 period.
119
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0122.png
Figure 8.5
Monthly delta change factors for regional nitrogen loads between the present period (baseline) and the year 1900
period.
The runoff delta change factors from the western and central part of the coun-
try show lower winter runoff and similar summer runoff for the period
around year 1900 compared with the present. For the eastern part, summer
runoff is markedly larger around year 1900 compared with the present period,
while winter runoff is lower. As for the nitrogen loads, region 1-4 show mark-
edly lower loads around year 1900 compared with the present period, with
winter loads being mostly below half of present period loads. For region 5, 7,
8 and 9, summer and early autumn loads are higher during the year 1900 pe-
riod than during the present period, while winter loads are lower. Only for
region 9, the island of Bornholm, around year 1900 loads are similar (winter
and spring months) or higher (July – October) than the present period loads.
At the national and mean annual scale, the two NNMs yield the results pre-
sented in Table 8.1.
120
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0123.png
Table 8.1.
Mean national differences (absolute and in %) in mean annual runoff, flow-weighted nitrogen concentration and nitrogen
load between the year 1900 period and the present period as calculated by NNM
1900
and NNM
present
.
NNM
1900
Runoff (mm/yr)
Flow-weighted total nitrogen concentration (mg-N/L)
Total nitrogen load (ton N/yr)
292
2.7
34,026
NNM
Present
333
3.9
56,283
Percent difference
14% (-12%)
45% (-31%)
65% (-40%)
8.3.2 Year 1900 total nitrogen and freshwater load to the sea
The monthly, regional delta change factors for runoff (Fig 8.4) and nitrogen
loads (Figure 8.5) were used in calculating the year 1900 runoff, flow-
weighted nitrogen concentrations and nitrogen loads at the fourth order
catchment level and mean monthly and mean annual level. The fourth order
mean monthly national nitrogen load estimate for 2008-2019 is used as the
baseline for the calculations (Thodsen et al., 2021). However, utilisation of the
fourth order coastal catchment scale is limited by the national-scale delta
change method used for year 1900 precipitation and by using the regional ra-
ther than the fourth order coastal catchment scale for the Q and TN delta
change factors (Figure 8.4 and 8.5). This method transfers a mean regional dif-
ference in runoff and nitrogen load, between the year 1900 and the present, to
each fourth order catchment, with the national nitrogen load estimate as the
baseline (Thodsen et al., 2021). As the results presented in Table 7.10 and Table
8.1 are aggregated from the ID15 scale, and the year 1900 results in Table 8.2
are based on the regionalised delta change factors, the ratios between the year
1900 and the present period nitrogen loads and runoffs (presented in Table
8.1 and 8.2) cannot be expected to match completely.
Mean annual year 1900 national freshwater and nitrogen loads as well as flow-
weighted nitrogen concentrations are calculated (Table 8.2) based on fourth
order catchments and on the principals shown in Figure 8.1.
Table 8.2.
Mean national differences (absolute and in %) in mean annual runoff, direct point sources to the sea, total inland
nitrogen load, flow-weighted nitrogen concentration and nitrogen load between the year 1900 period and the present period
(Present = 2008-2019) after applying the delta change parameters (Figure 8.4 and Figure 8.5) to the NNM
1900
values.
1900
Runoff (mm/yr)
Point sources directly to the sea (ton N/yr)
Total inland nitrogen load (ton N/yr)
Flow-weighted total nitrogen concentration (mg N/L)
Total nitrogen load (ton N/yr)
297
2,518
33,550
2.8
36,068
Present/baseline
335
2,710
56,310
4.1
59,020
Percent difference
13% (-11%)
8% (-7%)
68% (-40%)
47% (-32%)
66% (-40%)
The geographical pattern (fourth order catchments) of flow-weighted nitro-
gen concentrations (Figure 8.6), area-specific nitrogen loss (Figure 8.7) and
runoff/freshwater load (Figure 8.8) are shown below.
121
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0124.png
Figure 8.6.
Mean flow-weighted nitrogen concentration for fourth order catchments. The change in single 4
th
order catchments
is less certain than at the national scale, both because of the smaller scale and because of use of a similar national delta
change procedure for around year 1900 precipitation.
The flow-weighted nitrogen concentration in most fourth order catchments is
shown to be lower or in the same category in 1900 as in present. In both periods,
the highest concentrations are seen on the islands and in the northern part of
Jutland. Large parts of central and western Jutland are shown to have concen-
trations below 2 mg N/L, but also northeastern Zealand has low concentrations
(Figure 8.6). In 1900, most of the country lost 5-10 kg N/ha, while during the
present period the most common class is 10-20 kg N/ha (Figure 8.7).
Figure 8.7.
Mean area-specific nitrogen loss for fourth order catchments. The change in single fourth order catchments is less
certain than that at the national scale, both because of the smaller scale and because of use of a similar national delta change
procedure for around year 1900 precipitation.
122
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0125.png
Figure 8.8.
Mean runoff (mm/yr) for fourth order catchments. The change in single fourth order catchments is less certain than
at the national scale, both because of the smaller scale and because of use of a similar national delta change procedure for
around year 1900 precipitation.
The runoff (Figure 8.8) has generally increased since year 1900 due to in-
creased precipitation. The increase is shown to be larger in the western than
in the eastern part of Denmark. A few eastern catchments show decreases.
However, the regional pattern is very uncertain because of the similar na-
tional delta change procedure (single monthly value for the whole country)
used for calculating the year 1900 precipitation.
If the ratio between the whole-country NNM
1900
and NNM
present
(shown in Ta-
ble 7.10 and Table 8.1) of 0.60 is multiplied with the baseline nitrogen load to
the sea of approximately 56,000 ton N/yr (Table 8.2) and adding to this the di-
rect input to the sea from point sources of approximately 2,500 ton N/yr, the
resulting year 1900 nitrogen load to the sea is approximately 36,000 ton N/yr.
8.4
References
Andersson, L. and Arheimer, B., 2003. Modelling of human and climatic im-
pact on nitrogen load in a Swedish river 1885-1994. Hydrobiologia 497, 63-77.
Højberg, A.L., Windolf, J., Børgesen, C.D., Troldborg, L., Tornbjerg, H., Bli-
cher-Mathiesen, G., Kronvang, B., Thodsen, H. and Erntsen V., 2015. National
kvælstofmodel. Oplandsmodel til belastning og virkemidler. De Nationale
Geologiske Undersøgelser for Danmark og Grønland – GEUS, 111pp.
Thodsen, H., Tornbjerg, H., Bøgestrand, J., Larsen, S.E., Ovesen, N.B., Blicher-
Mathiesen, G., Rolighed, J., Holm, H. and Kjeldgaard, A., 2021. Vandløb 2019
- Kemisk vandkvalitet og stoftransport. NOVANA. Aarhus Universitet, DCE
– Nationalt Center for Miljø og Energi, 74 s. - Videnskabelig rapport nr. 452.
http://dce2.au.dk/pub/SR452.pdf
123
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
van Roosmalen, L., Christensen, J.H., Butts, M.B., Jensen, K.H., Refsgaard and
J.C., 2010. An intercomparison of regional climate model data for hydrological
impact studies in Denmark. Journal of Hydrology 380, 406-419.
doi:10.1016/j.jhydrol.2009.11.014.
Windolf, J., Thodsen, H., Troldborg, L., Larsen, S.E., Bøgestrand, J., Ovesen,
N.B. and Kronvang, B., 2011. A distributed modelling system for simulation
of monthly runoff and nitrogen sources, loads and sinks for ungauged catch-
ments in Denmark. Journal of Environmental Monitoring 13, 2645-2658.
doi:10.1039/C1em10139k.
124
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
9
Phosphorus losses from the Danish land
area to the sea around year 1900
Authors: Hans Estrup Andersen
1
, Goswin Heckrath
2
and Gitte Rubæk
2
Quality assurance: Jørgen Windolf
1
1
DCE,
2
DCA,
Aarhus University, Department of Ecoscience
Aarhus University, Department of Agroecology
Abstract
Purpose:
We aim to estimate the total P load to the sea around the year 1900
including contributions from background, other diffuse sources, inland and
direct point sources and taking P retention in lakes into account.
Materials and methods:
The stream water background concentration of total
phosphorus is derived from present-day measurements in natural streams.
Other diffuse phosphorus sources are sought quantified using historical in-
formation and expert judgements and are added to the background contribu-
tion. Around the year 1900, discharge is calculated by the National Water Re-
source Model forced by historical climate data. Water is routed through a net-
work of connected catchments. The streams in each catchment are assigned
specific geo-regional year 1900 stream water TP concentrations. P retention is
included for larger lakes. Finally, contributions from direct point sources are
added to the calculated phosphorus transport, thus arriving at estimates of
total phosphorus loading to the sea around the year 1900. The available data
only allow calculations at a geo-regional scale (Denmark subdivided into nine
geo-regions).
Results and discussion:
Around year 1900, the stream water TP concentration
is estimated to 0.062-0.075 mg P/l, including contributions from natural back-
ground sources, other diffuse sources and inland point sources. The year 1900
TP concentration is equivalent to 60-70% of the present-day stream water TP
concentration (0.1 mg P/l). The contribution from inland and direct point
sources is estimated to 471 ton P, 65-70% of the present-day contribution from
point sources (704 ton P, average 2014-2018). The total around year 1900 P load-
ing to the sea including background, other diffuse sources, indirect and direct
point sources and subtracted P retention in lakes is estimated to 1200-1340 ton
P, which is 60-65% of the present-day P loading (2,021 ton, average 2014-2018).
9.1
Introduction
9.1.1 Background
Measurements of phosphorus (P) concentrations in streams around the year
1900 are scarce. We are only aware of one study (Westermann, 1898), which
includes a limited number of measurements in a few streams with old meth-
odologies, making its representativeness uncertain. Consequently, the P load
to coastal waters around the year 1900 is largely unknown.
Background P concentration is defined as the concentration that can be meas-
ured in anthropogenically undisturbed streams. However, such streams do
125
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0128.png
not exist in intensively cultivated Denmark. Thus, here background concen-
tration is defined as the least anthropogenically disturbed streams.
Bøgestrand et al. (2008) deduced the regional background P concentration
from current P concentrations in streams from minimally disturbed catch-
ments. These concentration estimates were used in the River Basin Manage-
ment Plans 2015 -2021 to calculate the land-based P loading to the sea around
the year 1900. However, the data material in Bøgestrand et al. (2008) is limited
to only 19 streams, and no assessment of around year 1900 diffuse and point
sources was included in the calculation of P loading.
The population in 1900 was lower than today, but a larger proportion lived in
rural areas where facilities to collect human waste were poorly developed.
Farming was already intensive, covering 76.6% of the land area (Table 5.4,
chapter 5) and with a rather high livestock density of 0.89 units/ha (Danmarks
Statistik, 1969), a large population of cattle (1.4 million LU, chapter 5) and
more widespread grazing. Manure storage facilities were poor (chapter 5) and
handling of nutrients in manure much less sophisticated than today, suggest-
ing that P losses from storages, through simple drainage installations around
the farms and from manure deposited or applied on the soil surface, were
larger in 1900 than today.
Currently, the quantification of P transfer from agricultural land to surface
waters along different transport pathways is associated with many difficulties
and large uncertainties (Andersen and Heckrath, 2020). These difficulties are
even more pronounced when trying to quantify losses around year 1900. We
can, however, point towards major differences in factors affecting P losses
from agriculture between today and around year 1900 and suggest whether
these differences would have caused either larger or smaller contributions of
agricultural P to surface waters in 1900 compared with today (Table 9.1).
Table 9.1.
Qualitative comparison of factors affecting the P loss around year 1900 with today.
Increased
contribution
External inputs and thus the net surplus of P in Danish agriculture was low before 1900 (Fig-
ure 9.1), and the stock of P in agricultural soils was much lower than today.
There were many small and scattered farms in 1900, most of them with livestock, which an-
nually must have cycled considerable amounts of P on-farm through livestock in spite of lit-
tle use of inorganic fertilisers.
o
o
Poor facilities for storage of animal manure resulted in losses by runoff from ma-
nure storages and farmyards.
Poor facilities for storage of animal manure, resulting in manure application on
the soil surface outside narrow windows for incorporation, manure application
also in autumn and winter.
o
Animal manure was handled less efficiently than today, i.e., often it was not in-
corporated into the soil soon after application, increasing the risk of surface run-
off losses, and it was typically applied to land close to the farm, leading to local P
enrichment of soils.
No or poor facilities for collecting and treating waste and wastewater in rural areas.
Negligible use of inorganic fertiliser and feed phosphates.
Grazing and thus deposition of manure excreted onto the soil surface was more widespread
than today, increasing the risk of surface runoff-induced P loss.
Organic lowland soils were not yet intensively drained.
Streams and lakes were typically not fenced in, i.e., cattle had direct access to surface wa-
ters, causing poaching and increased sediment and faeces transfer to water.
+
+
+
+
+
+
Decreased
contribution
+
126
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0129.png
Figure 9.1.
Phosphorus balance
for Danish agriculture 1900-2005.
After Kyllingsbæk (2008).
9.1.2 Objectives
The overall objective of this chapter is to estimate the P loading from the entire
Danish land area to the sea around year 1900. To do this, we estimate stream
water background P concentrations using several approaches, assuming that
today’s background concentration resembles the background concentration
in 1900. Subsequently, and to the extent possible, we estimate and add inputs
from additional diffuse and point sources to arrive at ranges of likely stream
water concentrations of total phosphorus (TP) around year 1900. Stream water
transport of P at catchment scale is then calculated using modelled stream
water discharge from historical climate data. The P load is routed downstream
in a connecting river network including P retention in larger lakes, resulting
in a final estimate of P loading to the sea.
9.2
Material and methods
9.2.1 Background phosphorus concentrations in Danish streams
Background concentration inferred from stream measurements (according
to Bøgestrand et al., 2008)
Bøgestrand et al. (2008) did a comprehensive study to determine the back-
ground concentration of P in Danish streams. They mapped the P concentra-
tion in deep reduced groundwater (i.e. both oxygen and nitrate concentra-
tions are below 1 mg/l) and in shallow oxidized groundwater (concentrations
of oxygen > 3 mg/l or nitrate > 1 mg/l) based on 4,406 and 2,211 samples,
respectively (Figure 9.2 and 9.3).
The P concentration in deep, reduced groundwater (Figure 9.2) is high at Ærø,
Als, near Ribe and Esbjerg and towards Ringkøbing, in Vendsyssel and in
parts of western and northern Zealand. These areas are characterized by in-
terglacial marine deposits, which may have a natural high content of P. The P
concentration in the shallow oxidized groundwater (Figure 9.3) is lower and
seemingly without correlation with geological deposits. For the majority of
Denmark, the P concentration in the shallow oxidized groundwater is below
0.06 mg/l.
127
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0130.png
Figure 9.2.
The concentration of
total dissolved phosphorus in re-
duced groundwater (Bøgestrand
et al., 2008).
The P concentrations observed in “nature streams”, i.e., streams draining
catchments without point sources and with less than 10% agriculture, were,
however, not related to the mapped P concentrations in shallow groundwater
as was otherwise initially hypothesized (Bøgestrand et al., 2008). The missing
correlation was explained by a coarse mapping of groundwater (2 km grid),
possible inputs of reduced groundwater, sorption of P in riparian sediments,
declining P concentrations in surface waters and inputs of organic P from wet-
lands.
As an alternative to model background stream water P concentration as a
function of P concentration in groundwater, Bøgestrand et al. (2008) calcu-
lated geo-regional estimates of background concentration of TP as discharge-
weighted means from measurements in 18 “nature streams” (Figure 9.4). P
concentrations ranged from 0.029 mg P/l on Bornholm to 0.089 mg P/l in
Himmerland. For geo-regions without “nature streams”, the overall median
concentration value (0.055 mg P/l) was applied. These concentration esti-
mates were used nationally in the River Basin Management Plans 2015-2021
to calculate the land-based P loading to the sea around the year 1900.
128
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0131.png
Figure 9.4.
Location of the 18
“nature streams” draining catch-
ments without point sources and
with agriculture covering less
than 10%. Also shown are the
nine geo-regions.
Update and expansion of background concentration inferred from stream
measurements
For this analysis, we selected 414 streams included in the Danish national en-
vironmental monitoring program (NOVANA) where water discharge and TP
concentrations were measured during 2010-2017 and calculated the annual
mean discharge-weighted concentration of TP by dividing annual TP
transport with annual discharge. Since TP transport is based on bi-weekly
grab samples, there is a well-known risk of underestimating the true transport
and thus the true discharge-weighted TP concentration. Andersen and Heck-
rath (2020) estimate that the TP transport based on grab samples on average
is underestimated with 30% in small streams and 5% in larger streams. We
derived the annual mean discharge-weighted concentration of TP associated
with diffuse sources for each stream by subtracting the P contributions from
point sources and scattered dwellings not connected to a central sewer (both
reported by Miljøstyrelsen, 2019) from the measured TP concentration. Thus,
in theory, the main remaining stressor causing a P concentration above the
background concentration should be agriculture. Erosion of stream banks and
the associated P loss are considered parts of the background loading. For each
catchment, we collected land use information, including the area occupied by
agriculture. Stream networks including a lake with an area above 5% of the
total catchment area and streams with a catchment area larger than 100 km
2
were removed from the initial dataset to reduce the influence of in-catchment
retention of P. We also excluded catchments with impervious surfaces (e.g.
129
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0132.png
built-up areas) occupying more than 10% of the total catchment area (one
catchment has 13% built-up area), focusing on catchments dominated by na-
ture, forest or agriculture. The reduced dataset comprised 246 streams. Since
this dataset is dominated by streams with catchments with a high proportion
of agriculture, we supplemented the dataset with data from 21 streams iden-
tified by the former counties as the least anthropogenically impacted streams
and with agriculture occupying less than 40% of the total area. These 21
streams were, however, only sampled four times during 2004-2005. From this
dataset of 267 streams, we selected streams where agriculture occupies less
than 20% of the total catchment area. This criterion was met by 26 streams,
including 17 of the “nature streams” analysed by Bøgestrand et al. (2008).
Figure 9.3.
The concentration of
total dissolved phosphorus in oxi-
dised groundwater (Bøgestrand
et al., 2008).
The criterion of 20% agriculture was set based on an initial analysis that
showed two distinct groups having, respectively, below and above 20% agri-
culture, with the lowest TP concentrations in the first group. Allowing the
share of agriculture to increase from 10% to 20% only enhanced the overall
mean TP concentration in the group of streams with 2%, however the number
of streams included increased from 18 to 26. For that reason, it was decided to
include also streams with up to 20% agriculture in the catchment. We then
calculated overall mean values per station so that each station has an equal
weight irrespective of the number of years that the station has been sampled.
Finally, we calculated mean values per geo-region as estimates of background
130
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0133.png
TP concentration. Table 9.2 compares the geo-regional estimates of back-
ground TP concentration from Bøgestrand et al. (2008) with estimates from
the present analysis. In the beginning, NOVANA also comprised monitoring
of water quality of springs. P concentrations in spring water in nature areas
may be considered representative for groundwater-fed streams of nature ar-
eas and thus conservative (minimum) estimates of background concentration
(since contributions from, for instance, bank erosion are not included). TP con-
centrations in 12 springs located in nature areas (Larsen et al., 1995) are shown
as averages for the period 1989-1994 in Table 9.2.
Table 9.2.
Geo-regional estimates of annual discharge-weighted background concentration of total phosphorus (mg TP/l) from
Bøgestrand et al. (2008) and this study; 95% confidence intervals included for the latter study. Also shown is average total
phosphorus concentration (mg TP/l) measured in springs in nature areas during 1989-2004 (Larsen et al., 1995). The number of
streams/springs behind each estimate is given in parentheses.
Georegion
Bøgestrand et al.
ams”
1. Thy
2. North Jutland
3. West Jutland
4. Himmerland
5. North Djursland
6. Central Jutland
7. East Denmark
8. North Zealand
9. Bornholm
0.058 (1)
0.055 (0)
0.043 (3)
0.089 (2)
0.055 (0)
0.040 (2)
0.054 (5)
0.069 (2)
0.029 (3)
0.010 (1)
0.100 (2)
0.043 (4)
0.019
0.061 +/- 0.006
0.066 +/- 0.050
0.020 +/- 0.005
(2)
(7)
(3)
(4)
Larsen et al. (1995): This study: 26 streams with agriculture < 20%
areas
0.090 (1)
0.070 (1)
0.065 (2)
0.010 (1)
sources and scattered dwellings. 95% con-
fidence interval indicated where applicable
0.085
0.074
0.043 +/- 0.020
0.073 +/- 0.015
(1)
(1)
(4)
(4)
(2008): 18 ”nature stre- 12 springs in nature and corrected for P contributions from point
Average and median values in the dataset of Bøgestrand et al. (2008) are 0.052
and 0.055 mg TP/l, respectively. The corresponding values in the dataset from
Larsen et al. (1995) are 0.057 mg TP/l and 0.040 mg TP/l. Average and median
values in our updated analysis including 26 streams with agriculture < 20%
are 0.053 and 0.050 mg TP/l, respectively. Thus, at the national level there is
reasonable agreement between the three sets of estimated background P con-
centrations. However, compared with Bøgestrand et al. (2008), our new anal-
ysis yields considerably higher estimates of background P concentrations for
geo-region 1 and 2 (Thy and Nordjylland). The relatively high background P
estimates in these two geo-regions are supported by findings of Thodsen et
al. (2019), who provided a national overview of suspended matter in streams
based on measurements at 572 stations during 1976-2016 (> 100,000 water
samples). Suspended matter is often perceived as an important carrier of P.
Thodsen et al. (2019b) found very high values of sediment yield in geo-region
1 and especially in geo-region 2 (Figure 9.5). Furthermore, Figures 9.2 and 9.3
demonstrate that also the P concentration in both the reduced and the oxi-
dized groundwater is high in these two geo-regions.
131
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0134.png
Figure 9.5.
Sediment yields for
132 catchments (after Thodsen et
al., 2019).
International data on background concentration
A few studies from neighboring countries with similar geology report back-
ground concentration of P (Table 9.3). The values are generally lower albeit in
the same order of magnitude as the Danish estimates (average values) of back-
ground concentration. Note that the Swedish estimates of background con-
centration do not include contributions from stream bank erosion.
Table 9.3.
Stream water background concentration of total phosphorus (dissolved inorganic phosphorus concentration in paren-
theses) measured in neighbouring countries.
Country
NE Germany
Location
Catchment of
River Spree
Estonia
mg TP/l (mg PO
4
-P/l)
0.030-0.050
(0.010-0.020)
0.031-0.095, avg. 0.045
(0.006-0.045, avg. 0.019)
Latvia
Lielupe River
Basin District
Sweden
Southern
Skåne
0.020-0.056, avg. 0.037
(0.008-0.037, avg. 0.020)
Sand: 0.04
Sandy loam: 0.03
Loamy sand: 0.02
Stream characteristics
Brooks in small unpolluted catch-
ments, paleolimnological investi-
gations in river valley
8 streams in nature areas w. agri- Arvo Iital, Tallinn University
culture < 20%, 1994-2017, sizes
from tens to hundreds of km
2
Source
Gelbrecht et al. (2005);
Schönfelder et al. (2002)
of Technology, pers. comm.
Ainis Lagdzins, Latvia Uni-
versity of Agriculture, pers.
comm.
Karin Blombäck, Swedish
Agricultural University, pers.
comm.
3 streams w. agriculture 10 –
29%, 2005 – 2018, sizes 9 – 28
km
2
Background concentration is cal-
culated with ICECREAMDB
(Persson et al., 2007) for exten-
sive grassland.
132
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0135.png
Which background concentration estimates to use in further calculations?
Based on the review of several attempts to estimate background P concentra-
tion in streams, we choose to base our work on the updated stream-based geo-
regional analysis. However, for geo-region 5, for which there are no observa-
tions, we apply the overall median value of 0.050 mg P/l (Table 9.4). It is im-
portant to stress that (i) these estimates are geo-regional averages and that
several streams with both lower and higher background concentrations can
probably be found within each geo-region, depending on, for instance, con-
nectivity to groundwater systems, and (ii) these estimates come with a very
high degree of uncertainty as they are based on a limited number of streams.
Table 9.4.
Updated geo-regional estimates of present annual discharge-weighted back-
ground concentration of total phosphorus (mg TP/l) including 95% confidence intervals
where applicable. The number of streams behind each estimate is given in parentheses.
Georegion
This study: 26 streams with agriculture < 20%
and corrected for P contributions from point
sources and scattered dwellings. 95% con-
fidence interval indicated where applicable.
1. Thy
2. Nordjylland
3. Vestjylland
4. Himmerland
5. Norddjursland
6. Midtjylland
7. Østdanmark
8. Nordsjælland
9. Bornholm
0.085
0.074
0.043 +/- 0.020
0.073 +/- 0.015
0.050
0.019
0.061 +/- 0.006
0.066 +/- 0.050
0.020 +/- 0.005
(1)
(1)
(4)
(4)
(0)
(2)
(7)
(3)
(4)
9.2.2 Input related to soil drainage
Around the year 1900, an area of 659,000 ha was artificially drained (Aslyng,
1980). Most of the draining was carried out between 1860 and 1900 (Jensen,
1988; Olesen, 2009). Artificial drainage was primarily established on the
loamy soils on the Danish islands and in East Jutland (see Figure 3.4). The tile-
drained area is distributed on geo-regions as follows: Thy 40,000 ha; North
Jutland 11,000 ha; West Jutland 33,000 ha; Central Jutland 47,000 ha; East Den-
mark 509,000 ha; Bornholm 19,000 ha. Draining of low-lying organic soils, pri-
marily situated in Jutland, took place after 1900 and most intensively between
the two world wars and up to 1960 (Olesen, 2009; Olesen, 2019). Hence, we
will not consider P losses from draining of low-lying organic soils further.
Tile drains represent a direct transport pathway from the root zone to surface
waters, and installment of tile drains may thus increase the P loading of surface
waters. However, even today the P concentration in most tile drains is low (An-
dersen et al., 2016), and in 1900 the soil P content at the depth of the tile drains
would not have been enriched above the natural content. However, structured
soils have macropores that may convey water, dissolved and particulate sub-
stances directly from the surface and the topsoil to tile drains during the runoff
season (autumn and winter). Freshly added manure constitutes a mobile source
of P for transport in macropores (Soupir et al., 2006). In 1900, most of the animal
manure was applied in the autumn and winter periods (chapter 5). Poulsen and
Rubæk (2005) estimate that tile drains on minerogenic “low risk” soils today
have loss rates of 20-80 g P/ha and that tile drains on ‘high risk’ soils with active
133
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0136.png
macropores have loss rates of 100-500 g P/ha. Poulsen and Rubæk (2005) esti-
mate that 85% of tile-drained minerogenic soils are “low risk” and 15% are
“high risk”. Assuming similar loss rates and a similar ratio between low and
high-risk soils in year 1900 as today, total P losses from tile-drained soils would
have contributed 21- 94 ton P annually.
9.2.3 Input related to land reclamation
During 1860-1900, about 40,000 ha of land were reclaimed of which 7,500 ha
were drained lakes. The remainder was fjords and coastal areas (Hofmeister,
2004). We assume that P losses from cultivation of former fjords and coastal
areas were low and on par with losses from “low-risk” minerogenic soils.
Lakes, however, have served as sinks in the landscape for particulate P for a
long time, and hence draining of lakes may result in increased P losses. We
have access to data from one drained lake, the 11 ha artificial Brande Elkraft
Sø, which was created in 1910 and drained in November 2013 due to a dam
burst (Hoffmann, 2014). Hoffmann (2014) studied the P loss for three months
from the lake sediments in augered sediment columns (30 x 30 cm) taken im-
mediately after drainage of the lake. The columns were kept at a temperature
corresponding to the average temperature in June, July and August and irri-
gated with artificial rain at rates corresponding to the average precipitation
for these months. Hoffmann (2014) measured an average TP loss of 0.07 kg
P/ha from lake sediments under conditions excluding plant growth and with
favorable conditions for mineralization. Phosphorus losses from the remain-
ing part of the year were not simulated in this laboratory study. Due to lack
of further data from drained lakes, we have additionally included information
on P losses from drained organic soils (Table 9.5). At Skovsbjerggård, Volsted
and Gøderup, the soils were drained 1-2 years prior to the start of the moni-
toring, whereas drainage at Fussingø took place several decades before. The
P losses from these soils exhibit a large variation with highest losses from the
newly drained soils. Based on the data presented here, we estimate the addi-
tional annual loading from drained lakes to range from 0.5 to 1 kg P/ha/yr,
which translates into 3.75-7.5 ton P annually for all drained lakes in Denmark.
Lakes drained during 1860-1900 are located in all geo-regions except Born-
holm (no. 9).
Table 9.5.
Losses of P monitored from drained organic soils. The soils at Skovsbjerggård, Volsted and Gøderup were drained
1-2 years prior to the start of the monitoring, whereas drainage of the soils at Fussingø took place several decades before (after
Andersen et al., 2016).
Location
Skovsbjerggård
Volsted
Gøderup
Soil type
Deep peat, pumped
Deep peat, pumped
40 cm peat on sand, pumped
Period
1988/89
1989/90
1988/89
1989/90
1989/90
1998-2000
Fussingø, Vest, inte-Outlet from deep peat area with tile
grated investigation drains and ditches, permanent grass,
grazing
Fussingø, Øst, inte- Outlet from deep peat area with tile
grated investigation drains and ditches, permanent grass,
grazing
1998-2000
-
0.920
Hoffmann & Ovesen
(2003)
Total-P
mg/l
0.320
(avg. 1988-90)
0.660
(avg. 1988 – 90)
0.100
-
Total-P loss Reference
kg/ha
1.2
0.9
7.0
4.3
1.2
0.670
Hansen et al. (1990)
Hoffmann & Ovesen
(2003)
Hansen et al. (1990)
Hansen et al. (1990)
134
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0137.png
9.2.4 Inputs related to grazing animals and surface-applied animal ma-
nure
The agricultural sector included 2.6 million livestock units in 1898 (Danmarks
Statistik, 1969) with 54% cattle, 16% pigs, 15% horses, 8% poultry and 7% sheep
and goats. The livestock density was 0.89 units/ha on land under agricultural
use, almost equal to the density of 0.86 units/ha in 2017 (Blicher-Mathiesen et
al., 2019). Fresh applications of P on the soil surface excreted by grazing livestock
or spread by man from manure storages may cause “incidental” losses of dis-
solved and particulate P forms in surface runoff when rainfall interacts directly
with manure. Rates of P loss are temporarily and spatially very variable depend-
ing on the amount of P applied, the P release properties of the manure, the timing
of storm events after application and the amounts of runoff generated (Withers
et al., 2003). Table 9.6, which compiles results from international studies on inci-
dental P losses, also demonstrates this. The landscape in Denmark, which is
characterized by flat terrain or mainly gentle slopes and soils with a relatively
high infiltration capacity (especially when covered by grass) and with low inten-
sity rainfall being the most common, suggests that infiltration excess surface run-
off would have been rare in 1900. Therefore “incidental” losses of excreted ma-
nure from grazing animals and surface-applied manure on arable fields are con-
sidered of minor importance around year 1900.
Table 9.6.
Examples of total phosphorus (TP) concentration and loss in surface runoff following precipitation on grazed fields.
Location
Ohio, USA
UK
UK
Nebraska, USA
Precipitation
+1,000 mm/yr
(6- yr avg.)
12.5 mm/hr for 4
hrs
12.5 mm/hr for 4
hrs
689 mm/yr
(3. yr avg.)
39 mm/yr
(3 yr.avg.)
Ungrazed
ca. 1 LU/ha
1.28 mg TP/l
2.14 mg TP/l
Heavily grazed
Lightly grazed
event
event
Surface runoff
mm/yr
Stocking rate
LU/ha
0.6 LU/ha
P concentration
mg TP/l
< 0.1 mg TP/l
P loss
kg TP/ha
0.1 kg TP/ha/yr
Owens et al.
(1989)
0.08 kg TP/ha per Heathwaite et al.
(1999)
(1999)
Schepers & Fran-
cis (1982)
2.9 kg TP/ha per Heathwaite et al.
Reference
However, grazing animals were to a lesser extent than today fenced off from
streams and lakes. Today, at least 2 m wide riparian buffer zones are mandatory
along all natural streams and lakes (> 100 m
2
) in Denmark. Where relevant, the
buffer zones comprise fences preventing livestock access to surface waters. Mis-
management of fencing of stock can be detrimental to water quality, as demon-
strated in many British and New Zealand catchments where cattle activity in
and by rivers, particularly trampling (poaching, treading) of river margins with
associated defecation, is problematic even today (Wilson and Everard, 2017;
McDowell et al., 2003). Line et al. (2000) reported a 76% and 82% decrease in
weekly TP and sediment loads, respectively, following livestock exclusion from
streams in a North Carolina (USA) catchment. Muenz et al. (2006) compared
water quality between three streams without buffer zones and two fenced
streams at a farm in Georgia (USA) with a diversified row crop and beef cattle
operation. They found lower and more stable concentrations of suspended sol-
ids (SS) and dissolved P in the fenced streams; 0.4 mg/l vs. 4.1 mg/l SS and 0.01
mg/l PO
4
-P vs. 0.02 mg/l PO
4
-P in fenced vs unfenced streams, respectively.
Thus, undoubtedly, in 1900 access by livestock to surface waters contributed
significantly to total P load in surface waters. Unfortunately, we do not have
any means to quantify this contribution.
135
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
9.2.5 Input related to soil erosion by water
Water erosion transports dissolved and particulate P from land to water bod-
ies. Today, water erosion on farmland occurs in all regions of Denmark de-
spite the low relief and the typically low to moderate rainfall erosivity in the
country. Water erosion varies strongly in space and time depending on com-
plex interactions between factors related to climate, topography, soil and
cropping. Erosion modelling involving these factors currently provides the
only means of quantifying the risk of soil redistribution by water. The extent
of water erosion and sediment delivery to water bodies has recently been es-
timated by fine-scale spatial modelling for all of Denmark (Onnen et al., 2019).
The model is calibrated with data on riverine sediment yields in several catch-
ments in Denmark and characterizes the average erosion risk over a period of
about 10 years. With the help of calibration, it is possible to account for land-
scape- and land use-specific conditions that are not directly represented by
the input data. Subsequently, the erosion modelling results have been used to
predict erosional P transfer from land to water based on mapped soil P status
information and assumptions of sediment P content (Andersen and Heckrath,
2020). However, due to the lack of observational data on such P transfer, the
model predictions could not be validated.
In principle, a similar approach could be used for estimating P delivery to
waterbodies around the year 1900. However, this would require spatially ex-
plicit model input data for all of Denmark being representative of a period of
about 10 years around year 1900 of similar quality as in 2019. While topogra-
phy and soil properties will not have changed very much, most of the other
input data are not available in necessary detail. Importantly, data for model
calibration are not available either. This makes a sophisticated modelling ap-
proach redundant. In the following, it is considered how different conditions
around the year 1900 may have influenced the risk of sediment and thus P
delivery to surface waters in general.
Climate
Annual rainfall was in general lower around the year 1900 than today (chapter
1), suggesting a lower risk of surface runoff. However, the climate factor of
the erosion model is based on hourly rainfall intensity data from several cli-
mate stations in Denmark and periods of about 10 years. These data are not
available, and it is hence not possible to estimate whether rainfall was locally
more or less erosive than today.
Cropping
Different crops and cropping practices affect the erosion risk differently. Ero-
sion risk assessment requires sound information on typical crop rotations. The
widespread occurrence of bare fallow and ploughed fields during winter is
expected to have increased the risk of water erosion. However, a lower pro-
portion of winter cereals, the absence of maize and especially a larger propor-
tion of grass on farmland compared with today (chapter 5) would have low-
ered the risk.
Landscape structure and landscape elements
The spatial arrangement of different land uses in landscapes and the land-
scape structure in general have an essential influence on the soil redistribution
by water and sediment delivery to streams. Grassland along streams would
act as buffer zones retaining sediment. Likewise, field borders and narrow,
vegetated strips between fields act as barriers for runoff and sediment trans-
fer. Smaller fields with greater length of field borders as well as the frequency
136
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0139.png
of riparian grasslands around the year 1900 would have reduced hydrological
connectivity in landscapes and hence the risk of sediment transfer.
Soil P content
No historical measurements of soil P are available. However, mass balance
calculations for Danish agriculture indicate an average P accumulation in ag-
ricultural soils of ca. 1,100 kg P/ha during 1900-1987 (Kyllingsbæk, 2008). The
average soil P content measured in 338 Danish agricultural soils in 1987 was,
respectively, 526 mg P/kg and 381 mg P/kg in topsoil (0-25 cm) and subsoil
(25-50 cm) (Rubæk et al., 2013). A back-calculation suggests an average topsoil
concentration of ca. 340 mg P/kg around the year 1900.
Although erosion and sediment delivery to water bodies were possibly lower
around the year 1900 than in the early 21
st
century, it was not possible to quan-
tify by how much. It is therefore our best, albeit rough, estimate to assume the
same regional sediment delivery to waters at both times. The potential P load
is then calculated assuming a total P (TP) concentration in sediment of 340 mg
P/kg (Table 9.7).
Table 9.7.
Estimated annual phosphorus input by water erosion to surface waters per
geo-region assuming a phosphorus content of 340 mg TP/ kg in the eroded sediment. The
regional sediment delivery is based on Onnen et al. (2019).
Georegion
Model-estimated sediment
input to surface waters,
ton
1. Thy
2. Nordjylland
3. Vestjylland
4. Himmerland
5. Norddjursland
6. Midtjylland
7. Østdanmark
8. Nordsjælland
9. Bornholm
Total
8,400
15,000
15,000
6,400
1,500
14,000
27,000
2,300
2,300
Model-estimated phosphorus
load to surface waters,
ton P
2.8
5.1
5.1
2.2
0.5
4.7
9.3
0.8
0.8
31.3
Input related to uncovered animal manure storages, farmyards, roads
and tracks
Around the year 1900, storage tanks for liquid animal manure and roof-covered
manure heaps were uncommon: 28,000 tanks in 1896 and 16,500 roof-covered
manure heaps in 1907 (Iversen, 1944), whereas the number of farms and small
holdings was 237,000 to which should be added 35,000 holdings without land
(Christensen, 1985 and chapter 5). Such unprotected manure storages will serve
as local micro point sources of P to surrounding surface waters if connected by
a transport pathway. Additionally, farmyards, outdoor areas with high density
of traffic (cattle tracks etc.) close to farms where manure will be deposited and
where trampling and traffic compact the soil, may serve as local hotspots/point
sources. Today, contributions from storage of animal manure are practically
eliminated due to improved handling of storage facilities for animal manure. It
was also assumed that the contribution from farmyards is lower today com-
pared with around year 1900 due to better handling of manure and collection
and treatment of runoff water from impervious surfaces.
137
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0140.png
Unfortunately, no data allowing proper quantification of the contribution
from such micro hotspots in Denmark are available, neither from more recent
times nor from around the year 1900. Some examples of P concentrations from
such areas monitored recently and reported in studies from UK and USA
(summarised in Table 9.8) show that the concentration of P leaving such areas
can be very high but also that the variation in concentrations is large. The con-
centrations leaving similar areas in Denmark would probably be of the same
range and variation, but we do not have data on the number and the sizes of
such areas and to which extent manure storages, farmyards, roads and tracks
were hydrologically connected to surface waters.
Table 9.8.
Concentrations of P measured in case studies in the UK and US (Delaware).
Reference
Hively et al. (2005)
Site description
Dairy farm, heifer
barnyard with
heavy manure de-
posits
Hively et al. (2005)
Cow path leading
up from heifer
stream crossing
Edwards and Withers
(2008)
Edwards and Withers
(2008)
Edwards and Withers
(2008)
Edwards and Withers
(2008)
Edwards and Withers
(2008)
Edwards and Withers
(2008)
Edwards and Withers
(2008)
Withers et al. (2009)
Sheep handling ar- UK
eas
General purpose
machine area
Roads
UK
2,030 (61-
17,272)
Withers et al.,( 2009)
*Dissolved reactive P
Farmyards
UK
2,710 (72-20,010) 1,108 (61-
5,680)*
381(4-1,330)* 2 years measurements from 3
micro watersheds, 29 observa-
tions
2 years measurements from 3 mi-
cro watersheds, 24 observations
UK
3 (0-5)
0.1 (0-0.2)*
Data from DEFRA WAO523
13
12*
Data from DEFRA WAO523
Dairy cow yards
UK
54 (12-115)
28 (9.82)*
Track runoff
UK
2.7 (0.24-7.3)
Road runoff
?
0.3 (0.26-0.34)
Pig slurry
Farmyard runoff
Mainly
UK
?
41.1 (39.4-43.6)
30.8 (0.02-247)
-
Delaware,
US
1.0
0.2
Country
Delaware,
US
Total P
mg/L
13.2
Total dis-
solved P mg/L
11.6
Overland flow measurements
1x2 m plots simulated rainfall for
38 mm for 30 minutes, repli-
cated plots 2 time points
Overland flow measurements
1x2 m plots simulated rainfall for
30 minutes
33 observations gathered from
other sources
Data from Lee et al. (2004), 3
observations
2 observations from Mitchell
(2001)
13 observations (Withers pers.
Com.)
Data from DEFRA WAO523
Remarks on measurement
9.3
Results and discussion
9.3.1
Stream water concentrations of total phosphorus (TP) around the
year 1900
Regional estimates of stream water concentrations of TP are calculated in the
following way: first geo-regional TP inputs from diffuse and inland point
sources are converted to discharge-weighted average annual TP concentra-
tions by dividing the inputs with annual geo-regional water discharge (aver-
age for the period 1890-1910, chapter 3). Secondly, these concentrations are
added to the geo-regional background TP concentration estimated in section
9.2.1. Tables 9.9a and 9.9b give an overview of the estimated contributions
138
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0141.png
from diffuse sources and point sources (chapter 4) and the resulting stream
water TP concentrations.
Table 9.9a.
Estimated P inputs from diffuse sources per geo-region. Minimum and maximum values are calculated, respec-
tively, by assuming no inputs from grazing livestock and manure storages, and by assuming that inputs from these two sources
each are at an equivalent size as input from soil erosion.
Geo-region
Soil
drainage
ton P
1. Thy
2. North Jutland
3. West Jutland
4. Himmerland
5. North Djursland
6. Central Jutland
7. East Denmark
8. North Zealand
9. Bornholm
Total
1.3-5.7
0.4-1.6
1.1-4.7
0
0
1.5-6.7
16.3-72.8
0
0.6-2.7
21-94
Land
reclamation
ton P
0.3-0.5
0.3-0.6
0.9-1.8
0.4-0.7
0.1-0.2
0.4-0.8
1.4-2.7
0.1-0.2
0
3.75-7.5
Grazing
livestock
ton P
0-2.8
0-5.1
0-5.1
0-2.2
0-0.5
0-4.7
0-9.3
0-0.8
0-0.8
0-31.3
Soil
erosion
ton P
2.8
5.1
5.1
2.2
0.5
4.7
9.3
0.8
0.8
31.3
Manure storages
etc.
ton P
0-2.8
0-5.1
0-5.1
0-2.2
0-0.5
0-4.7
0-9.3
0-0.8
0-0.8
0-31.3
Table 9.9b.
Summary of calculations of year 1900 stream water TP concentration. Annual water discharge calculated with the
National Water Resources Model (DK-model) forced by historical climate data (chapter 3). Inputs from diffuse sources from Ta-
ble 9.9a. Inputs from inland point sources from chapter 4 (Table 4.3 includes five major inland towns, data shown here include
inputs from additionally 75 inland towns). Increase in TP concentration: The sum of diffuse and inland point sources divided by
the annual water discharge. Background TP concentration: Table 9.4.
Geo-region
Area
km
2
1. Thy
2. North Jutland
3. West Jutland
4. Himmerland
5.North Djursland
6. Central Jutland
7. East Denmark
8. North Zealand
9. Bornholm
Total
1) Area-weighted average concentration.
3,059
3,582
9,972
4,155
899
4,412
15,324
994
589
Annual
discharge
1,000 m
3
838,924
1,037,032
4,383,591
1,236,308
212,630
1,567,855
3,282,947
164,726
131,121
Diffuse
sources
ton P
4.4-14.7
5.8-17.5
7.0-21.8
2.6-7.3
0.6-1.7
6.6-21.6
26.9-103.4
0.9-2.6
1.4-5.1
56.1-195.6
Inland po- Increase in TP Background TP
int sources concentration
ton P
0.2
0.7
2.9
0.3
0.0
1.6
4.1
0.5
0.1
10.5
mg P/l
0.005-0.018
0.006-0.018
0.002-0.006
0.002-0.006
0.003-0.008
0.005-0.015
0.009-0.033
0.008-0.019
0.011-0.040
concentration
mg P/l
0.085
0.074
0.043
0.073
0.050
0.019
0.061
0.066
0.020
Resulting TP
concentration
mg P/l
0.090-0.103
0.080-0.092
0.045-0.049
0.075-0.079
0.053-0.058
0.024-0.034
0.070-0.094
0.074-0.085
0.031-0.060
0.062-0.075
1
The national average stream water TP concentration around the year 1900 was
in the range 0.062 – 0.075 mg P/l, which is 60-70% of the present-day stream
water TP concentration of 0.1 mg P/l (median of more than 200 stream water
stations in NOVANA, Thodsen et al., 2019a).
9.3.2 Comparison of estimated year 1900 stream water concentration
values with lake P concentrations estimated in paleolimnological
studies and to historical data
Amsinck et al. (2003) analyzed sediment samples representing year 1900 from
17 Danish lakes for diatom subfossils. By applying existing transfer functions,
they reached estimates of average annual in-lake TP concentrations at year
139
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0142.png
1900 (Table 9.10). Average and median TP concentrations for the 17 lakes were
91 µg/l and 78 µg/l, respectively.
Table 9.10.
Estimates of average annual in-lake concentrations of TP in Danish lakes
around the year 1900 inferred from sediment diatom analysis (Amsinck et al., 2003).
Lake name
Geo-region
Estimated in-lake total phos-
phorus concentration year 1900
µg/l
Helle Sø
Møllesø
Sjørupgårde Sø
Vallum Sø
Velling Igelsø
Vedsted Sø
Skærsø
Ormstrup Sø
Hostrup Sø
Hvidsø
Agsø
Avnsø
Huno Sø
Søbo Sø
Vedsø
Sønderby sø
Agersø
1
2
4
5
6
6
6
6
7
7
7
7
7
7
7
7
8
175
45
48
150
15
78
70
130
80
53
25
25
145
100
160
220
27
Historical measurements of P concentrations in stream water are extremely
scarce. However, Professor T. Westermann collected water samples during
November 1889 from several large streams in Denmark. He took care to col-
lect the samples upstream of cities to avoid influence from sewage. The sam-
ples were analyzed with an old method whose results we judge to resemble
the TP content as measured by today’s method fairly well, albeit it has a rela-
tively high detection limit (based on personal communication with Professor
Ole Borggaard, Copenhagen University). Due to the elevated detection limit,
it was only possible to quantify the P concentration in samples taken during
autumn but not from samples taken at other times of the year (Table 9.11).
Seven out of 17 investigated lakes (Table 9.10) have year 1900 TP
concentrations lower than today (0.015-0.053 mg TP/l). The remaining lakes
have TP concentrations resembling present-day in-lake concentrations
(compared withto NOVANA lakes (Johansson et al., 2021). The historical
measurements of TP concentrations in streams are, on average, 22% lower
than present-day values. The lake sediment studies and the historical stream
measurements support our finding that stream water TP concentrations
around the year 1900 were lower than today but considerably higher than the
background concentration.
140
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0143.png
Table 9.11.
P concentration in water samples collected in November 1889 (Westermann, 1898) compared with water sam-
ples collected in November 2013-2017 at the same sampling sites.
Station id
Geo-region
Westermann (1898)
mg TP/l
Gudenå upstream Silkeborg
Storå
Kongeå
Suså
Outlet Søborg Sø
Odense Å
210087
220057
360009
570058
480010
450003
6
3
3
7
8
7
0.090
0.130
0.090
0.090
0.070
0.110
NOVANA 2013 –
2017
mg TP/l
0.069
0.122
0.140
0.119
0.190
0.102
9.3.1 Phosphorus transport to the sea
The around year 1900 discharge is calculated (chapter 3) with the National
Water Resources Model (DK-model) forced by historical climate data (chap-
ters 1 and 2) and routed through a network of connected catchments. Average
catchment size is 15 km
2
, Denmark being subdivided into more than 3,000
catchments. The streams in each catchment are assigned the specific geo-re-
gional year 1900 TP concentration range (Table 9.9b). The monthly stream wa-
ter TP transport is calculated by multiplying TP concentration by stream wa-
ter discharge aggregated to monthly values. During transport, P retention in
611 larger lakes is taken into account. Retention is calculated on a per quarter
basis as a percentage reduction in input to the lake (Table 9.12). Retention
rates are based on a study of 16 shallow lakes (Søndergaard et al., 2001). The
overall P retention may, however, be underestimated since small lakes are
omitted from the calculation. P retention by sedimentation of particulate P
during flooding of riparian areas is not considered due to lack of data.
Table 9.12.
Per quarter retention of P in lakes (Søndergaard et al., 2001).
Quarter
1
2
3
4
Lake P retention, % of input
14.3
-15.2
-13.9
14.1
Inputs from direct point sources (coastal cities, chapter 4) are finally added at
the outlets to the sea arriving at figures for the total P transport from the Dan-
ish land area. Overall, TP transport to the sea around year 1900 was within
the range 1,200-1,340 ton P (Table 9.13). Present-day TP loading to the sea is
2,021 ton P (average 2014-2018; Andersen and Heckrath, 2020): thus the
around year 1900 TP load constitutes 60-65% of the present-day loading. The
total loading to the sea around year 1900 translates into an average TP con-
centration of 93-104 mg TP/l, which can be compared with a present-day av-
erage TP concentration in the total load of 130 mg TP/l (Thodsen et al., 2021).
The relatively high around year 1900 TP concentration in the total load com-
pared with today is partly explained by the lower year 1900 freshwater runoff
(approx. 12% lower than today, chapter 3). Table 9.13 summarizes the calcu-
lations.
141
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0144.png
Table 9.13.
Summary of calculation of total P loading to the sea around the year 1900.
Background concentration
Diffuse sources on top of background contribution
Inland point sources
Stream water TP concentration
P transport to the sea from inland sources
P retention in lakes
Direct point sources (coastal cities)
Total P loading to the sea AD 1900
Total freshwater runoff, Denmark
Average TP concentration in total load to the sea
0.050 mg TP/l
56-196 ton P
11 ton P
748-887 ton P
6-8 ton P
460 ton P
1,200 – 1,341 ton P
12,855 mio. m
3
Median value, area-weighted average for Den-
mark (Table 9.4)
Table 9.9a
Chapter 4 and Table 9.9b
This chapter
This chapter
Chapter 4
This chapter
Chapter 3
0.062-0.075 mg TP/ l Average 1890-1910, Table 9.9b
93 – 104 mg TP/l
9.4
Conclusion
The around year 1900 stream water TP concentration is estimated at to 0.062-
0.075 mg P/l, which is equivalent to 60-70% of the present-day stream water
TP concentration (0.1 mg P/l). The contribution from inland and direct point
sources is estimated to 471 ton P, 65-70% of the present-day value (704 ton P,
average 2014-2018). The total around year 1900 P loading to the sea, including
background, other diffuse sources, inland and direct point sources and sub-
tracted P retention in lakes, is estimated to 1,200-1,340 ton P, 60-65% of the
present-day P loading (2,021 ton, average 2014-2018). The corresponding av-
erage TP concentration in the total load to the sea around the year 1900 was
93-104 mg TP/l, 70-80% of the present-day average TP concentration (130 mg
TP/l).
9.5
References
Andersen, H.E. and Heckrath, G. (red.), 2020. Fosforkortlægning af dyrk-
ningsjord og vandområder i Danmark. Aarhus Universitet, DCE - Nationalt
Center for Miljø og Energi, 340 s. - Videnskabelig rapport nr. 397.
Andersen, H.E., Baatrup-Pedersen, A., Blicher-Mathiesen, G., Christensen,
J.P., Heckrath, G., Nordemann Jensen, P. (red.), Vinther, F.P., Rolighed, J., Ru-
bæk, G. and Søndergaard, M., 2016. Redegørelse for udvikling i landbrugets
fosforforbrug, tab og påvirkning af Vandmiljøet. Aarhus Universitet, DCE –
Nationalt Center for Miljø og Energi, 86 s. - Teknisk rapport fra DCE - Natio-
nalt Center for Miljø og Energi nr. 77
http://dce2.au.dk/pub/TR77.pdf
Amsinck, S.L., Johansson, L.S., Bjerring, R., Jeppesen, E., Søndergaard, M.,
Jensen, J.P., Jensen, K., Bradshaw, E., Anderson, N.J., Bennike, O., Nielsen,
A.B., Rasmussen, P., Ryves, D., Stavngaard, B., Brodersen, K., McGowan, S.,
Odgaard, B.V. and Wolin, J., 2003: Vandrammedirektivet og danske søer. Del
2: Palæoøkologiske undersøgelser. Danmarks Miljøundersøgelser. 120 s. –
Faglig rapport fra DMU nr. 476 (in Danish).
Aslyng, H.C., 1980. Afvanding i jordbruget. København, DSR Forlag, Den
Kongelige Veterinær- og Landbohøjskole.
Blicher-Mathiesen, G., Holm, H., Houlborg, T., Rolighed, J., Andersen, H.E.,
Carstensen, M.V., Jensen, P.G., Wienke, J., Hansen, B. and Thorling, L., 2019.
142
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Landovervågningsoplande 2017. NOVANA. Aarhus Universitet, DCE – Na-
tionalt Center for Miljø og Energi, 222 s. - Videnskabelig rapport nr. 305.
Bøgestrand, J., Kronvang, B., Ovesen, N.B., Nyegaard, P. and Troldborg, L.,
2008. Baggrundskoncentrationen af næringsstoffer i grundvand og vandløb.
Vand & Jord 15, 113-116 (in Danish).
Christensen, J. 1985. Landbostatistik. Håndbog i dansk landbohistorisk stati-
stik 1830-1900. Landbohistorisk Selskab, København.
Danmarks Statistik, 1969. Statistisk årbog 1969. Danmarks Statistik, Køben-
havn 1969.
Edwards, A.C. and Withers, P., 2008. Transport and delivery of suspended
solids, nitrogen and phosphorus from various sources to freshwaters in the
UK. Journal of Hydrology 350, 144-153.
Gelbrecht, J., Lensfeld, H., Pöthig, R. and Opitz, D., 2005, Temporal and spa-
tial variation of phosphorus input, retention and loss in a small catchment of
NE Germany. Journal of Hydrology 304, 151-165.
Hansen, B., Hansen, A.C., Hoffmann, C.C. and Nielsen, H. ,1990: Vand- og
stofbalance på lavbundsjord. – NPo-forskning fra Miljøstyrelsen, nr. C14.
Heathwaite, L. and Sharpley, A., 1999. Evaluating measures to control the im-
pact of agricultural phosphorus on water quality. Water Science and Technol-
ogy 39, 149-155.
Hively, W.D., Bryant, R.B. and Fahey, T.J., 2005. Phosphorus concentrations
in overland flow from diverse locations on a New York dairy farm. Journal of
Environmental Quality 34, 1224-1233.
Hoffmann, C.C. and Ovesen, N.B., 2003: Næringsstofomsætning og -tab ved
ekstensiv græsning på lavbundsarealerne ved Fussingø. I: Hald, A.B., Hoff-
mann, C.C. and Nielsen, L. (red.): Ekstensiv afgræsning af ferske enge – bota-
nisk diversitet, småpattedyr, miljø og produktion. – DJF rapport, Markbrug 91.
Hoffmann, C.C., 2014. Udvaskningsforsøg med sedimentsøjler fra Brande El-
værk Sø. Unpublished note, Aarhus University, Institute for Bioscience (in
Danish).
Hofmeister, E. (ed.), 2004. De ferske vandes kulturhistorie i Danmark. Aqua,
Ferskvands Akvarium. Danmark. 407pp.
Jensen, S.P., 1988, Statistisk oversigt. Side 250-265 i red. C. Bjørn: Det danske
landbrugs historie III – 1810-1914. Landbohistorisk Selskab.
Johansson, L.S., Søndergaard, M. and Andersen, P.M., 2021. Søer 2019. NO-
VANA. Aarhus Universitet, DCE – Nationalt Center for Miljø og Energi, 68 s.
- Videnskabelig rapport nr. 417.
Kyllingsbæk, A. 2008. Landbrugets husholdning med næringsstoffer 1900 -
2005. DJF Markbrug nr. 18, aug. 2008 (in Danish).
143
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Larsen, S.E., Erfurt, J., Græsbøll, P., Kronvang, B., Mortensen, E., Nielsen, C.,
Ovesen, N.B., Paludan, C., Rebsdorf, Aa, Svendsen, L.M. and Nyegaard, P.,
1995, Ferske vandområder - Vandløb og kilder. Vandmiljøplanens Overvåg-
ningsprogram 1994. Danmarks Miljøundersøgelser. Faglig rapport fra DMU
nr. 140 (in Danish).
Line, D.E., Harman, W.A., Jennings, G.D., Thompson, E.J. and Osmond, D.L.,
2000. Nonpoint-source pollutant load reductions associated with livestock ex-
clusion. Journal of Environmental Quality 29, 1882-1890.
McDowell, R.W., Drewry, J.J., Muirhead, R.W. and Paton, R.J., 2003. Cattle
treading and phosphorus and sediment loss in overland flow from grazed
cropland. Australian Journal of Soil Research 41, 1521-1532.
Miljøstyrelsen. 2019: Punktkilder 2017. (in Danish)
https://www2.mst.dk/Udgiv/publikationer/2019/02/978-87-7038-042-3.pdf
Muenz, F.K., Golladay, S.W., Vellidis, G. and Smith, L.L., 2006. Stream buffer
effectiveness in an agriculturally influenced area, Southwestern Georgia: Re-
sponses of water quality, macroinvertebrates, and amphibians. Journal of En-
vironmental Quality 35, 1924-1938.
Olesen, S.E., 2009. Kortlægning af potentielt dræningsbehov på landbrugsare-
aler opdelt efter landskabselement, geologi, jordklasse, geologisk region samt
høj/lavbund. DJF Intern Rapport Markbrug nr. 21, marts 2009, Det Jordbrugs-
videnskabelige Fakultet, Aarhus Universitet, Tjele.
Olesen, S.E., 2019. Afvanding i Den Store Danske, Gyldendal. Hentet 2. juni
2019 fra
http://denstoredanske.dk/index.php?sideId=33889
Onnen, N., Heckrath, G., Olsen, P., Greve, M., Pulens, J.W.M., Kronvang, B.
and Van Oost, K., 2019. Distributed water erosion modelling at fine spatial
resolution across Denmark. Geomorphology 342, 150-162.
Owens, L.B., Edwards, W.M and Van Keuren, R.W., 1989: Sediment and nu-
trient losses from an unimproved, all-year grazed watershed. Journal of En-
vironmental Quality 18, 232-238.
Persson, K., Larsson, M., Johnsson, H. and Lindsjö, A., 2007. ICECREAMDB
1.0.34 - Technical de-scription.
Teknisk Rapport 117.
Avdelningen för vat-
tenvårdslära, Swedish Agricultural University, SLU.
Poulsen, H.D. and Rubæk, G.H. (eds.), 2005. Fosfor i dansk landbrug. DJF rap-
port Husdyrbrug nr. 68. 2005. 211pp.
Rubæk, G.H., Kristensen, K., Olesen, S.E., Østergaard, H.S. and Heckrath, G.,
2013. Phosphorus accumulation and spatial distribution in agricultural soils
in Denmark. Geoderme 209-210, 241-250.
Schepers, J.S. and Francis, D.D., 1982: Chemical water quality of runoff from
grazing land in Nebraska: I. Influence of grazing livestock. Journal of Envi-
ronmental Quality 11, 351-354.
144
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Schönfelder, I., Gelbrecht, J., Schönfelder, J. and Steinerg, C.E.W., 2002. Rela-
tionships between littoral diatoms and their chemical environment in North-
eastern German lakes and rivers. Journal of Phycology 38, 66-82.
Soupir, M.L., Mostaghimi, S. and Yagow, E.R., 2006. Nutrient transport from
llivestock manure applied to pastureland using phosphorus-based manage-
ment strategies. Journal of Environmental Quality 35, 1269-1278.
Søndergaard, M., Jensen, J.P. and Jeppesen, E., 2001. Retention and internal
loading of phosphorus in shallow, eutrophic lakes. The Scientific World Jour-
nal 1, 427-442.
Thodsen, H., Tornbjerg, H., Rasmussen, J.J., Bøgestrand, J., Blicher-Mathiesen,
G., Larsen, S.E., Ovesen, N.B., Windolf, J. and Kjeldgaard, A., 2019a. Vandløb
2017. NOVANA. Aarhus Universitet, DCE – Nationalt Center for Miljø og
Energi, 74 s. - Videnskabelig rapport nr. 306 (in Danish)
http://dce2.au.dk/pub/SR306.pdf
Thodsen, H., Rasmussen, J., Kronvang, B., Andersen, H.E., Nielsen, A. and
Larsen, S.E,. 2019b: Suspended matter and associated contaminants in Danish
streams: a national analysis. Journal of Soils and Sediments.
https://doi.org/10.1007/s11368-019-02320-8.
Thodsen, H., Tornbjerg, H., Bøgestrand, J., Larsen, S.E., Ovesen, N.B., Blicher-
Mathiesen, G., Rolighed, J., Holm, H. and Kjeldgaard, A., 2021. Vandløb 2019
- Kemisk vandkvalitet og stoftransport. NOVANA. Aarhus Universitet, DCE
– Nationalt Center for Miljø og Energi,74 s. - Videnskabelig rapport nr. 452
http://dce2.au.dk/pub/SR452.pdf
Westermann, T., 1898. Om indholdet af plantenæring I vandet fra vore vand-
løb. Tidsskrift for Landbrugets Planteavl 4, 157-165. (in Danish)
Wilson, J.L. and Everard, M., 2017. Real-time consequences of riparian cattle
trampling for mobilization of sediment, nutrients and bacteria in a British
lowland river. International Journal of River Basin Management, TBD, 1-37.
Withers, P., Ulen, B., Stamm, C. and Bechmann, M., 2003. Incidental phospho-
rus losses – are they significant and can they be predicted? Journal of Plant
Nutrition and Soil Sciience 166, 459-468.
Withers, P., Jarvie, H., Hodgkinson, R., Palmer-Felgate, E., Bates, A., Neal, M.,
Howells, R., Withers, C. and Wickham, H., 2009. Characterization of phospho-
rus sources in rural watersheds. Journal of Environmental Quality.
doi10.2134/jeq2008.0096
145
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
10 Uncertainty and sensitivity
Authors: Hans Thodsen
1
, Jørgen Eriksen
2
, Lars Troldborg
3
, Martin Rygaard
4
, Flem-
ming Vejen
5
, Hans Estrup Andersen
1
Quality assurance: Dennis Trolle
1
Aarhus University, Department of Ecoscience
Aarhus University, Department of Agroecology
3
GEUS, Geological Survey of Denmark and Greenland
4
DTU-Technical University of Denmark
5
DMI- Danish Meteorological Institute
2
DCA,
1
DCE,
In this chapter, some considerations on uncertainty and sensitivity are given.
For some variables, the considerations have been quantified, but for most var-
iables only qualitative considerations are presented.
Not all aspects concerning the uncertainty and sensibility of the model com-
plex have been sought to be quantified as most of the uncertainty analysis was
exempted from the project due to limitation of resources. The uncertainty
deals with the degree of certainty/precision to which a parameter, variable or
a process can be estimated/determined. Sensitivity is the effect of a parame-
ter, variable or process on the result; in this case, the result is the estimated
national nitrogen or phosphorus load to the sea around the year 1900.
The focus on the year 1900 (120 years ago) of course makes most aspects of
calculating the national nitrogen load more uncertain than when calculating
it for the present. For some variables, present measurements are available,
and these can be used either directly for modelling the nitrogen load or as
calibration and validation data. For the time around year 1900, much less data
on many fewer variables are available. Therefore, parameter and variable
value estimates and modelling results are more uncertain for this period. The
knowledge of how, for example, the farming systems worked or how point
sources were managed around year 1900 is more limited than for the present,
and the uncertainty around year 1900 is consequently larger than for the pre-
sent period (sections 4.3.5 and 5.2). Precipitation data do exist for the period
1890-1910, but as shown in section 1.4 the uncertainty of these data, the asso-
ciated metadata and other weather/climate data necessary to calculate the
corrected precipitation is relatively large, and the data can therefore only be
employed as monthly national averages over the entire period. This, of course,
introduces a larger regional uncertainty compared with precipitation data
with full spatial resolution.
The overall sensitivity is believed to be almost the same for the period around
the year 1900 and for the present as the modelling approach for the two peri-
ods is almost identical: however, in some model's parameters have proven
sensitive to changes in climate (e.g. Melsen and Guse, 2021)
The uncertainty of model outputs increases with both decreasing spatial
scale and time scale. Because of the propagation of the error principle, the un-
certainty is larger when assessing single/few or small fourth order catch-
ments rather than large or multiple catchments. The same principle is valid
when assessing a single month rather than a year or multiple years.
146
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0149.png
Some uncertainty originates from the model structure of the NNM, which is
a complex modelling system with interdependent modules. There is a risk
that the ID15-subcatchments (n > 3,000) in a few places are not correctly con-
nected. Also, there is a risk that, in a few places, the routing of the hydrological
model (DK-model) and the ID15 map does not match completely. The differ-
ent modules of the NNM all have an associated uncertainty, and the sequence
of calculations influences the model outputs. The NNM and its submodules,
will not perform equally well in all fourth order catchments as it does not per-
form equally well for all catchments upstream monitoring stations (Højberg
et al., 2015). The individual modules (sources and sinks) will perform differ-
ently across both location and time.
10.1 Quantitative uncertainty and sensitivity - nitrogen
The sensitivity of a few model parameters and input variables was quantified
by running the NNM
1900
with different parameter settings (Table 10.1). These
model runs were not originally designed as a sensitivity analysis, however,
and therefore only relative changes are presented. Potentially, a sensitiv-
ity analysis could have been made for more model parameters, but this was
beyond both the scope and the resources of this project.
Table 10.1.
Sensitivity of some parameters to the year 1900 modelled (NNM
1900
) national nitrogen load to the sea.
Baseline
Natural wetland source area (3 × natural wetland
area)
Natural wetland source area (3 × natural wetland
area)
Double of present, small lake area
120% length of present small watercourses
100% precipitation on Zealand
Scenario
Natural wetland source area (2 × natural wetland
area)
Natural wetland source area (4 × natural wetland
area)
Present, small lake area
100% length of present small watercourses
90% precipitation on Zealand
2
1
-4
-6
Percent change
national N load
6
The natural wetland source area is a relatively sensitive parameter as chang-
ing the extent from 3 to 2 or 4 times the natural wetland area changes the N-
load by 6% (Table 10.1). The parameter value “3
times the natural wetland
area
“was found through a calibration procedure and is believed to be a good
estimate (see section 7.3). The area of small lakes and the length of small wa-
tercourses (<2 m width) are shown to represent a relatively low sensitivity.
As expected, the precipitation is a relatively sensitive model input variable.
The 4% reduction in national nitrogen load originating from a 10% reduction
in precipitation on about 17% of the land area is a relatively large uncer-
tainty/sensitivity. However, the nitrogen loss would have been above aver-
age from Zealand as both the fraction of agricultural area and the percentage
of tile-drained area were larger than the average for the whole country. There-
fore, it would be an overestimation to extrapolate this reduction to the whole
country. Besides, it is unlikely that precipitation has a large bias for the whole
country for the time around year 1900.
The uncertainty of the NNM is tested in Højberg et al. (2015). The nitrogen
load error is reported to deviate 2% from the measured load across 169 mon-
itoring stations covering 57% of the Danish land area for the period between
1990 and 2010. The mass balance error is larger for individual years –31% to
15% (measured – simulated)/measured x 100%) – and at smaller geographical
scale. The version of the NNM is not the same in this study as in Højberg et
147
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0150.png
al. (2015), but the uncertainty is believed to be comparable, the main differ-
ence is the inclusion of the “natural wetlands” module. The uncertainty of a
newer version of the NNM
2020
(Højberg et al., 2021) also including the natural
wetlands module is comparable to the Højberg et al. (2015) NNM
2015
version.
It is not possible to make a direct comparison between the NNM
present
used in
this study and the measured loads for the period 1990 with 2010 because the
root zone N leaching in this study only represents 2011 and not the continuous
period 1990-2010.
10.2 Qualitative uncertainty and sensitivity -nitrogen
Evaluation of qualitative uncertainty and sensitivity is provided in Table 10.2.
The uncertainty of variables varies according to the knowledge and assump-
tions on which both the year 1900 and the present period are based. Cli-
mate inputs are based on observations from around the year 1900 period,
though for precipitation a “delta change” procedure based on 1990-2010 grid-
ded precipitation is used. Year 1900 temperature observations are thought to
be relatively certain as a thermometer is easy to read and quite accurate, while
for example wind data are given in a visual, Beaufort-like, scale and have a
large uncertainty compared with presently observed wind speed. For the year
1900 precipitation, a delta change approach (see section 1.3.8), with the 1990-
2010 precipitation as a baseline, is used, meaning that the relative geograph-
ical precipitation pattern is almost identical for the two periods. This is prob-
ably not the case in reality. Therefore, it is likely that regional/local biases
(both over- and underestimations) are introduced, but the locations and
amounts are unknown. These biases are transferred to the rest of the
NNM
1900
and the model results.
N leaching from the root zone is estimated from land use documented in ag-
ricultural statistics from 1896 and 1900 at the church parish level combined
with measurements from recent experiments with around year 1900-relevant
N concentrations in root zone percolates. As the parish boundaries have re-
mained almost unchanged since the time around year 1900, the parish proba-
bly represents the most conservative land area unit; the present study relies
on 1766 parish units. The estimated N concentrations for root zone percolates
for the land use categories autumn- and spring-sown crops, grass-clover and
root crops are derived from well-monitored field experiments under organic
farming regimes and with an animal stocking density of 0.9 unit/ha. Alt-
hough organic farming excludes the use of mineral fertilisers and chemical
plant protection measures, the present crop varieties are superior compared
with those used around year 1900. Further, a large part of the animal manure
around year 1900 was applied at times associated with a substantial possibil-
ity of N leaching losses. The increased N use efficiency associated with mod-
ern crops grown on well-drained soils and spring application of animal ma-
nure leads to higher yield levels than around year 1900. The higher N use ef-
ficiency recorded in current organic field experiments with year 1900-relevant
stocking density may lead to underestimation of N leaching for the agricul-
ture around year 1900. However, an inferior quality of manure associated
with the low productivity around year 1900 animal husbandry may counter-
balance this discrepancy.
The main uncertainties in relation to point sources for N and P are investi-
gated in section 4.3.5. The most important factors are identified as the initial
load of nutrients from human and animal excrements and the minimum pop-
ulation required (>5,000 in habitants) for inclusion of towns in the assessment.
148
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Of these, changes in the initial load of nutrients in animal and human excre-
ments affected the resulting nutrient emissions most significantly; thus,
changing the baseline value between +/- 25% resulted in a +/- 30% change in
emissions (see chapter 4.)
Even though there is a rather large uncertainty regarding the emissions from
individual towns, the values are, as such, a qualified indication of the magni-
tude of nutrient loads from point sources around year 1900 and a robust na-
tional estimate of emissions.
As the point sources constitute a minor contribution to the total nitrogen load
in the NMN model (2,500 out of 36,000 ton), it is assumed that the uncertainty
regarding the point sources has a small effect (sensitivity) on the estimated
total N load (Table 10.2).
The national hydrological model (DK-model) was used for simulation of cur-
rent and around year 1900 conditions of stream discharge and nitrogen reten-
tion in groundwater. For the current conditions, the model was calibrated
against observations of head and stream discharge (water and nitrogen) data
covering a period of 20-30 years. The hydrogeology properties are not ex-
pected to change, except for the near surface geology in the – -since then –
urbanised areas. The overall effect of these changes is assumed to be insignif-
icant in this context. The two main sources of uncertainty related to change
in groundwater transport are the year 1900 climate input and the year
1900 drainage conditions. Uncertainty related to the year 1900 climate in-
put is partly dealt with using the delta change approach, where the nation-
wide change in precipitation is assumed to be equally distributed across the
entire country.
Today, a large part of the water and nitrogen load to streams
flows through the tile drainage system, where no denitrification is assumed
to take place once the water has entered the tile drains. For the present-day
situation, almost all areas that need drainage are covered by tile drainage,
which makes it relatively easy to simulate in a hydrological model, but this
was not the situation around year 1900. Both density and the drainage type af-
fect the nitrogen load to the streams, and they are both subject to high uncer-
tainty for the year 1900 period. The water balance errors simulated using the
best estimate of drainage density and precipitation for the period just after
year 1900 indicate that the drainage density used in the hydrological simula-
tions might have been overestimated for Zealand plus islands and underesti-
mated for the western part of Jutland. (Table 3.10).
The surface water nitrogen retention modules are also associated with uncer-
tainty. Thus, all modules have a model uncertainty (e.g. as to kg/ha retention)
and an uncertainty originating from the applied maps (e.g. whether the length
of a stream in an ID15 catchment is the same on the map as in reality). On top
of these uncertainties, there is uncertainty concerning the assumptions made
about the changes between the two periods; for example, it is assumed that
the area of the year 1900 small lakes was double that of the present area. The
uncertainty of both the model and the applied maps is, by way of example,
believed to be smaller for larger lakes than for small lakes (Table 10.2) as the
applied model for larger lakes is considered the better model, and the
knowledge of which lakes existed in the year 1900 is also more extensive for
larger lakes.
149
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0152.png
Table 10.2.
Qualitative uncertainty, sensitivity and known bias of parameters involved in calculating the national nitrogen load
around year 1900. Uncertainty = separate parameter uncertainty. Sensitivity = the sensitivity of the national nitrogen load to a
bias of the given parameter. Comp. present is the around year 1900 uncertainty compared with the present period uncer-
tainty. x = small uncertainty/sensitivity; xx = medium uncertainty/sensitivity; xxx = large uncertainty/sensitivity. + = little-, ++ =
medium-, +++ = much larger uncertainty for the around year 1900 period than for the present period, equal = uncertainty is
about the same for the present and the around year 1900 period.
Parameter
Climate
Corrected precipitation
Xx
xxx
+++
Uncertainty primarily connected with regional precipitation
pattern.
Uncertainty Sensitivity
Comp.
present
Comments
Temperature
Solar radiation
Wind speed
Potential evaporation
Nitrogen sources
X
x
x
xx
xx
xxx
xx
x
+
+++
+++
++
+
Equal
+++
Uncertainty almost as today but with some differences in
agricultural management – see text.
Only a few measurements exist.
The main uncertainty on the atmospheric nitrogen deposi-
tion around year 1900 originates from the emission scenar-
ios. The uncertainty on the year 1900 nitrogen deposition is
substantial (Thomas Ellermann, Aarhus University, pers.
comm.).
New method for estimating wind speed gives a 2.7% bias
on corrected precipitation compared with the used values.
Xx
Xxx
Xx
Root zone N leaching (ag- x
riculture)
Root zone N leaching
(other)
Atmospheric deposition
xx
xx
Organic nitrogen
Point sources
xx
xx
xx
x
++
+
The historical loads and urban pathways of nutrients
are highly uncertain at local scale, although robust at na-
tional scale.
Runoff
xx
xxx
+++
Primarily depending on drainage, the corrected precipita-
tion and potential evaporation. Larger uncertainty regionally
than nationally.
Land use
Agricultural area
Nitrogen retention
Groundwater
xx
xxx
+++
Larger regional uncertainty, originating from regional uncer-
tainty on drainage and precipitation, affecting primarily run-
off and secondary possible denitrification in the drainage
system.
Small lakes
Streams
Natural wetlands
Rivers
Larger lakes
xxx
xx
xx
xx
x
x
x
xx
x
xx
++
+
+
+
++
Some uncertainty on the estimates of year 1900 lake area
and Nret rates.
x
xxx
Equal
As today.
Parameters not included in the NNM
150
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0153.png
Meadow irrigation
xx
x
Not included in the NNM. There would be a substantial un-
certainty about potential modelling results, and the sensitiv-
ity would be small as meadow irrigation only occurred on
about 10,000 ha in the year 1900 (Rasmussen, 1964).
There is, though, a small bias, as N retention is likely
to have taken place on meadow-irrigated areas, primarily
river valleys in heath areas in central and western Jutland,
fed by stream water with relatively low nitrogen concentra-
tions.
Modelling approach
x
x
As the NNM
1900
and NNM
present
are near parallel/identical
approaches, there is little relative difference in uncertainty
or sensitivity between the two. As a “delta change” method,
based on a separate estimate of the present nitrogen load
is applied for estimating the year 1900 nitrogen load to the
sea, there is little sensitivity in the modelling approach.
The surface water nitrogen retention calculations are not expected to have se-
rious geographical biases. However, the development, between 1900 and the
present as to, for instance, the area/number of small lakes will differ substan-
tially between, for instance, intensely agricultured areas (expected large
change) and a more or less unchanged heathland (expected low change). As
the same modelling approach and the same map inputs (except for natural
wetlands) are used for both time periods, the uncertainty is believed to be
comparable between the two. For natural wetlands, two separate estimates of
the wetland area are applied, which in itself introduces some uncertainty, but
this uncertainty is believed to be compensated for by avoiding introducing an
assumed change in wetland area.
Generally, the sensitivity is dependent on the size of the nitrogen
source/sink and with the major components of the water balance. Therefore,
the root zone leaching, the ground water nitrogen retention, the precipitation
and the runoff are the most sensitive parameters/parts of the modelling com-
plex.
Comparing the uncertainty of the around year 1900 nitrogen load (as pre-
sented in section 8.3.2) with the 1990-2019 load (Thodsen et al., 2021), the
around year 1900 load has a larger uncertainty. Because of the delta change
approach, the larger uncertainty originates from the larger uncertainty on the
NNM
1900
, originating from the larger uncertainty on model input data, than
for the NNM
present
(Table 10.2). The difference between the year 1900 nitrogen
load (as presented in section 8.3.2) and the 1990-2019 load (Thodsen et al.,
2021) is therefore only attributed to the difference between NNM
1900
and
the NNM
present
and the differences in point sources. The 1990-2019 national ni-
trogen load estimate also has an uncertainty originating from laboratory and
hydrometric field measurements and data interpolation uncertainties (esti-
mated to 4-7%, Søren E. Larsen, pers. comm). Besides this uncertainty, uncer-
tainties from data management errors, model uncertainty for models used in
ungauged areas and some dependency on the chosen load estimation
method add to the overall uncertainty. However, as stated earlier, the differ-
ence between the NNM
1900
and the NNM
present
is independent of these uncer-
tainties as well as the uncertainties existing in equal measures in both the
NNM
1900
and the NNM
present
model.
151
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
10.3 Uncertainties on phosphorus load around year 1900
The phosphorus load calculation around year 1900 (chapter 9) relies on the
assumption that the background TP concentration can be inferred from pre-
sent-day measurements in streams draining catchments with low anthropo-
genic impact. However, there is a risk that present-day measurements in even
relatively undisturbed catchments overestimate the background TP concen-
tration since some level of anthropogenic contamination is unavoidable, at
least in densely populated, highly cultivated Denmark. A few studies from
neighboring countries with similar geology report background concentration
values that are generally lower but in the same order of magnitude as the
background estimates presented in this report (Table 9.3). The background TP
concentration estimate is considered medium uncertain. Since the back-
ground load dominates the total diffuse P load, it is therefore assumed that
the uncertainty on the background concentration has a high effect (sensitivity)
on the total P load estimate (Table 10.3).
Estimation of contributions from other sources than background is hampered
by limited and fragmented knowledge of farming practices around year 1900,
including handling of manure and fencing of livestock. The quantification of
P transfer from agricultural land to surface waters along different transport
pathways (tile drains, erosion) is even today associated with many difficulties
and large uncertainties (Andersen and Heckrath, 2020). These difficulties per-
sist in the attempt to quantify P losses around year 1900.
The overall size and the spatial distribution of the artificially drained area
around year 1900 are uncertain. No data exist on P leaching from drained soils
around year 1900. Alternatively, present-day leaching estimates were used.
For P leaching through the soil matrix pore system, this was justified by the
fact that even today, the P concentrations at tile drain depth are generally very
low and close to the concentration in uncultivated soils. For macropores
where P losses are mobilized from the topsoil, this was justified by the year
1900 practice of applying most of the animal manure in the autumn and win-
ter period, i.e. the period for macropore flow.
A major P input around year 1900 was undoubtedly caused by access of live-
stock to surface waters. Unfortunately, no data are available to quantify this
input, which makes our estimate very uncertain. We attempted to estimate
the order of magnitude by equating the importance of this source to water
erosion of cultivated fields.
For water erosion of agricultural fields, use of a regular modelling approach
was not possible due to lack of historical data. Therefore, through a number
of considerations, the same sediment input to streams as today is assumed.
This was combined with an average year 1900 topsoil P content back-calcu-
lated from the present-day soil P content minus the estimated P accumulation
in agricultural soils since 1900.
Uncovered manure storages, farmyards and outdoor areas with high density
of traffic (cattle tracks etc.) may serve as local hotspots/point sources if con-
nected to surface waters by a hydrological transport pathway. No data are
available to allow a proper quantification of the contribution from such micro
hotspots, making the estimate very uncertain.
Compared with the N load, point sources contribute a relatively large fraction
of the total P load to the sea (471 ton out of 1,200-1,340 ton, see chapter 9);
152
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0155.png
therefore, it is assumed that the uncertainty on the point sources has a me-
dium effect (sensitivity) on the estimated total P load (table 10.3)
Overall, the estimates of the various diffuse P inputs are considered medium
to highly uncertain. Still, the lake sediment studies and the historical stream
measurements reported in section 9.3.2 support our finding that the stream
water TP concentrations around the year 1900 were lower than today but con-
siderably higher than the background concentration.
Table 10.3.
Uncertainty, sensitivity and bias of parameters involved in calculating the na-
tional phosphorus load around year 1900. Uncertainty = separate parameter uncertainty.
Sensitivity = the sensitivity of the national phosphorous load to a bias of the given parame-
ter. x = small uncertainty/sensitivity; xx = medium uncertainty/sensitivity; xxx = large un-
certainty/sensitivity. Known bias positive (+) or negative (-)
Parameter
Climate
Precipitation
Temperature
Solar radiation
Wind speed
Potential evaporation
Point sources
Runoff
Phosphorus concentration/load
Background TP concentration
Soil drainage
Land reclamation
Grazing livestock and surface-applied manure
Soil erosion
Manure storage, farm yards etc.
P retention in lakes
Not included retention by flooding of riparian areas
xx
xx
xx
xxx
x
xxx
xx
xx
xxx
x
x
x
x
x
x
x
-
-
+
Uncertainty Sensitivity Known bias
xx
x
xx
xxx
xx
xx
xxx
x
x
xx
xx
xxx
xxx
xx
10.4 References
Andersen, H.E. and Heckrath, G. (red.), 2020. Fosforkortlægning af dyrk-
ningsjord og vandområder i Danmark. Aarhus Universitet, DCE - Nationalt
Center for Miljø og Energi, 340 s. - Videnskabelig rapport nr. 397.
Højberg, A.L., Windolf, J., Børgesen, C.D., Troldborg, L., Tornbjerg, H., Bli-
cher-Mathiesen, G., Kronvang, B., Thodsen, H. and Erntsen V., 2015. National
kvælstofmodel. Oplandsmodel til belastning og virkemidler. De Nationale
Geologiske Undersøgelser for Danmark og Grønland – GEUS, 111pp.
Larsen, S.E., Senior scientist at Department of Ecoscience, Aarhus University,
Denmark, pers. comm.
Melsen, L.A. and Guse, B., 2021 Climate change impacts model parameter sen-
sitivity – implications for calibration strategy and model diagnostic evalua-
tion. Hydrology and Earth Systems Science 25, 1307-1332.
153
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Thodsen, H., Tornbjerg, H., Bøgestrand, J., Larsen, S.E., Ovesen, N.B., Blicher-
Mathiesen, G., Rolighed, J., Holm, H. and Kjeldgaard, A., 2021. Vandløb 2019
- Kemisk vandkvalitet og stoftransport. NOVANA. Aarhus Universitet, DCE
– Nationalt Center for Miljø og Energi, 74 s. - Videnskabelig rapport nr. 452
http://dce2.au.dk/pub/SR452.pdf
154
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
11 Synthesis of results
Background
The nutrient load from land to the sea around year 1900 has been estimated
considering the various factors affecting the nutrient inputs and transport
based on data from that time, literature, comparative analysis methods and
modelling tools. The main factors to consider include climate, hydrology, land
use, agricultural practices and drainage, urban development and landscape
(e.g. nutrient retention in groundwater, wetlands, lakes and streams). The am-
bition was to use as much data and information as possible from around year
1900 considering the quality and representativeness and use modelling and
GIS tools to provide a geographically distributed estimate of total nitrogen
(TN) and total phosphorus (TP) concentrations and the load from the root
zone to the sea.
Climate
The climate was colder and drier around year 1900 compared with the pre-
sent-day. The estimated average annual precipitation around year 1900 was
about 60 mm, or 7% lower than today. Digitized climate data from around
year 1900 at observation points across the country, including temperature,
wind and rainfall, were used to find monthly values of bias-corrected precip-
itation. The correction approach was evaluated for the period 1917-1950 using
water balance modelling of discharge. At national level, a water balance error
of 3 % indicated reasonable correction estimates, but large regional differences
in error level was found. To obtain spatially distributed corrected precipita-
tion for the period 1890-1910 a delta change climate factor approach was used.
In this approach national monthly correction factors were calculated based on
corrected precipitation for 1890-1910 compared with 1989-2010. These na-
tional factors were then applied to the present daily corrected precipitation
assuming a similar geographical distribution of precipitation around year
1900 as in the present time reference period (1989-2010) to provide a spatially
distributed daily time series of precipitation for the period 1890-1910.
To be able to model nitrate leaching, global radiation and potential evapotran-
spiration must be calculated. By using the measured minimum and maximum
air temperatures for 1890-1950, the global radiation and potential evapotran-
spiration can be calculated and used in the simulation of nitrate leaching in
this period. The modelled global radiation and potential evapotranspiration
around year 1900 are in good agreement with values measured at Foulum
from 1987-2013.
Hydrology
The total discharge based on these precipitation estimates, and drainage den-
sity estimated for the historical time was on average 292 mm/yr for the period
1890-1910 compared with 333 mm/yr for the present period (1990-2010) as
calculated by the hydrological model (DK-model, chapter 3). After applying
a delta change method, the total discharge is recalculated to 335 mm/yr for
the present period and 297 mm/yr for the year 1900 period (chapter 8). This
means that for the total average, annual discharge was about 11% lower
around year 1900 compared with the present time. The change in discharge
155
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
for the two periods largely reflects the changes in precipitation but is ampli-
fied in some areas due to the lower density of drainage in the historical period.
The calculated change in discharge comparing the present to the time around
year 1900 was found to be 0-20% for most of the country, which agrees with
the trend analysis of long discharge time-series presented in Jensen (ed.) 2017.
However, for the western part of Zealand observed data indicate an increase
in discharge between the two periods of approximately 30%, while the simu-
lation resulted in a decrease in discharge of approx. 5%. The methodology
used implies that the total discharge to the sea may resemble the conditions
around year 1900, but it cannot be expected to reproduce the local conditions
at that time.
Nutrient inputs
The nitrogen reaching the sea from land mainly originated from agricultural
activities and dwellings across the country around year 1900. For phosphorus,
the main inputs were natural background loading (mainly from groundwater
and erosion of stream banks) and direct point sources, i.e. coastal cities.
Agriculture
The area in agricultural use increased dramatically during the last half of the
19th century and accounted for close to 3/4 of the area under Danish admin-
istration around year 1900. Crop production differed significantly from cur-
rent agriculture for virtually all growth factors: inferior crop varieties, higher
weed pressure, lack of chemical crop protection and inferior plant nutrient
supply, including the absence of mineral fertiliser. The main sources of nutri-
ents were solid farmyard manure, liquid manure and nitrogen fixation by leg-
ume crops. The number and categories of livestock as well as the farm struc-
ture and management practices around year 1900 also differed from today’s
practices.
Parish level statistics from around year 1900 for the area under current Danish
administration were unified into eight categories (winter and spring crops,
grass, root crops, fallow, nature and forest), and for each category a nitrogen
concentration was ascribed to the root zone percolate. The nitrogen root zone
concentrations were set using data from studies of organic farming as a proxy
for the past-time situation. Literature data were found for the remaining cat-
egories. These values were applied in the nitrogen modelling. The calculation
of the area-weighted average nitrogen concentration for land in agricultural
use (78% of the land area) resulted in a value of 12 mg N/l, while the value in
root zone percolate (inorganic nitrogen) for the entire land area was 9.6 mg
N/l.
The estimation of agricultural sources for phosphorus considered factors such
as soil drainage, land reclamation, grazing animals, soil erosion and manure
storage. These factors were difficult and uncertain to determine, leading to an
estimated range from 56 to 196 ton P annually around year 1900.
Point sources
Sewer systems were increasingly implemented in towns, but still without
wastewater treatment in year 1900. Therefore, towns acted as significant point
sources around year 1900, with 4,261 ton N/yr and 764 ton P/yr emitted in
excrements from humans and animals and industrial wastewater. These fig-
ures indicate that most of the nutrients from point sources were discharged
156
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
directly to receiving waters (55%), but emissions to landfills (20%) and agri-
cultural soil (25%) were significant as well. The total contribution from inland
and direct point sources to water was estimated to 471 ton P, about 65-70% of
the present-day value (704 ton P, average 2014-2018) and 2,531 ton N, about
47% of the present-day TN point sources (5,400 ton N, 2020).
Other sources including background nutrient concentration
Nitrogen inputs from the atmosphere were estimated by multiplying EMEP
simulations for year 2000 by 0.3 (Jensen (ed.), 2017). Organic nitrogen origi-
nating from landscape sources and internal surface water sources was in-
cluded to be able to calculate total nitrogen concentrations. Estimates based
on literature studies assume that the organic nitrogen concentration around
year 1900 was about 20% below the current levels. Furthermore, it is assumed
that the current geographical distribution of organic N is valid for the time
around year 1900.
A literature review and measurements from largely undisturbed streams al-
lowed estimation of background TP stream concentration, and an area-
weighted TP median value at 0.052 mg/l was estimated.
Nutrient transport to the sea
Nitrogen percolates through the soil and reaches the groundwater where re-
duction (retention/removal) of nitrogen under oxygen-free conditions takes
place before the remaining nitrogen ends up in surface waters (wetlands,
lakes, streams). The National Nitrogen Model simulates transport and reten-
tion in groundwater based on water discharge and the nitrogen percolate in-
put. The surface water component calculates the nitrogen retention in wet-
lands, streams and lakes, while also considering point source inputs, atmos-
pheric inputs and the contribution of organic nitrogen. Landscape changes
between the time around year 1900 and the present time were handled by
modifying the current landscape maps based on various information sources
on the past landscape related to rivers and lakes. For wetlands, different maps
were used.
The phosphorus load calculation was based on total phosphorus considera-
tions and used an approach of a ”background” or ”nature” concentration
level, on top of which the relevant additional agricultural and point sources
were added and retention in lakes was subtracted. The transport and routing
of phosphorus through the catchments were simulated using the same water
discharge as in the nitrogen modelling.
The nitrogen retention in inland surface water was shown to be higher in the
present period (28,000 ton N) than in the 1900 period (26,000 ton N) due to a
larger present-day nitrogen load. However, the relative nitrogen retention
was higher around the year 1900 as 43% of the load was removed compared
with 33% for the present period.
The total nitrogen load is modelled to be approximately 36,000 ton N/yr
around year 1900, which is approximately 42% less than for the present period
(59,000 ton N/yr). The nitrogen concentration is modelled to be around 2.8
mg N/l around year 1900 compared with 4.1 mg N/l in the present period.
The national nitrogen model yields regional results, which are utilised for es-
timating regional year 1900 nitrogen and freshwater loads.
157
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
The average stream water phosphorus concentration around year 1900 was
estimated to 0.062-0.075 mg P/l, equivalent to 60-70% of the present-day
stream water TP concentration (0.1 mg P/l). The TP values were calculated
for nine geographical regions across the country. The total TP loading to the
sea, including background concentration, other diffuse sources, inland and
direct point sources and subtracted phosphorus retention in lakes, was esti-
mated to 1,200-1,340 ton P, 60-65% of (or 35-40% less than) the present-day
phosphorus loading (2,021 ton, average 2014-2018).
Uncertainties
Working with a period 120 years ago naturally makes most aspects of calcu-
lating the national nitrogen and phosphorus load more uncertain than when
calculating it for the present period. An in-depth analysis of uncertainties of
data layers, variables, model and model assumptions was not a part of the
present study however, some considerations regarding uncertainty and sen-
sitivity (the effect of a given parameter on the result) have been made.
Most of the parameters used to estimate the nitrogen loads around the year
1900 are considered to have “medium” uncertainty (on a three-step scale from
low to high). The uncertainty of the nitrogen loads is influenced by a variety
of factors, the most important being the uncertainty of the estimates of pre-
cipitation, run-off, root zone concentration of nitrogen and retention in sur-
face and groundwater.
Overall, the model concept used to calculate the year 1900 nitrogen load is
considered relatively robust and the overall uncertainty at national scale ac-
ceptable. However, the uncertainty increases with decreasing geographical-
and timescales.
Most of the parameters used to estimate phosphorus loads are considered
“medium” to “highly” uncertain. The uncertainty of phosphorous loads is es-
pecially influenced by the uncertainties of the estimates of precipitation, run
off, P input from point sources and the background TP concentration. Despite
the many uncertainties the results of this study are believed to be the best
possible estimate of the year 1900 phosphorous loads. Furthermore, the re-
sults are supported by historical lake measurements that also find the histor-
ical TP-concentrations to be lower than today but considerably higher than
the background concentration, though.
158
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
12 Perspectives of results
Authors: Hans Thodsen
1
, Jørgen Eriksen
2
Quality assurance: Dennis Trolle
1
1
DCE,
Aarhus University, Department of Ecoscience
2
DCA, Aarhus University, Department of Agroecology
12.1 Comparisons to other root zone nitrate leaching studies
Mean annual nitrate leaching from agriculture is estimated to 36 kg N/ha in
year 1900, based on measurements in modern organically farmed crop rota-
tions similar to those around 1900, with nitrate concentrations in leachate of
12 mg N/l and the year 1900 percolation/runoff of 297 mm/yr. For the pre-
sent period (2007) the comparable value is about 58 kg/ha (Børgesen et al.
(eds.), 2009). This is comparable to the concentrations determined in the Ag-
ricultural Catchment Monitoring Program (LOOP) during 2004/05-2015/16
of 14 mg N/l (Blicher-Mathiesen et al., 2019) and lower than the 1990/91-
93/94 mean number of 28 mg/l. As nitrate leaching is closely connected to
drainage volume (Blicher-Mathiesen et al., 2019), the same concentrations in
year 1900 would lead to lower leaching losses than today. For south and cen-
tral Sweden, similar results were found by Hoffmann et al. (2000) when com-
paring nitrate leaching in the late part of the 19th century to the 1980s. They
found that specific leaching rates were approximately the same as those today
using the SOIL/SOILN model to calculate N leaching.
Opposite to the two studies for Denmark and Sweden, a study for the UK
found that modelled annual N loss by leaching, runoff and soil erosion in-
creased from 15 to 52 kg N/ha in arable farming in the periods 1800-1950 to
1970-2010 and from 18 to 36 kg N/ha in grassland for the same periods (Mu-
hammed et al., 2018). Arable and grassland leaching were modelled and cali-
brated separately on data from the Broadbalk and Park Grass long-term ex-
periments for arable and permanent, semi-natural grassland, respectively,
during 1800-1950. This setup was chosen as only after 1950 semi-natural grass-
land was converted to improved grass and arable land (Muhammed et al.,
2018). This situation is very much different from the conditions in Denmark,
where in year 1900, 69% of the total grassland area was in rotation with arable
crops (Danmarks Statistik, 1968). As grassland cultivation is a main source of
N leaching when combined with bare fallow and winter cereals (see chapter
5), these differences in crop rotations between UK and Denmark/southern
Sweden probably explain the differences in historical nitrate leaching.
The N surplus as calculated from simple input/output to Danish agriculture
was estimated to 28 kg N/ha in year 1900 in a previous study (Kyllingsbæk,
2008). This cannot be directly linked to losses for several reasons (Christensen
et al., 2017), one being that it does not include mineralisation of the soil or-
ganic N pool, which was estimated to 39-67 kg/ha in the UK study for 1800-
1950 (Muhammed et al., 2018). Furthermore, in Kyllingsbæk (2008) biological
N
2
fixation by legumes in grasslands was estimated to 50 kg N/ha, which
seems to considerably underestimate the N input from this source. As average
of two soil types during 1907-1922, the yields of grass-clover leys in animal-
manured plots of the Askov long-term field experiments were 4,765 kg DM
159
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
with 42% clover (Iversen and Dorph-Petersen, 1951). Using an empirical
model (Høgh-Jensen et al., 2004) for grass-clover leys, the N2-fixation based
on the Askov yields and legume contents can be estimated to 117 and 183 kg
N/ha for clay and sandy soil, respectively. This gives an average input of 150
kg N/ha, which is three times higher than the estimate from the balance by
Kyllingsbæk (2008).
12.2 Comparisons to other river load studies
This section provides a few comparisons to other studies attempting to esti-
mate the year 1900 riverine nitrogen concentrations or loads.
Danish studies
Jensen (ed.) (2017) conducted a study, to which some of the authors of this
report contributed, estimating the Danish riverine concentrations in year 1900
using several approaches. In contrast to the present study, Jensen (ed.) (2017)
did not attempt to estimate any loads (national, regional or local) or to calcu-
late monthly N concentrations/loads but estimated a national N concentra-
tion range of 1.2 to 2.2 mg N/l (rounded off to 1-2 mg N/l). This range is based
on a total retention range of 76-87% for 1900 and an average leaching concen-
tration of 9.25 mg N/l. In this study, the total retention rate was estimated to
78% (Table 7.10) and thus the higher end of the concentration range at 2.0 mg
N/l. However, the estimate of Jensen (ed.) (2017) is only based on nitrate
leaching through soils and retention in groundwater and surface water and
does not consider the organic nitrogen fraction. Including the organic fraction
would increase the total nitrogen concentration by approximately 0.5 mg N/l,
as calculated for 1900 in this study (chapter 6). This would increase the total
N concentration range in Jensen (ed.) (2017) to 1.7-2.7 mg N/l, ending up 2.5
mg N/l using the 78% total retention rate found in this study. In Jensen (ed.)
(2017), point sources were considered to be small and therefore not quantified.
However, in this study point sources have been quantified to about 2,500 ton
N/yr, which corresponds to about 0.20 mg N/l (with no retention on inland
point sources). Thus, if including point sources, the Jensen (ed.) (2017) esti-
mate range rises to 1.9-2.9 mg N/l. The total load concentration of 2.8 mg N/l
modelled in this study thus falls within the range of the Jensen (ed.) (2017)
range corrected for the organic nitrogen fraction and point sources.
Jensen ed. (2017) (see Figure 5.5) uses a relation where the annual nitrogen
surplus in Danish agriculture between 1990 and 2014 is used for calculating
the normalised diffuse nitrogen concentration for 1900 (note that the equation
as presented in Jensen (ed.) (2017) is not correct, a corrected version is pre-
sented below). Jensen (ed.) (2017) gets a diffuse nitrogen concentration be-
tween 2.0 and 2.2 mg N/L using a nitrogen surplus around year 1900 (agro-
hydrological year (1 April to 31 March) 1900-01 – 1903-04) ranging between
69,000 ton and 87,000 ton (Kyllingsbæk, 2008). However, the Kyllingsbæk
(2008) estimate of the year 1900 nitrogen surplus is underestimated due to the
use of too low N fixation rates (see section 12.1). A new nitrogen balance was
calculated based on Kyllingsbæk (2008) (graphs read with graphreader.com)
and multiplying the fixation amount by three. In the new estimate based on
the exact agricultural year 1900-1901, the nitrogen surplus is about 176,000 ton
N. Using the range of the period 1900-01 to 1903-04, as in Jensen (ed.) (2017),
the nitrogen surplus is about 216,000 ton N. Based on an updated version of
Figure 5.5 in Jensen (ed.) (2017), the above-mentioned relation (updated equa-
tion, diffuse N concentration = 0.0106 x nitrogen surplus + 1.486; as in Jensen
160
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0163.png
(ed.) (2017) 1995-96 is omitted from the dataset) yields a year 1900-1901 diffuse
nitrogen concentration of 3.3 mg N/l. Again, the year 1900 point source con-
centration of about 0.20 mg N/l needs to be added to obtain the total load
concentration of about 3.5 mg N/l (Table 12.1). Note that the national farm
nitrogen surplus estimate for the present has changed a little due to a change
in method (Blicher-Mathiesen et al., 2015). Also note that this approach is an
extrapolation of an empirical relation, which is a method with considerable
uncertainty, as mentioned in Jensen (ed.) (2017). The results in Table 12.1 are
comparable to the 2.8 mg N/l concentration and 36,000 ton N/yr load to the
sea calculated in this study. The results presented here are in the middle range
of the estimated loads shown in Table 12.1.
Table 12.1
Agricultural field nitrogen surplus year 1900 and corresponding diffuse and total river nitrogen concentrations and
total nitrogen load to the sea. *Total concentration is diffuse + 0.2 mg N/l from point sources.
1
is the low limit, and
2
is the high
limit of the agricultural nitrogen surplus range given in Kyllingsbæk (2008) for the period 1900-1901 to 1903-1904.
3
is the low
limit, and
4
is the high limit of the recalculated agricultural nitrogen surplus given in Kyllingsbæk (2008) but with a corrected nitro-
gen fixation component for the period 1900-1901 to 1903-1904. See text above for more information.
Agricultural nitro-
gen surplus
Ton N/yr
0
69,000
1
87,000
2
Diffuse nitrogen river con-
centration
mg N/l
1.5
2.2
2.4
3.3
3.8
Total nitrogen river
concentration*
mg N/l
1.7
2.4
2.6
3.5
4.0
Year 1900 runoff
mm
297
297
297
297
297
Total nitrogen load to
sea
ton N/yr
22,000
31,000
33,000
45,000
51,000
176,000
3
216,000
4
European studies
Savchuk et al. (2008) ran the SANBALTS (marine model) to simulate annually
averaged coupled nitrogen and phosphorus cycles in the major basins of the
Baltic Sea around year 1900 (“a century ago”). As the riverine nutrient forcing
the assessment by Schernewski and Neumann (2005) is used, they estimate
background inorganic (DIN) concentrations from the 15 largest rivers around
the Baltic Sea. They report year 1900 (“about one century ago”) DIN concen-
trations between 0.06 mg/l in northern Sweden and Finland and 1 mg/l in
the Vistula and Oder rivers in Poland/Germany; no TN concentrations are
given. However, the TN concentration will be >1 mg/l in Oder and Vistula as
the DIN concentration is given as 1 mg N/l, and there is always an organic
fraction. None of the 15 rivers are Danish. The German/Polish catchments are
probably the ones that are most comparable to the Danish catchments because
of the similarities in climate and agricultural use (more comparable than most
other parts of the Baltic Sea catchment), even though the Oder and Vistula
catchments have mountainous upland areas and more forest than the Danish
catchments around year 1900. Gadegast et al. (2012) reports 24% forest in the
Oder catchment, for Denmark the corresponding figure is about 7% (sec-
tion 5.2). Therefore, the concentration in the Oder is expected to be lower than
the 2.8 mg N/l for Denmark, the difference is relatively large, though.
Savchuk et al. (2008) reports around year 1900 nitrogen loads for the “Danish
straits” of 27% and for “the Kattegat” of 31% of contemporary loads (1997-
2003). These results are as mentioned above based on Schernewski and Neu-
mann (2005). However, to some degree, the results contrast the recognition of
Savchuck et al. (2008) of the Hoffmann et al. (2000) paper that states that the
N leaching from agriculture in southern Sweden was about the same around
year 1900 (and in the mid-19th century) as in the late 1990s.
161
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0164.png
From the water discharge data presented in Schernewski and Neumann
(2005), a runoff of 177 mm/yr can be calculated for the Oder River (1970-
1990). This is about half of the 2008-2019 runoff from Denmark, 335 mm/yr.
In itself, the difference in runoff probably yields a difference in nitrogen load
as the amount of runoff and the soil water percolation will be closely linked,
as will the percolation and the soil nitrogen leaching. The dependence of the
annual nitrogen load on the annual runoff is clear in the Danish assessment
(Thodsen et al., 2021).
Gustafsson et al. (2012) calculated monthly mean river flows for the period
1850-2009 to the Baltic Sea. The N loads around year 1900 are based on
Savchuk et al. (2008) but with reevaluated point source contributions from
larger coastal cities, however, how this affects the nitrogen load is not dis-
cussed.
Gadegast et al. (2012) modelled the change in nitrogen emissions into the
Oder River system (136,528 km
2
) between 1875 and 1944 using the MONERIS
model (Venohr et al., 2011). During this period, the Oder watershed was part
of Germany but is now mostly part of Poland. The climate is comparable to
Denmark, though drier and with a more continental temperature regime. The
geology, especially in the northern part of the Oder basin, is comparable to
that of Denmark, while the southern parts contain some mountainous land-
scapes, which are not present in Denmark. Gadegast et al. (2012) found that
the load about doubles from 25,300 ton N/yr in 1880 to 46,600 ton N/yr in
1940. It is modelled that 57% of the load in 1880 is derived from point
sources/urban sources, although only about 10% of the population were con-
nected to sewer systems. This is markedly higher than estimated for Denmark,
where the point source load is calculated to be about 2% (not including direct
point sources to the sea) (Table 7.10). Between 1880 and 1910, the land use of
the Oder catchment is reported to be relatively stable with around 58% arable
land, 9% grassland, 26% forest and 2% urban areas. In comparison, the Danish
arable land constitutes about 67% and about 10% permanent grass and 7%
forest in 1896/1901. Approximately 13% of the arable land in the Oder catcht-
ment would have been tile drained around year 1900, assuming a 1.9% annual
increase between the reported values of 9% in 1880 and 28% in 1940. This is
lower than the values for tile draining in Denmark (26%) around year 1900
(section 5.2). Gadegast et al. (2012) estimates a nitrogen surplus around -5
kg/ha (agricultural area) for year 1900, calculated using the OECD method,
but they do not specify whether the “farm gate balance” or “soil surface bal-
ance” is used or which components are included in the calculation (OECD,
1997). It seems likely, though, that the “farm gate balance” approach is used
since negative N surplus values are reported for 26 years in a row, between
1893 and 1918. The leaching of nitrogen is around 40 kg N/ha (section 12.1).
The high Danish leaching value is largely due to high N fixation rates in clover
grass and other N fixation crops utilised in the cropping system of the time.
N fixating crops will have increased the fixation rates with low fertilizer ap-
plication rates compared with a situation with high fertilisation rates. Nitro-
gen fixation is not mentioned in Gadegast et al. (2012); therefore, a comparison
cannot be made for this important parameter. Some TN measurements from
between 1877 and 1881 are presented for the Oder River, around the city of
Wroclaw in southwest Poland, showing 1.5 mg N/l (at two locations) and a
modelled concentration of 2.3 mg-N/l at the same location around 1880. The
modelled outflow concentration of the Oder River for year 1900 is modelled
to 1.1 mg N/l (Gadegast et al., 2012) (Markus Venohr, pers. comm.). For Den-
mark, the corresponding value is 2.8 mg N/L.
162
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0165.png
Many of the European studies report nitrogen concentrations around the year
1900 that are considerably lower than in the present study. The reasons for
this are probably differences in landscape, land use, farming practices and
runoff between the investigated areas and Denmark. It probably also reflects
the degree to which agricultural practices and nutrient dynamics are included
in the study. In our study, a year 1900 root zone leaching is calculated, and
nitrogen fixation, the main nitrogen source in Danish agriculture in year 1900,
is considered, which is not the case in most other studies. In Denmark, the
present measurement of “undisturbed” streams (streams draining catchments
with <10% agricultural land use (Bøgestrand et al., 2014) display nitrogen con-
centrations ranging between 0.61 and 1.48 mg N/l. If these concentrations
were representative for year 1900 as well, the national nitrogen load could be
calculated by multiplying these concentrations by runoff for year 1900 (the
297 mm/yr at the fourth order coastal catchments scale (Table 8.2 and Figure
8.3). This would give a nitrogen load of about 15,000 ton N/yr, with a flow-
weighted concentration of 1.1 mg N/l, which is considerably lower than the
36,000 ton N/yr and 2.8 mg N/l found in this study. This emphasises that the
nutrient concentrations around year 1900 in Denmark were influenced by hu-
man activity. To a certain degree, this study builds on Jensen (ed.) (2017),
which gives a concentration interval of 1.2 and 2.2 mg N/l (not including
point sources and organic nitrogen) and thus also higher estimated concen-
trations than background concentrations.
A key strength of the present study compared with most of the studies men-
tioned above is the wide range of the authors’ scientific backgrounds, ranging
from climatologists, agronomists specialised in nitrogen leaching, groundwa-
ter modellers, surface water specialists, specialists in landscape and river
phosphorus dynamics and urban water/point source specialists. Such wide
diversity of scientific backgrounds is fundamental for conducting a study like
this. Therefore, we perceive our study to be more thorough than the other
studies mentioned, thus giving a more accurate picture of the year 1900 con-
ditions.
12.3 References
B
licher-Mathiesen, G., Holm, H., Houlborg, T., Rolighed, J., Estrup Andersen,
H., Vodder Carstensen, M., Grewy Jensen, P., Wienke, J., Hansen, B. and
Thorling, L., 2019. Landovervågningsoplande 2017. Scientific Report from
DCE – Danish Centre for Environment and Energy No. 305.
Blicher-Mathiesen, G., Rasmussen, A., Andersen, H.E., Timmermann, A., Jen-
sen, P.G., Hansen, B. and Thorling, L., 2015. Landovervågningsoplande 2013.
NOVANA. Aarhus Universitet, DCE – Nationalt Center for Miljø og Energi,
154 s. - Videnskabelig rapport fra DCE - Nationalt Center for Miljø og Energi
nr. 120
Bøgestrand, J., Kronvang, B., Windolf, J. and Kjeldgaard, A., 2014. Baggrunds-
belastning med total N og nitrat-N. Notat fra Aarhus Universitet, DCE - Na-
tionalt Center for Miljø og Energi. 11pp.
https://dce.au.dk/fileadmin/dce.au.dk/Udgivelser/Notater_2014/Bag-
rundsbelastning_med_total_N_opdatering.pdf
163
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0166.png
Børgesen, C.D., Waagepetersen, J., Iversen, T.M., Grant, R., Jacobsen, B. and
Elmholt, S. 2009. Midtvejsevaluering af vandmiljøplan III – hoved- og Bag-
grundsnotater. DJF rapport, Markbrug 142. August 2009. 238pp.
https://pure.au.dk/ws/files/2841678/djfma142.pdf.pdf
Christensen B.T., Olesen J.E. and Eriksen, J., 2017. Year 1900: Agriculture and
leaching of nitrogen from the root zone. In “Estimation of Nitrogen Concen-
trations from root zone to marine areas around the year 1900” Ed. Jensen, PN.
Scientific Report from DCE – Danish Centre for Environment and Energy No.
241, 37-46.
Danmarks Statistik, 1968. Landbrugsstatistik 1900-1965. Bind 1, Landbrugs-
areal og høstudbytte samt gødningsforbrug. Danmarks Statistik, København.
Gadegast, M., Hirt, U., Opitz, D. and Venohr, M., 2012. Modelling changes in
nitrogen emissions into the Oder River System 1875-1944. Regional Environ-
mental Change 12, 571-580. doi:10.1007/s10113-011-0270-5
Gustafsson, B. G., F. Schenk, T. Blenckner, K. Eilola, H. E. M. Meier, B. Müller-
Karulis, T. Neumann, T. Ruoho-Airola, O. P. Savchuk and E. Zorita, 2012. Re-
constructing the Development of Baltic Sea Eutrophication 1850–2006. Ambio
41, 534-548. doi: 10.1007/s13280-012-0318-x.
Hoffmann M., Johnsson H., Gustafson A. and Grimvall A., 2000. Leaching of
nitrogen in Swedish Agriculture – a historical perspective. Agriculture, Eco-
systems and Environment 80, 277-290.
Høgh-Jensen. H., Loges. R., Jørgensen. F.V., Vinther F.P. and Jensen E.S., 2004.
An empirical model for quantification of symbiotic nitrogen fixation in grass-
clover mixtures. Agricultural Systems 82, 181-194.
Iversen, K. and Dorph-Petersen, K. 1951. Forsøg med staldgødning og kunst-
gødning ved Askov 1894-1948. 440. Beretning fra Statens Forsøgsvirksomhed
i Plantekultur.
Jensen, P.N. (Ed.), 2017. Estimation of Nitrogen Concentrations from root
zone to marine areas around the year 1900. Aarhus University, DCE – Danish
Centre for Environment and Energy, 126 pp. Scientific Report from DCE –
Danish Centre for Environment and Energy No. 241.
http://dce2.au.dk/pub/SR241.pdf
Kyllingsbæk, A., 2008. Landbrugets husholdning med næringsstoffer 1900-
2005. DJF markbrug nr. 18.
Muhammed S.E., Coleman K., Wu L., Bell V.A., Davies J.A.C., Quinton J.N.,
Carnell E.J., Tomlinson S.J., Dore A.J., Dragosits U., Naden P.S., Glendining
M.J., Tipping E. and Whitmore A.P., 2018. Impact of two centuries of intensive
agriculture on soil carbon, nitrogen and phosphorus cycling in the UK. Sci-
ence of the Total Environment 634, 1486-1504.
OECD, 1997. Environmental indicators for agriculture, volume 1 concept and
ramework. OECD, Paris. 45pp. Environmental Indicators for Agriculture
– Vol.I Concepts and framework (oecd.org)
164
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0167.png
F., Hille, S., Humborg, C. and Pollehne, F., 2008. The Baltic Sea a century ago
- a reconstruction from model simulations, verified by observations. Journal
of Marine Systems 74, 485-494. doi:10.1016/j.jmarsys.2008.03.008.
Schernewski, G. and Neumann, T., 2005. The trophic state of the Baltic Sea a
century ago: A model simulation study. Journal of Marine Systems 53, 109-
124. doi:10.1016/j.jmarsys.2004.03.007.
Thodsen, H., Tornbjerg, H., Bøgestrand, J., Larsen, S.E., Ovesen, N.B., Blicher-
Mathiesen, G., Rolighed, J., Holm, H. and Kjeldgaard, A. 2021. Vandløb 2019
- Kemisk vandkvalitet og stoftransport. NOVANA. Aarhus Universitet, DCE
– Nationalt Center for Miljø og Energi, 74 s. - Videnskabelig rapport nr. 452
http://dce2.au.dk/pub/SR452.pdf
Venohr, M. et al., 2011. Modelling of Nutrient Emissions in River Systems -
MONERIS - Methods and Background. International Review of Hydrobiology
96(5), 435-483. doi:10.1002/iroh.201111331
165
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0168.png
Appendix to chapter 4 (4.1 and 4.2).
Appendix 4.1
Further documentation is found here: https://www.mdpi.com/2073-
4441/12/3/789 where a spreadsheet appendix is available as Supplementary
Material. The spreadsheet contains background calculations and resulting
emissions to different environmental compartments from each town and
slaughterhouse.
Appendix 4.2
Flow charts illustrating the alternative disposal routes tested in the uncer-
tainty analysis (Figures A2-A4)
166
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0169.png
Figure A1.
Alternative flow chart assuming 50% less pipes and more buckets in coastal and inland towns (Table 4.1). Changes
relative to the assumed baseline scenario (Figure 4.2) are highlighted in red.
Figure A2.
Alternative flow chart assuming 50% more pipes and less buckets in coastal and inland towns (Table 4.1). Changes
relative to the assumed baseline scenario (Figure 4.2) are highlighted in red.
167
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0170.png
Figure A3.
Alternative flow chart assuming more buckets sold to Copenhagen farmers. Changes relative to the assumed base-
line scenario (Figure 4.2) are highlighted in red.
168
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0171.png
Appendix to chapter 5 (5.1)
Establishing land use and agriculture map at parish
level in 1900
Author: Birger Faurholt Pedersen
1
, Mette Balslev Greve
1
and Eva Overby Bach
1
1
DCA, Aarhus University, Department of Agroecology
From around year 1830 and every 5-10 years onwards, the Danish authorities
frequently gathered agricultural, arable land and land use statistics. In the pe-
riod 1864 to 1920, the southernmost part of Jutland (approx. 1/11 of the Dan-
ish territory today) was under German reign. The German authorities con-
ducted similar statistical work on the land use and arable land under their
reign.
The most detailed agricultural and land use statistics on the Danish territory
from around 1900 are available at parish (sogn/Amtsbezirk/Flecken) level
and aggregated at shire (herred) level as well. The data sources closest to year
1900 at parish level come from both the Danish Statistics Agency (year 1896
1
)
and the German Statistical Agency (year 1900
2
).
Figure A.1.
Extract from the Danish (left) and German (right) statistics.
Both statistical sources use parish level data. Most (around 90%) of the parish
data follow the single parish boundaries and contain only one parish, where
the areas recorded match the area of the parish. However, the statistical data
also contain some merged parishes and data on specific land use on, for in-
stance, contained and drained lakes, bogs between parishes and other areas
1
Statens Statistiske Bureau & Danmarks Statistik, 1898. Arealets Benyttelse i Danmark den 15.
juli 1896 (Statistisk Tabelværk Rk. 5 Litra C Nr. 1). København, Bianco Lunos Hof-Trykkeri (F.
Dreyer).
2
Engelbrecht, Th. H., 1907. Bodenanbau und Viehstand in Schleswig-Holstein nach Ergebnissen
der amtlichen Statistik. II, Anhang, Tabelle 1.
169
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0172.png
(Store Vildmose etc.) belonging to more parishes. For these 10% of the data,
we needed to adjust the numbers to keep the details and completeness.
A polygon layer of the parishes was available from the Ministry of Higher
Education and Science
3
. The parish boundaries are stored in a database with
all applicable parish boundaries from around 1800 and up to the current par-
ish boundary map. For the purpose of the project, we extracted the applicable
parish boundaries on 1 July 1896 to fit the statistical dataset.
The land register (cadastral map) forms the basis for the parish map. In some
cases, there is a delay in updating the land register and the cadastral map.
Furthermore, there may be a delay in updating the parish maps when matric-
ulating drained sea areas and internal lake areas that transition into active
agricultural land. The situation is similar along coastlines where land areas
either emerge or disappear. Moreover, in the period 1864-1920, the parish map
for the German-reigned areas was not updated.
Table A.1
Categories and areas from the Danish and German data.
DK category
DA3
DA4
DA5
DA6
DK_category_text
Barley
Oat
Mixed cereals (mature)
Buckwheat
Ha (S26)
301.057
479.084
129.831
14.530
Merged_category
S03
S04
S05
S06
Merged_category_text
Barley
Oats
Mixed cereals
Buckwheat
DK-MODEL
Ha (DK-model)
TY_category
TY6
TY7
TY2
TY4
TY8
TY9
TY10
TY11
TY12
TY13
TY14
TY15
TY16
TY32
TY25
TY39
TY42
TY27
TY28
TY45
Winter cereals
357.426
TY1
TY5
TY3
TYI
TY17
TY18
TY20
TY19
TY21
TY22
TYK
TY29
TY30
TY31
TY33
TY23
TY24
TY35
TY36
TY37
TY38
TY40
TY44
TYG
TYH
Fallow
267.843
TY43
no code
no code
TYJ
TYL
Ty_category_text
Spring barley
Oats
Spring wheat
Spring rye
Mixed cereal (winter)
Mixed cereal (summer)
Buckwheat
Peas
Fava bean
Vetch
Mixed cereals
Mixed pulses
Other types
Spurrey
Winter rape and radish
Mustard
Mustard
Flax
Other types
Gardens and fruit plantations
Winter wheat
Winter barley
Winter rye
Forests
Potatoes
Sugar beets
Carrots
Fodder beets
Fodder radish
Swedes
Uncultivated land
Clover (for forage)
Lucerne
Serradel
Seed production (Grass-clover,grass)
Field herbs and caddish
Other types
Vetch
Lupines (for forage)
Mixed legumes
Mixed vegetables (for forage)
Lupines
Cultivated grass
Meadows
Pastures
Fallow
no code
no code
Buildings and yards
Roads and lakes, ponds, streams
DA7
Pulses
11.600
S07
Pulses
Spring cereals
982.110
DA8
DA9
DA16
DA17
DA27
DA1
DA2
DA26
DA28
DA29
DA11
DA12
DA13
DA14
DA25
DA30
DA31
DA32
DA10
Spurrey (mature)
Caraway and oil-seed rape
Flax, hemp and tobacco
Garden crops
Gardens and plant nurseries
Wheat
Cereal rye
Hedgerows and shelters
Forest areas (planted)
Forest areas (unplanted)
Potatoes
Sugarbeets and Chicory
Carrots
Fodder beets
Moor and Peatland
Heathland
Shifting sands and sand dunes
Stone fields, swamps, foreshores etc
Seed production (clover, grass, beets and lupines)
7.641
525
293
37.549
41.428
315.998
295.580
56.743
13.583
6.535
76.910
S08
S09
S16
S17
S01
S02
S22
S11
S12
S13
S14
Spurrey
Caraway and rape
Flax, hemp and tobacco
Garden crops
Wheat
Cereal rye
Fores
Potatoes
Sugar beets
Carrots
Fodder beets
Forest
295.580
Roots
153.771
587.555
S21
Moors, peat- and
Nature
587.555
25.028
S10
Seed production
DA15
Green forage (mixed cereals, spurrey and lucerne)
51.277
S15
Green forage
Grass
1.537.640
DA21
DA22
DA23
DA24
DA18
DA19
DA20
DA33
DA34
no code
no code
Cultivated grass for hay
Cultivated grass for grazing
Meadows
Fens and commons
Black fallow land (vegetation free)
Black fallow land (green manure before ploughing)
Semi-black fallow land (early summer-crop)
Roads, building sites and storage areas
Lakes, ponds, streams (outside sea territory)
no code
no code
Total area on the ID15 map from 2015
1.053.714
407.621
267.843
88.336
13.086
2.967
17.441
4.303.755
S19
S20
S18
S23
S24
S25
S26
Cultivated grass
Meadows, fens and commons
Fallow
Roads and building sites
Lakes, ponds and stremas
Buildings and yards
Roads and water areas
Other
121830
4.303.755
3
Available at the website:
http://digdag.dk.
Funded and created by the Ministry of Higher
Education and Science, Denmark.
https://ufm.dk/en
170
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0173.png
Compared to the current parish map of Denmark and the current water catch-
ment map (ID15)
4
, there are some minor differences in the parish boundaries
along the coastlines and the inland lakes; these are included in the water
catchment map but are not part of the parish maps and statistics.
The statistical data from the Danish Statistics Agency include 34 different crop
types/land use types (DK categories in Table A.1), while the German statisti-
cal data include 51 categories (TY categories in Table A.1). Some of the 34/51
categories are merged, resulting in 26 (new) common groups (merged catego-
ries in Table A.1) that accommodate the differences in the provided data. Fur-
thermore, the number of categories was minimised for further calculations,
implying that the 26 categories were divided into eight major groups (DK-
model), and the data analyses conducted based on the DK-model use these
eight groups. Table A.1 also shows the total area of the 26 common categories
and the sum of the eight major categories.
Accordingly, the use of parish maps gives many details on the farming and
the actual land use in 1900, but a certain tweaking was needed to show the
details on a map that could be applied for comparisons with current maps and
statistics. Part of this tweaking process was to match and align the statistical
data with the parish maps and areas. Some of the data and polygons were
merged to give a more correct picture of the land use. The final map contains
1,702 parishes from the Danish data and 64 from the German data, 1,766 in
total.
Further details on how the statistical data and parish maps were made are
given in the next chapter.
Ad 1 Adapting the statistical data – description of methodology
Statistical data on agricultural and other land use from different data sources
created the basis for the survey. To sum up the data for the 316 coastal land
areas, the data needed alignment at the most detailed level.
The methodology used for establishing the background map for land use in
1900 is roughly described in Figure A.2., including also the distribution of
land use on the ID15 and the water catchment map. The available data are
also described in more detail. The data material included:
1. Digital map of parishes in 1900 - download from digdag.dk.
2. Statistics on agriculture and land use for the Danish part of Denmark as of
1 July 1896.
3. Statistics on agriculture and land use for the German part of Denmark as
of 1 July 1900.
4. Digital ID15 water catchment map.
4
https://www.geus.dk/media/13243/national-kvaelstofmodel-oplandsmodel-til-belastning-
og-virkemidler-sep2015.pdf
171
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0174.png
Figure A.2.
Methodology.
Re 1) - Digital map of parishes in 1900 - download from digdag.dk
From the digital parish map, the actual parish boundaries were extracted as
of 1 July 1896 for optimum match with the agricultural statistics for the Danish
part (2). In addition, a map for the German part calculated per unit was ex-
tracted as of 1 July 1900 (3). As it turned out that the German part had not
changed since the German takeover of the areas in 1864, the parish maps as of
1 July 1896 were used for the entire current Danish territory. There are a total
of 1,860 parish polygons in this cohesive parish map, of which 1,744 of the
parishes are located north of the Kongeåen, the border between Denmark and
Germany.
Figure A.3.
Danish (left) and
German (right) parishes in 1900.
172
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
Re 2) - Statistics on agriculture and land use for the Danish part of Den-
mark as of 1 July 1896
All data at Danish parish level were entered into and a table including 1,771
unique parishes with available area data. For most of the towns, the total area
covered both a “district” and a “land part”. To be comparable to the parish
map – usually containing only one parish polygon – the urban and rural par-
ish figures had to be merged.
Furthermore, the division into parishes is not fully identical with the parish
map and the table data, which is why both polygons and table data had to be
merged and added to create a map with coherent data. There was coincidence
between approx. 90% of the parishes and the table data, while the remaining
table data and corresponding polygons for the parishes had to be ”tweaked”.
Areas are also found that, for some reason, are included in or excluded from
the parish table data. For example, land use for a part of Store Vildmose was
calculated collectively instead of using the 3 + 2 parishes where Store Vild-
mose is located.
The overall process was further complicated because of name coincidences
and mistyping during the entry of around 100,000 single figures. In addition,
a number of input errors and deficiencies were identified and subsequently
rectified in connection with the quality check.
In connection with the quality check, it was also found that almost no parish
areas in the polygon areas in the parish maps and total area in the table data
for the same parishes were identical. Thus, there are a number of major dif-
ferences, particularly along the coasts, where ongoing erosion, containment
and drying created major changes in the landscape around the turn of the
century – for example, at Vejlerne in Hanherred and in southern Lolland.
However, most area deviations are within 5%.
For the Danish part, the total land use statistics yield the percentages shown
in Table A.2.
In order to create a uniform map, the table area was adjusted to ensure that it
was 100% aligned with the polygon area of the parish map, although this pro-
duced a change in the statistically recorded data. For the largest of these area
deviations, for example along the coast, an individual assessment was made
of whether or not the “water bodies” category should be increased to match
the areas. Furthermore, for some parishes recovered land was included in ta-
ble data but not in the parish polygons. For these areas, the table area was
reduced proportionally. See also Figure A.4.
The areas were calculated on 34 crops and other land uses and several sum-
maries. See also Table A.2.
The final map contained 1,702 parishes. A complete overview of the modifi-
cations made was prepared to facilitate other inventories, for example con-
cerning livestock statistics. This is not included in this annex, however.
173
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0176.png
Table A.2
Land use categories and distribution of the area under Danish administration in
1900: Parish data collected in 1896.
Code
DA1
DA2
DA3
DA4
DA5
DA6
DA7
DA8
DA9
DA10
DA11
DA12
DA13
DA14
DA15
DA16
DA17
DA18
DA19
DA20
DA21
DA22
DA23
DA24
DA25
DA26
DA27
DA28
DA29
DA30
DA31
DA32
DA33
DA34
Total
Land use category
Wheat
Cereal rye
Barley
Oats
Mixed cereals (mature)
Buckwheat
Pulses
Spurrey (mature)
Caraway and oil-seed rape
Seed production (clover, grass, beets, lupines)
Potatoes
Sugarbeets and chicory
Carrots
Fodder beets
Green forage (mixed cereals, spurrey, lucerne)
Flax, hemp and tobacco
Garden crops
Black fallow (vegetation-free)
Black fallow ( green manure before ploughing)
Semi-black fallow (early summer-crop)
Cultivated grass for hay
Cultivated grass for grazing
Meadows
Fens and commons
Moors and peatland
Hedgerows and shelters
Gardens and plant nurseries
Forest area (planted)
Forest area (unplanted)
Heathland
Sand dunes and shifting sands
Swamp, foreshores, stone fields etc.
Roads, building sites and storage areas
Lakes, ponds, streams (outside sea territory)
% of total area
0.9
7.6
7.4
11.6
3.1
0.3
0.2
0.2
0.0
0.1
1.4
0.3
0.2
1.8
1.3
0.0
0.0
5.1
0.1
1.4
6.9
17.9
6.0
2.5
2.0
0.2
0.9
6.3
0.7
9.2
1.1
0.4
2.3
0.3
100.0
Re 3) Statistics on agriculture and land use for the German part of Den-
mark as of July 1 1900
The German agricultural data were calculated as of 1 July 1900 in “Amtsbezirk”
and “Flecken”, which are approximately parishes and towns. The parish maps
contains 116 polygons, while the agricultural data contains 88 parishes and
towns. Of the 88 parishes in the table data, two small areas (less than 25 ha)
cannot be accurately located and are not included in the total area.
174
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
As for the Danish part of the German data, polygons and table data had to be
combined to create a coherent map. The combination process was compli-
cated as the old parish boundaries along the current state border did not fully
coincide with the current state border. For the areas containing data from both
sides of the current border, a proportionate part of the total land use was used.
In addition, there were large area deviations along the entire Wadden Sea
coast and in the Vidå area, partly due to a lack of or delayed updating of the
parish map in relation to the course of the coastline, including continuous im-
provement of coastal protection and containment and drying of former land
areas in the latter half of the 1800s. There may also be errors in data or at least
a difference in the cadastral area of the parishes and the aggregated area spec-
ified in table data.
The parental map cannot be customised to ensure 100% correspondence be-
tween data and polygons. Therefore, the polygon areas were maintained, and
the land use was distributed proportionally within the polygons. For a num-
ber of parishes along the Wadden Sea coast, an individual assessment of
whether the excess table area had to be adapted by reducing the category of
“water bodies” was made.
A complete overview of the adjustments was prepared. The final parish map
contained 64 parish polygons so that similar inventories on, for example, live-
stock statistics, were provided. The table of modifications is not included in
this annex.
The German data contained more crop categories than the Danish data. To
maintain the original data and to allow comparison, three code tables were
used, of which the original crop categories were used on the sub-maps (the
German and the Danish part), while a collective code table was then applied
to the resulting background map for the Danish Environmental Protection
Agency. The total code table contains 26 land uses. The German categories are
shown in Table A.3, which also shows the distribution among the categories.
175
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0178.png
Table A.3
Land use categories and distribution of the area under German administration in
1900.
Category Code Land use category
TY1
TY2
TY3
TY4
TY5
TY6
TY7
TY8
TY9
TY10
TY11
TY12
TY13
TY14
TY15
TY16
TY17
TY18
TY19
TY20
TY21
TY22
TY23
TY24
TY25
TY26
TY27
TY28
TY29
TY30
TY31
TY32
TY33
TY34
TY35
TY36
TY37
TY38
TY39
TY40
TY41
TY42
TY43
TY44
TY45
TYG
TYH
TYI
TYJ
TYK
TYL
TOTAL
Winter wheat
Spring wheat
Winter rye
Spring rye
Winter barley
Spring barley
Oats
Mixed cereals (winter)
Mixed cereals (summer)
Buckwheat
Peas
Fava bean
Vetch
Mixed cereals
Mixed pulses
Other types
Potatoes
Sugar beets
Fodder beets
Carrots
Fodder radish
Swedes
Field herbs and caddish
Other types
Winter rape and radish
Leindotter (Camelina sativa)
Flax
Other types
Clover (for forage)
Lucerne
Seradel
Spurrey
Seed production (clover, grass-clover)
Maize
Vetch
Lupines (for forage)
Mixed legumes
Mixed vegetables (for forage)
Mustard
Lupines
Mixed vegetables
Mustard
Fallow
Cultivated grass
Gardens and fruit plantations
Meadows
Pastures
Forests
Buildings and yards
Uncultivated land
Roads and lakes, ponds, streams
% of total area
1.6
0.0
5.6
0.0
0.0
4.4
8.9
0.0
2.1
0.7
0.1
0.0
0.0
0.4
0.0
0.0
1.0
0.0
0.6
0.1
0.1
1.3
0.0
0.0
0.0
0.0
0.0
0.0
1.0
0.0
0.0
0.0
4.2
0.0
0.0
0.0
0.2
0.0
0.0
0.0
0.0
0.0
3.3
25.0
0.7
10.7
11.1
3.6
0.7
5.8
6.7
100.0
176
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0179.png
After establishing both the totals from the Danish and the German statistics,
a total of the statistics on the Danish territory for 1900 could be made by merg-
ing some of the 34/51 categories into 26 groups using the code table in Table
A.4.
Table A.4
Combinations of the merged land use categories.
Code
S01
S02
S03
S04
S05
S06
S07
S08
S09
S10
S11
S12
S13
S14
S15
S16
S17
S18
S19
S20
S21
S22
S23
S24
S25
S26
Total
Land-use category
Wheat
Cereal rye
Barley
Oats
Mixed cereals
Buckwheat
Pulses
Spurrey
Caraway and rape
Seed production
Potatoes
Sugar beets
Carrots
Fodder beets
Green forage
Flax, hemp and tobacco
Garden crops
Fallow
Cultivated grass
Meadows, fens and commons
Moors, peat- and heathland
Forest
Roads and building sites
Lakes, ponds and streams
Buildings and yards
Roads and water areas
Area (ha)
41,428
315,998
301,057
479,084
129,831
14,530
11,600
7,641
525
25,028
56,743
13,583
6,535
76,910
51,277
293
37,549
267,843
1,053,714
407,621
587,555
295,580
88,336
13,086
2,967
17,441
4,303,762
% of total area
1.0
7.3
7.0
11.1
3.0
0.3
0.3
0.2
0.0
0.6
1.3
0.3
0.2
1.8
1.2
0.0
0.9
6.2
24.5
9.5
13.7
6.9
2.1
0.3
0.1
0.4
100.0
DA code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17, 27
18, 19, 20
21, 22
23, 24
25, 30, 31, 32
26, 28, 29
33
34
No code
No code
TY code
1, 5
3
6
7
2, 4, 8, 9
10
11, 12, 13, 14, 15, 16
32
25, 39, 42
29, 30, 31, 33
17
18
20
19, 21,22
23,24, 35, 36,37, 38, 40
27, 28
45
43
44
G, H
K
I
No code
No code
J
L
The map in Figure A.4 shows where polygon areas fitted with table data
within 5% (yellow), where the table data were lower by at least 5% compared
to the parish polygon (green) and where the table data were larger by at least
5% compared to the parish polygon (red).
After the creation of the common categories, a single map could be drawn for
the whole of Denmark for the year 1900, and area distribution was calculated
for the 1,766 parishes. For the areas in Hvidding and Skærbæk parishes south
of Ribe along the Wadden Sea coast, a specific ”sea area“ was withdrawn from
the table area to align it with the parish polygon area. For the rest of the par-
ishes, the areas were adjusted proportionally.
177
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0180.png
Figure A.4
Map showing the
alignment of table and polygon
data in the parishes map.
Re 4) - Digital ID15 water catchment map
After creating a parish map for the total Danish territory in 1900, it was dis-
/aggregated to the 3,134 ID15 areas and 316 coastal water areas. A compari-
son of the parish map with the ID15 map showed that the area of the ID15
map was 795 km
2
larger than the area of the parish map, while 139 km
2
in the
parish map were not included in the 3,037 km
2
large ID15 map. See the maps
in Figure A.5 – A.7.
Figure A.5
Map of areas in the
parish map not found in the ID15
map.
178
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0181.png
Figure A.6
Map of areas in the
ID15map not found in the parish
map.
Figure A.7
Map of Denmark
whole country (top) and zoomed
in at “Limfjorden” (bottom) show-
ing the difference between the
ID15 map and the parish map.
The differences are marked in
green and yellow. Green is
mainly flooded land and yellow
mainly lake area and former sea
territories (see text below).
Overall, it is the lake areas (yellow areas in Figure A.7) that are not included
in the parish map. Some of these areas have been reclaimed and dried out,
others are emerged sea areas along the coasts that have not yet been included
in the parish map. Finally, it is primarily flooded land that is not part of con-
temporary ID15 maps (green areas in Figure A.7.). In the aggregation to ID15,
a “water area” was therefore added to the area calculation to compensate for
any missing ID15 area. This is not entirely correct, but it difficult to get more
specific considering the uncertain statistics.
179
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0182.png
Figure A.8
Parish map versus ID 15 map (left) and coastal land map (right).
Table A.5
Categories and areas from the Danish and German data.
Combination Land use
categories
S01
S02
S03
S04
S05
S06
S07
S08
S09
S10
S11
S12
S13
S14
S15
S16
S17
S18
S19
S20
S21
S22
S23
S24
S25
S26
Total
Wheat
Cereal rye
Barley
Oats
Mixed cereals
Buckwheat
Pulses
Spurrey
Caraway and rape
Seed production
Potatoes
Sugar beets
Carrots
Fodder beets
Green forage
Flax, hemp and tobacco
Garden crops
Fallow
Cultivated grass
Meadows, fens and commons
Moors, peat- and heathland
Forest
Roads and building sites
Lakes, ponds and streams
Buildings and yards
Roads and water areas
Area DK
(ha)
41,428
315,998
301,057
479,084
129,831
14,530
11,600
7,641
525
25,028
56,743
13,583
6,535
76,910
51,277
293
37,549
267,843
1,053,714
407,621
587,555
295,580
88,336
13,086
2,967
17,441
4,303,762
1.0%
7.3%
7.0%
Winter crops
Winter crops
Spring crops
Pct
DK-MODEL
11.1% Spring crops
3.0%
0.3%
0.3%
0.2%
0.0%
0.6%
1.3%
0.3%
0.2%
1.8%
1.2%
0.0%
0.9%
6.2%
Spring crops
Spring crops
Spring crops
Spring crops
Spring crops
Grass
Root crops
Root crops
Root crops
Root crops
Grass
Spring crops
Spring crops
Fallow
24.5% Grass
9.5%
Grass
13.7% Nature
6.9%
2.1%
0.3%
0.1%
0.4%
100.0%
Forest
Other land use
Other land use
Other land use
Other land use
180
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0183.png
There were 1,766 parish polygons, where the areas had to be transferred to
3,134 ID polygons. As seen in Figure A.8 to the left, the red ID15 polygons
were joined by portions of the area distribution from the blue parish polygons.
In the Figure to the right, the ID15 polygons were replaced with the 316 coastal
catchments.
In the ID15 map, data were aggregated to the 316 coastal catchments. The total
statistics used in the calculations for the Danish EPA thus included the area
for 26 crop categories for each of the 316 coastal catchments. In order to run
the NNM-model calculations for the 500 m grid and 10 km DMI grid, were
needed. Land use data at these levels was calculated in the same way as from
parish to ID15, as described above.
The data processing allows calculation of the total land use in 1900 for the
various aggregate area categories related to the 316 coastal water catchment
area. Table A.5 shows that approx. 67% of the area was used for agriculture
or cultivation purposes (S1-S19).
For the use of GEUS’ work in the DK-model, a map based on 500 m grid (250
m grid on Bornholm) was used, see further in section A.2. The DK-model uses
only eight categories, and the 26 categories therefore had to be aggregated
into fewer groups. This was done following the model in Table A.5.
The following Figures (A.9-A.16) show the distribution of all eight DK-model
categories in 1900 for the whole country at parish level (upper) and at ID15
level in 2015 (lower).
Figure A.9.
Winter crops.
181
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0184.png
Figure A.10
Spring crops
182
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0185.png
Figure. A.11
Grass.
183
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0186.png
Figure. A.12
Root crops.
184
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0187.png
Figure A.13
Fallow.
185
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0188.png
Figure A.14
Nature.
186
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0189.png
Figure A.15
Forest.
187
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0190.png
Figure A.16.
Other land use.
A.2 DK-model
The DK-model is a National Water Resources Model for Denmark developed
by the Geological Survey of Denmark and Greenland in 1996
5
. The DK model
is grid based. For Bornholm (blue in Figure A.16.), a 250 m grid was used or
0.0625 ha per point/grid, and for the rest of Denmark (green in Figure A.16.)
a 500 m grid or 0.25 ha per grid/point was applied. The DK-model does not
cover the whole country as seen in Figure A.16, excluding, for instance, some
remote islands like Samsø and Læsø. This means that only 1,747 of the 1,766
parish polygons hold one or more grid points. The parish boundaries of 1896
are shown with red hatch and black outline in Figure A.17.
5
http://dk.vandmodel.dk/in-english/
188
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0191.png
Figure A.17.
DK-model coverage
and parish boundaries in 1896.
The model consists of 220,935 points/grids. A comparison between the grid
map and the parish polygons was made, and the number of points/grids per
parish was calculated.
A total of 175,251 points are found within the parish polygons. There are be-
tween 2 and 1,115 of these grid points per parish. Bornholm, covered by a 250
m grid, has the parishes with the most points, see Figure A.18.
Figure A.18
Example of the dis-
tribution of points within the par-
ish boundaries (Sæby Parish).
Vestermarie Parish thus has 1,115 points, corresponding to 279 points in a 500
m grid. The largest parish polygon in the rest of the country is Gram Parish
with 917 points. For all parishes, the number of points for each of the eight
categories were calculated (example in Table A.6, where the top line shows
hectares for each category the bottom line the number of points for the cate-
gories.
Table A.6.
Example of distribution of points/grids per parish.
Winter
Parish_ID Parish_Name
30921
30921
Sæby Sogn
Sæby Sogn
crop
74
3
Spring
crop
151
6
Grass
278
11
Roots
23
1
Fallow
34
1
Nature
37
2
Forest
10
0
Other
39
2
Ha
646
Number
26
26
189
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0192.png
For a number of parishes, minor differences (+/- 2) occurred between the cal-
culated number of points (total area per parish in ha divided by 0.25 ha out-
side Bornholm and by 0.0625 on Bornholm). Therefore, the final number of
grid point per parish was adjusted to the DK-model point numbers.
The distribution of points for the eight categories of the 1,747 parish polygons
was calculated and a long list of two variables, category and parish, was cal-
culated for the 175,251 points within the parishes.
Table A.7
Example of randomizing the order of the categories.
Category_distribution
ID
128387
128388
128389
128390
128391
128392
128393
128394
128395
128396
128397
128398
128399
128400
128401
128402
128403
128404
128405
128406
128407
128408
128409
128410
128411
128412
OBJ_id
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
30921
Parish_name
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Sæby Sogn
Category
Winter crops
Winter crops
Winter crops
Spring crops
Spring crops
Spring crops
Spring crops
Spring crops
Spring crops
Grass
Grass
Grass
Grass
Grass
Grass
Grass
Grass
Grass
Grass
Grass
Roots
Fallow
Nature
Nature
Other land use
Other land use
Order
0.983833727693675
0.135477146238217
0.848905959640247
0.722893372527041
0.748253493622873
0.410818219434944
0.147878651785771
0.275325349568977
0.930087926500275
0.513536983510489
0.853150756147839
0.155372852730436
0.273081025060814
0.867565884409888
0.894961922037269
0.570764009625787
0.961505090513572
0.743253093446627
0.172367963090125
0.899721533760677
6.04009572611242E-02
0.769277974608645
0.854412472136659
0.679376938655891
0.986322715960438
4.08918354819406E-02
Via the Excel function RAND (), the categories in the list were distributed
”randomly” within the parishes and according to the category attached to the
points lying within the individual parish via sorting of the “Order” column –
see Table A.7. This list of randomised categories was used in the further DK-
model analyses.
The total areas for parish and grid maps were then calculated without the ar-
eas (islands) that are not covered by the DK-model. As can be seen in Table
A.8, the percentages were very well met.
190
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0193.png
Table A.8.
Total areas after creating the randomised map for the DK-model
Area (km
2
)
Category
DK1
DK2
DK3
DK4
DK5
DK6
DK7
DK8
Winter crops
Spring crops
Grass
Roots
Fallow
Nature
Forest
Other land
Total
Legend
Parish map
3,560
9,793
15,358
1,530
2,668
5,000
2,950
1,221
42,081
Grid
3,553
9,788
15,374
1,525
2,663
4,988
2,936
1,224
42,052
difference
7
5
-16
5
5
12
14
-3
30
Parish map
8.5%
23.3%
36.5%
3.6%
6.3%
11.9%
7.0%
2.9%
100.0%
Percentage
Grid
8.5%
23.3%
36.6%
3.6%
6.3%
11.9%
7.0%
2.9%
100.0%
difference
0.0%
0.0%
-0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Finally, a map was prepared (see Figure A.19) based on the random distribu-
tion of crops/land use within the individual parishes from the 1900 parish
map.
Figure A.19
Map of the DK-
model with land use categories in
1896 from Table A.8.
191
 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900 MOF, Alm.del - 2021-22 - Bilag 605: Rapport fra Århus Universitet om estimerede udledninger af næringsstoffer omkring år 1900
2597473_0195.png
TRANSPORT OF NITROGEN AND
PHOSPHORUS FROM LAND TO SEA
AROUND YEAR 1900
The nutrient loads from land to sea around year 1900 in
Denmark are estimated using a delta change modelling
approach considering the numerous factors affecting the
nutrient inputs and transport. The estimates are based
on available data from that time, literature, comparative
analysis methods and modelling tools. The main factors
investigated are climate, hydrology, land use, agricul-
tural practices and drainage, urban developments and
landscape (e.g. nutrient retention in groundwater, wetland,
lakes and streams). Nutrient loads around the year 1900
were affected by human activity, with total nitrogen and
phosphorous loads being approx. 40% and 25-40% less
than presentday loadings, respectively.
ISBN: 978-87-7156-691-8
ISSN: 2244-9981