Miljø- og Fødevareudvalget 2017-18
MOF Alm.del Bilag 35
Offentligt
1803667_0001.png
Bæredygtigt Landbrug
General comments to the evaluation of the models
Comments
Questions
Overall consideration of the N: P ratio
It is important to remember that the water environment target for
Denmark as well as EU is ”good ecological status”.
Although N restrictions dominate the Danish environment strategy, it
is worth mentioning that neither EU nor WFD have similar
requirements.
The high demands for reduced N discharge from cultivation surfaces
are solely based on Danish model calculations that, besides lacking
statistic quality, do not involve the other main nutrient P – an
important factor according to international scientific material - and
therefore have poor scientific founding.
Further to this, DCE - Danish Centre for Environment and Energy,
has overestimated the Danish share of N in the Danish water body
areas and thus the benefit of N restrictions.
On top of this, the administration has, without political approval,
changed the calculation method, changing the reference level from
“water environment” to “marine areas”, which means an increase in
the reduction requirements of approx. 30%.
N:P Interactions
A fundamental problem in Danish water management is a lacking
recognition of phosphorous (P) impact on the eutrophication status in
the water bodies.
General
Does the
reviewing group agree that
"N limitation" means that the
ecosystem receives too much P
(from wastewater, run off and
sediment) - not that nitrogen must be
controlled?
Especially the well documented interaction between N and P in
marine environments is completely underestimated. This has caused
a disproportionate focus on the effect of nitrogen, while the effect of
phosphorous has not been included in the equation.
Focus has been missing on phosphorous wastewater emissions from
Danish wastewater treatment plants that have lacked the efficiency of
equivalent plants in our neighbouring countries. Furthermore, a series
of wet weather overflow have been grossly underestimated as their
occurrence exceeds the statistics with around 50%.
The fact that the P emissions from city wastewater lead to
requirements for substantial agricultural N restrictions based on
unqualified nitrogen calculations has had great impact on the
agricultural sector in Denmark and has resulted in tremendous costs
N and P in marine environments
Increased focus on the function between the two main nutrients,
nitrogen (N) and phosphorous (P) and their interaction is crucial.
The following statement illustrates the abovementioned approach:
“Nitrogen is the determining factor for our marine environment. It is without
any doubt. It is an absolute fact”
(Stiig Markager, DCE, Aarhus University, who is involved in ministerial
consultancy).
Such statements have misled the politicians and the public. The fact
is that opinions differ on this subject. Among international scientists
there is a different view (please see appendixes A) Ecological
1
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0002.png
Bæredygtigt Landbrug
Stoichiometry, p. 324-329, B) 1999 Downing Ecology 80.1157-1167,
C) 1997 Downing Marine np stoichiometry).
For years, DCE has misled the public with one-track minded
statements on nitrogen. This has created an artificial opposition
between production and environment interests.
The resulting administration focusing solely on N restrictions has not
led to the expected improvements for the water environment but has
had great negative impact for the farming industry.
DCE has assessed the water environment incorrectly. The model
attempting to link the spread of eelgrass with nitrogen emissions is
today being called into question by most scientists but is still playing a
key role for environmental targets and actions.
Phosphorous in wastewater creates problems
One of the world’s leading scientists on nutrients in marine
environments, professor John A. Downing, Iowa State University, has
carried out a substantial analysis, gathering and analysing scientific
studies from all over the world.
Downing documented that nitrogen sensitivity is present in
phosphorous polluted water systems only. Unfortunately, almost all
research regarding eutrophication problems has been done in
phosphorous polluted coastal waters where the eutrophication issues
are most evident. Therefore, sufficient light has not been shed on the
interaction between nitrogen and phosphorous, which shows that
nitrogen is only problematic in phosphorous polluted waters.
DCE at Aarhus University has not acknowledged this interaction
effect.
In below schematic illustration by Downing of the N:P cycle in the
water system, the red arrow shows how the Danish strategy with N
restrictions counteracts/delays the natural adaption towards the
Redfield Comfort Zone.
2
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0003.png
Bæredygtigt Landbrug
Conclusion
It has been established that P is the limiting nutrient in water
environments under natural conditions. In inland rivers and lakes, it is
widely recognised that P is the limiting nutrient. In the oceans, P is the
limiting nutrient in most cases. Only around populated areas with
wastewater emissions to fjords and coastal waters, do we have
nitrogen limitation. The reason for this is wastewater with low N:P
ratio, due to anthropogenic pollution.
Local nitrogen limitation in fjords and coastal waters are a result of
this low N:P wastewater pollution combined with natural
denitrification/retention of nitrogen, that lower the N:P ratio further.
None of these processes are caused by nitrogen from cultivation
surfaces.
Still this nitrogen limitation has caused faulty conclusions in most of
the world, however most significantly in Denmark where DCE for
three decades has supported a strict nitrogen strategy and lost focus
on phosphorous.
General comments to the scientific documentation
1. Prologue
Comments
Questions
General
3
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0004.png
Bæredygtigt Landbrug
2. Introduction
Chapter/section
Comments
Questions
2.1
Uncertain nitrogen assessment methodology behind the River
Basin Management Plans
(Development of models and methods to support the establishment of
Danish River Basin Management Plans)
The assessment methodology used by the Ministry of Environment and
Food of Denmark for river basin management in coastal waters, which is
used for determining the environmental objectives and mitigation
demands in the river basin management plans, is based on
unsatisfactory groundwork which does not meet the Water Framework
Directive (WFD).
The assessment methodology is being exploited as a means to stop
agricultural use of nitrate fertilisers and to force the farmers in the
direction of organic farming.
In the report “Development of models and methods to support the
establishment of Danish River Basin Management Plans – Part 1
Methods for determination of target loads” carried out for the Danish
Nature Agency (Harley Bundgaard Madsen, Stig Eggert Pedersen) by
Aarhus University (Karen Timmermann, Stiig Markager, Jesper
Christensen, Ciaràn Murray) 22/12/2014*, the following is stated
(translation by Bæredygtigt Landbrug as no English version is available):
“the ecological status in the WFD is defined in terms of three
biological quality elements: Phytoplankton, benthic vegetation and
benthic flora. Multiple indicators are used for each quality element.
Presently in Denmark intercalibration has been done for only one
indicator of each quality element. The model tools developed are
focused on the two following quality elements and accompanying
indicators:
Phytoplankton, described by the concentration of chlorophyll
Benthic vegetation, described by eelgrass depth limit
The intercalibration indicator for aquatic fauna is based on species
composition, which is not readily applicable in the development of
the model.
In addition to the quality elements, the models can describe
physico-chemical supporting parameters, which are relevant to take
into consideration when evaluating the ecological status. The
following supporting parameters play a key role in the model tools:
Nutrient loads
Light conditions
2.2
4
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0005.png
Bæredygtigt Landbrug
Many other parameters are influencing the eco systems, including
presence of harmful algal blooms, spread and abundance of
eelgrass, lack of oxygen, organic substances in sediments, etc.
However, as the intercalibrated biological indicators are primarily
used in defining the River Basin Management Plans, focus is on
these indicators in the model tools development.”
Neglect of the WFD requirements
The WFD requires that the classification of the ecological status in costal
zones is to be based on the following three biological quality elements
with quantity and density as determining factors:
Phytoplankton (chlorophyll)
Benthic vegetation (eelgrass depth limit)
Benthic fauna (species composition)
In the WFD, the physico-chemical supporting parameters (including
limitation of nutrient loads to the marine areas) are secondary to the aim
of obtaining “good ecological status”.
However, in the Danish implementation of the directive, the supporting
parameters have been given more importance than the quality elements.
The limitation of nitrates has been made the main target while two of the
three biological quality elements (phytoplankton and benthic vegetation)
have been made less important targets and the third indicator (benthic
fauna) has been entirely eliminated because it is not “model friendly”.
The model uses the following calculation to determine the environmental
targets (=target loads):
mitigation demand (%)
Target Load = 1 –
(
100
)
x present status
Based on the nitrogen input from 2007 to 2012 (61 ktons N/year), the
target load (environmental target) for all water bodies has been defined
to 44.5 ktons N/year in 2021.
Thus, instead of the three quality elements laid down in the directive, the
environmental target/target load of the Danish water body management
plans are based on one secondary supporting parameter; nitrogen
reduction. This is not according to the Water Framework Directive.
* http://naturstyrelsen.dk/media/131361/3_1_modeller-for-danske-fjorde-og-
kystnaere-havomraader-del1.pdf - Page1
Legal comments on the International Nitrogen Assessment
The International Nitrogen Assessment originates from the Water
Framework Directive (WFD), which is why we will deal with a few legal
points which are relevant in connection with the assessment of the
nitrogen models.
General
The term “water services”, which is defined in the WFD, has been
translated incorrectly from English to Danish. This causes a limitation to
the field of application in Danish law. The consequence is that the
required economic analysis does not include important matters such as
drainage.
The principle of proportionality is a legally binding EU principle. A
specific requirement in the principle is that the content and form of the
5
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0006.png
Bæredygtigt Landbrug
action must be in keeping with the aim pursued. The more far-reaching
the effect of the models, the more important the accuracy, and most
importantly that the responsibility for the recipient not being able to
achieve “good ecological status” does indeed originate from the
agricultural discharges of nitrates.
Water Services contra the Danish translation
“forsyningspligtydelser”
In the English version of the Water Framework Directive, the term “Water
Services” has been translated to the Danish word
“forsyningspligtydelse”. The English version is as follows as in article
2.1.38:
”‘Water services’ means all services which provide, for
households, public institutions or any economic activity:
a)
abstraction, impoundment, storage, treatment and distribution
of surface water or groundwater, 22.12.2000 EN Official
Journal of the European Communities L 327/7
waste-water collection and treatment facilities which
subsequently discharge into surface water.”
b)
The Danish version has the following wording:
”’Forsyningspligtydelse’: alle ydelser, som for husholdninger,
offentlige institutioner eller økonomiske aktiviteter af enhver art
stiller følgende til rådighed:
indvinding, opmagasinering, oplagring og behandling af samt
forsyning med overfladevand eller grundvand 22.12.2000 DA De
Europæiske Fællesskabers Tidende L 327/7
b) anlæg til opsamling og rensning af spildevand med efterfølgende
udledning til overfladevand.”
The language concept of the word ”water services” is far wider than the
Danish translation “Forsyningsforpligtelser”. The definition of the
translation is wrong. In English, the meaning of the word “water services”
is:
”‘Water services’ means all services which provide[...]any economic
activity:[...]”
The English definition includes any service of water for any economic
activity. This wide definition also includes the sewage and drainage, as
drainage quite obviously fulfills article 2.1.38 a:
”[...]distribution of surface water[...]”
One of the definitions used for the Danish word “forsyningspligtydelse” is
water use, see article 2.1.39. Water use has the following wording in
Danish:
”’[...]Water use: forsyningspligtydelser together with any other
activity as laid down according to article 5.1.II which has impact on
the status of surface waters and on groundwater.
This definition is applicable for article 1 and for the economic
analysis to be done according to article 5.1.III. [...]”
From article 5 the following can be highlighted:
”’[...]Characteristics of the river basin district,
review of the
environmental impact of human activity and economic
analysis of water use
6
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0007.png
Bæredygtigt Landbrug
1. Each Member State shall ensure that for each river basin
district or for the portion of an international river basin
district falling within its territory:
- an analysis of its characteristics,
-
a review of the impact of human activity on the status
of surface waters and on groundwater, and
-
an economic analysis of water use
is undertaken according to the technical specifications set
out in Annexes II and III and that it is completed at the latest
four years after the date of entry into force of this Directive.
2. The analyses and reviews mentioned under paragraph 1
shall be reviewed, and if necessary updated at the latest 13
years after the date of entry into force of this Directive and
every six years thereafter.
With a correct understanding of “water services”, the authorities are
obliged to carry out an economic analysis of the impact of different ways
of water use, such as agricultural drainage, and an evaluation of how the
human activities will impact surface waters and ground water. It may be
that general estimates of impact sources have been carried out, but the
impact sources are general and do not take human development into
account, let alone drainage (water use). So to be absolutely clear; No
economic analysis of the water use has been made.
A correct implementation of the WFD requires that an economic analysis
of the water use has been carried out. This has not happened. Nor has a
correct economic consequence evaluation of the water basin
management plans and the accompanying directive been carried out.
Proportionality
The principle of proportionality is used in most EU countries. It is
however, the EU principle of proportionality that applies in connection
with EU law. Hence the principle applies to the WFD including the
models.
The principle has been established in a range of rulings with varying
wording. In T-290/12 the wording is as follows:
”80. It is settled case-law that the principle of proportionality, which
is one of the general principles of EU law, requires that measures
implemented by acts of the European Union are appropriate for
attaining the objective pursued and do not go beyond what is
necessary to achieve it (see judgments of 6 December 2005 in
ABNA and Others, C
‑
453/03, C
‑
11/04, C
‑
12/04 and C
‑
194/04,
ECR, EU:C:2005:741, paragraph 68; of 7 July 2009 in S.P.C.M.
and Others, C
‑
558/07, ECR, EU:C:2009:430, paragraph 41; and of
8 June 2010 in Vodafone and Others, C
‑
58/08, ECR,
EU:C:2010:321, paragraph 51).”
The principle underlines that
1)
the content and form of the action must be in keeping with the
aim pursued
2)
the action must be limited to what is necessary
7
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0008.png
Bæredygtigt Landbrug
In connection with the nitrogen regulation carried out on basis of the
WFD, the principle means that:
1) the nitrogen reduction must be an action that is in keeping with
the aim
2) if no. 1 is applicable, the required nitrogen reductions should not
supersede what is necessary to obtain “good ecological status”.
If other sources than nitrogen from agricultural production occur,
such as wastewater, “sins of the past” (phosphorous reused year
after year as opposed to nitrates), xenobiotics, climate changes,
precipitation, etc., this means that it cannot be taken for granted
that there is a direct connection between agricultural outlet of
nitrogen and “good ecological status”. Furthermore, we cannot be
certain that the recipients (especially referring to the fjords on the
east coast of Jutland) have ever been in a very good ecological
state or that they ever will be, or that obtaining a “good ecological
status” takes longer than expected and that the present limits
have no environmental effect and therefore are not an action that
is in keeping with the aim of obtaining “good ecological status”.
Furthermore, reference is made to a previous ruling from the EU Court of
Justice, C-293/97 from 29
th
April 1999 (“Stanley”). The case is regarding
the nitrates directive and includes the principle of proportionality. Based
on the proportionality and “polluter pays” principles, the ruling states that
the general burden of removing nitrate pollution should not be placed on
agricultural production where the pollution does not come from farming.
Hence, it is just as important to establish the source of the nitrates.
Nitrates come from farming, but there are also other nitrate sources such
as old forests, wastewater, nitrate emission from nature areas, etc.
Reference is made to a new report from 2017 with the Danish title:
“Landbruget og vandområdeplanerne: omkostninger og implementering
af virkemidler I oplandet til Norsminde Fjord”
*
In English: Agriculture and the water basin management plans: Costs
and implementation of actions in the catchment of Norsminde Fjord.
The following extract of the conclusion** is a translation by Bæredygtigt
Landbrug as no English version is available:
”’[...]The costs of meeting the demands in 2021 and 2027 totals
approximately DKK 800 and 2,900 per HA. Of this the costs for
required catch crops, livestock catch crops, and ecological focus
areas are approx. DKK 100 per HA.
These requirements will mean radical changes in the farming
conditions, especially in 2027 where considerable areas are
appointed for fallowing. However, what has not been considered or
included in the analytics, are the full economic consequences,
welfare, regional, and sector wise of these changes; In terms of
more nature areas, reduced husbandry breeding, loss of
workplaces, and costs for land allocation etc. Therefore, especially
for 2027, the calculated costs are subject to strong reservations.
On the other hand it can be concluded with great certainty that the
2021 target will be considerably (three times) more expensive [than
calculated] and require fallowing of vast areas in order to achieve
the 2027 target. [...]”
The costs are significant, probably to the extent that agricultural
production in the affected areas will not be possible under those
conditions. This calls for similar strict demands to accuracy and statistic
8
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0009.png
Bæredygtigt Landbrug
accuracy of the models. The more drastic the consequences will be in
the end, the greater accuracy the models should have. This requires a
closer analysis.
*
https://curis.ku.dk/ws/files/178737610/IFRO_Rapport_258.pdf
**The report, page 41.
3. Danish marine waters
Chapter/section
Comments
Questions
3.1
3.2
General
4. Danish monitoring data DNAMAP (NOVANA)
Chapter/section
Comments
Questions
4.1
4.2
4.3
General
9
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0010.png
Bæredygtigt Landbrug
5. Overview of WFD tool development in a Danish context
Chapter/section
Comments
Questions
5.1
5.2
General
6. Statistical model development
Chapter/section
Comments
Questions
6.1
6.2
6.3
6.4
In more than 40% of the cases, no relation was found
between N load and Kd. How is it then possible to
assume that the relation is still valid by using the so-
called meta model?
6.5
General
10
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0011.png
Bæredygtigt Landbrug
7. Mechanistic model development
Chapter/section
Comments
Questions
7.1
7.2
7.3
7.4
General
11
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0012.png
Bæredygtigt Landbrug
8. Model application
Chapter/section
Comments
Questions
No data – only uncertain model calculations
The quality element phytoplankton (measured as chlorophyll
concentration) is problematic. As there are no chlorophyll
measurements from the reference period around year 1900, there is no
measured reference level. At the same time, there are no data for the
natural load anno 1900.
Therefore, the experts employed by the Ministry of Environment and
Food of Denmark have – contrary to good scientific practice –
constructed a reference level: The experts have in 2015 made a model
calculation of a reference level in relation to the water basin
management plans 2015 and going forward.
Furthermore, chlorophyll measurement is a simplified operational
foundation as the measurements show no information about the quality
of phytoplankton and its value in the food web.
How were reference
conditions for chlorophyll
and Kd estimated in the
mechanistic model
approach?
The reference conditions
for chlorophyll (defined by
chlorophyll concentrations
more than 100 years ago?)
are directly related to the
target concentrations for
the different water bodies,
and directly builds (applying
modelling?) on the level of
nutrient loads required to
reach target
concentrations. Without
(any) old measurements of
nutrient concentrations or
chlorophyll, the methods to
estimate nutrient loads from
Danish catchment areas as
they were around year
1900 must have been a
formidable task.
Unfortunately, we cannot
find information in the
report how these “ancient”
nitrogen and phosphorus
concentrations in water
courses emptying into
fjords and coastal waters
were estimated.
We suggest it would be
advisory to submit this
important documentation to
avoid any suspicions of
data-related misconduct or
other misunderstandings.
If existing nutrient
concentrations measured in
streams draining
uncultivated soil
(“naturvandløb”) were used
to represent reference
concentrations, we must
remind the reviewing group
that the majority of Danish
land was cultivated and
fertilizer (manure) was used
also more than 100 years
ago. Therefore, one cannot
assume that conditions in
“naturvandløb” directly can
8.1
12
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0013.png
Bæredygtigt Landbrug
represent streams draining
farmed land more than 100
years ago! If other methods
were used to estimate
ancient nutrient
concentrations we would
like to know the details.
Secondly, irrespectively of
the methods used to
estimate “reference”
concentrations of nutrients
in streams 100 years ago
such estimates inherently
must be associated with
very large uncertainties; ±
50% will not be an
unreasonable uncertainty
range. We would like to
know the influence on
reference
chlorophyll of
applying ± 50% nutrient
concentration as reference,
and how recalculated target
chlorophyll and “required”
nutrient reductions will be.
8.2
It is stated that
phosphorous loading has
Behind the decision of equating the WFD quality element benthic
significant positive
vegetation with the secondary supporting parameters nutrients and light
coefficients to Kd in spring-
conditions/chlorophyll for determination of environment targets and
early summer in 14 out og
mitigation demands, is a simplified understanding that less N equals
22 stations.
more eelgrass. As the use of nutrients was very limited around year
Is it the opinion of the
1900 and as by coincidence substantial historical information about the
reviewing group that lack of
spread of eelgrass was available for the years 1883-1929, a reference
funds and time is an
value and an environment target (74% of the reference value, ref. EU
acceptable reason for
intercalibration) for eelgrass spread was constructed in the 1990’ies.
ignoring the phosphorous
effect in the model?
The undocumented chain of events, that this strategy is based upon,
assumes:
The Model Tools
8.3
that the eelgrass has disappeared due to light attenuation
that the light has disappeared due to shading phytoplankton
(chlorophyll)
that phytoplankton has increased due to more nitrates in the
water
and that the increase of nitrates in the water is caused by the
increased use of nutrients in agricultural production since year
1900
At least two of the four prerequisites are problematic – for instance more
than 80% of light attenuation is due to other factors than chloroplhyll.
*
Hence the model has also been rejected as steering parameter, yet the
Ministry of Environment and Food of Denmark still uses it as basis for
determination of environmental requirements.
13
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0014.png
Bæredygtigt Landbrug
The model completely neglects the fact that the phosphorous
concentration has risen dramatically in coastal areas since year 1900
due to wastewater discharge and that the cause for low secchi depth is
a high concentration of phosphorous rather than nitrate. As it is, the
model – the so-called Laurentius relation – does not take phosphorous
into account at all. Evidently, if you do not look for a connection, you will
not find it
*In an article from January 2016, Recovery of Danish Coastal Ecosystems After
Reductions in Nutrient Loading: A Holistic Ecosystem Approach, written by 13
scientists with professor Bo Riemann, DCE as main author:
“In fact, results have shown that chlorophyll accounted for a small fraction, often
less than 20 % of light absorption and
attenuation”.
http://link.springer.com/article/10.1007/s12237-015-9980-0
8.4
8.5
8.6
8.7
8.8
Exploitation of the Water Framework Directive (WFD)
Instead of ensuring “good ecological status” according to the three
quality elements (phytoplankton, benthic vegetation, and benthic fauna),
the EU Water Framework Directive is being exploited to enforce a
limitation of agricultural use of nitrogen fertilizers.
General
The annual NOVANA status reports from the Ministry of Environment
and Food of Denmark as well as analytics carried out by the Danish
Hydraulic Institute (DHI) for the years 1987-2007, have documented that
a 50% reduction of the nitrogen runoff to the marine areas have had no
effect on the depth limit of the eelgrass (incidentally nor on secchi
depths and oxygen conditions). The Model basis is not tenable: The link
14
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0015.png
Bæredygtigt Landbrug
between nitrogen input and eelgrass abundance has not been
scientifically justified.
Despite this, the Ministry of Environment and Food of Denmark
continues with the existing nitrogen strategy. This alternative Danish
implementation of the WFD is not aiming for “good ecological status” in
the marine areas as much as aiming for preventing the farmers from
being able to fertilize the crops in a professional and scientifically well-
grounded manner. In reality, the Ministry of Environment and Food of
Denmark is working on enforcing conversion to organic farming
(Although not openly).
There is no legal or political basis for such a Modus Operandi in Danish
legislation or the EU WFD.
Non-compliance with the Water Framework Directive
The WFD requirement that determination of references must be trough
scientific objectivity has not been fulfilled.
In 2002 the Danish National Environment Research Institute (DMU)
stated (report no. 390 – extract translated by Bæredygtigt Landbrug as
no English version is available):
“It is however not possible to quantify connections between species
composition of phytoplankton, aquatic plants, and benthic fauna
and input of nutrients from the catchment the way it has been laid
down in the Water Framework Directive”.
*
So DMU has not been able to describe a connection between the
nutrient input from the catchment and the biological quality elements
despite the fact that this connection has been the prerequisite for the
strategy employed by the Danish Ministry of Environment for 30 years.
The WFD requirement for a scientific and objective process apposed
with the scientists’ acknowledgement that they “cannot quantify
connections between species composition of phytoplankton, aquatic
plants, and benthic fauna and input of nutrients from the catchment
area”, documents that the Water Framework Directive is not being
followed.
An example
That the nutrient models can have substantial consequences can be
seen in an example from the water body area of Skive Fjord. Here the
model shows that the end target of nitrogen input in 2021 is 6 kg N/HA.
The background contribution from nature is 3.5 kg N/HA, which leaves
2.5 kg N/HA available for farming – the main occupation in the area.
This means that the local authorities will close down farming entirely in
the 144,000 HA coastal zone. It will have no improved influence on
Skive Fjord. It might even be that the situation will be worsened due to
declining plankton quality caused by lower N:P ratio as a result of the
nitrogen limitation. This will again mean that non-edible plankton will be
left to rot, thus promoting oxygen depletion.
The situation is the same for many other water body areas.
It is indisputable that the livelihood of thousands of families will be
removed in these areas based on a scientifically uncertain model.
*
http://www.dmu.dk/1_viden/2_publikationer/3_fagrapporter/rapporter/FR390.pdf
15
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0016.png
Bæredygtigt Landbrug
9. Discussion
Chapter/section
Comments
Questions
9.1
9.2
9.3
9.4
General
16
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0017.png
Bæredygtigt Landbrug
10.-12 Conclusion, Epiloque, and References
Chapter/section
Comments
Questions
Conclusion
Epilogue
References
17
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0018.png
Bæredygtigt Landbrug
Chapter/section
Comments
Questions
Appendix A
Appendix B
Appendix C
General
18
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0019.png
Comments and questions
to
‘Development of models and methods to support the establishment
of Danish River Basin Management
Plan, Scientific documentation’
from
The Danish Society for Nature Conservation
Date: 30-06-2017
Written by: Lisbet Ogstrup, telephone: +45 3119 3209, mail:
[email protected]
General comments to the evaluation of the models
Comments
We find that the models and methods used to support the
establishment of Danish River Basin Management Plans
and described in the Scientific documentation are sturdy,
and give a scientific and objective
assessment of MAI to
each Danish water body.
Questions
General
General comments to the scientific documentation
Comments
Questions
General
1. Prologue
Comments
Questions
1
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0020.png
General
2
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0021.png
2. Introduction
Chapter/section
Comm
ents
Questions
2.1
2.2
With reference to section 2.1 (page 8) saying:
‘… a
conceptual chance has been introduced to create
scientific coherence between the goals of achieving a
certain
–
political defined
–
environmental quality, and the
required reduction of nutrients inputs.’ The essence
of The
Water Framework Directive (WFD) is, that all surface
waters shall achieve at least good ecological and chemical
status.
Do you find that the ecological status in the Danish plan period
2015-21 can be classified according to the three indicators
mentioned in section 2.2.1 (page 9
–
10)?
Do you find, that the ecological status in the Danish plan period
2015-21 could have been classified according to more or other
indicators, than the three indicators mentioned I section 2.2.1 (page
9
–
10)? If so, do you think, that it would have had influence of the
result for the maximum allowable nutrient input (MAI) due to the
models for calculation?
We notice, that the development of the marine model tools Do you find that the marine model tools founded on the
was largely founded on the recommendations of the
recommendations of ‘Eelgrass Working Group II’
are sufficient?
‘Eelgrass Working Group II’.
Can you recommend, that the marine model tools founded on the
recommendation of the ‘Eelgrass Working Group II’ is further
qualified?
General
3
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0022.png
3. Danish marine waters
Chapter/section
Comments
Questions
3.1
3.2
General
4
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0023.png
4. Danish monitoring data DNAMAP (NOVANA)
Chapter/section
Comments
Questions
4.1
4.2
4.3
In the chapter it is mentioned, that originally The Danish
National Aquatic Monitoring and Assessment Programme
(DNAMAP) probably was the most comprehensive
programmes in the world (page 19).
Do you find, that The Danish National Aquatic Monitoring and
Assessment Programme (DNAMAP) probably no longer is the most
comprehensive programmes in the world?
Do you overall find, that the DNAMAP is sufficient according to
numbers of stations and monitoring land-based loadings of N and
of P in Denmark?
General
Do you overall find that the data from DNAMAP can be used to
develop the marina modeling tools as done in the project?
Do you overall find that if the land-based loadings of N and P in
Denmark had been monitored further in DNAMAP in the period
used, it would have result in a greater strength of linear relationship
between modeled and observed data, than shown in the project? If
so, how much more should there have been monitored in order to
get a greater strength of linear relationship between modelled and
5
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0024.png
observed data, than shown in the project?
5. Overview of WFD tool development in a Danish context
Chapter/section
Comments
We notice the recommendation given by the Eelgrass
Working Group II about which models, there should be in
focus (page 24).
Questions
Are those models and methods
–
or similar models and methods -
used to support the establishment of Danish River Basin
Management Plans
–
been develop and used in other
countries/water bodies?
We notice, that both the budget and the time schedule
was taking into account when it was adopted an approach Are the models and methods used to support the establishment of
involving development of four mechanistic biogeochemical Danish River Basin Management Plans generally scientifically
models and statistical models (page 24).
accepted?
5.1
Should there have been developed more than four mechanistic
biogeochemical models and statistical models (if the budget and
the time had not to be taking into account) calculating nutrient
reduction requirement and corresponding MAI to obtain GES?
Considering the Danish water bodies do you assess, that the four
mechanistic biogeochemical models and statistical models
developed sufficient covers the Danish water bodies?
We notice, that it is mentioned, that for the Danish plan
period 2015-21, ecological status is classified to three
indicators (chlorophyll-a, eelgrass depth limit and a fauna
index (DKI). We furthermore notice, that not all of these
indicators can be linked to the model toolbox (page 25).
Do you agree that it was necessary to make the adjustments as
described in section 5.2 (page 25
–
26)?
What is your assessment of the adjustment described in section 5.2
(page 25
–
26)? Could the adjustment have influence on the result
of linear relationship between modeled and observed data?
5.2
6
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0025.png
General
7
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0026.png
6. Statistical model development
Chapter/section
Comments
Questions
Do you agree, that the PLS regression models are an appropriate
tool taking the argument for the chose mentioned in section 6.1 in
to consideration (page 27)?
6.1
Do you in overall find, that it would have been inadvisable, if PLS
regression models had not been chosen as a tool?
6.2
We notice, that the predictors was selected due to their
known ability to act as forcing factors on the indicators
(see table 6.2 at page 30).
Please comment figure 6.1 according to the numbers of official
stations compared to the Danish water bodies and their individual
characteristics (page 29).
Do you find it correct, that only monitoring stations within the zone
of WRD and data series with at least 15 years of data during the
period 1990 to 2012 with a minimum of one bimonthly observation,
has been used (page 29)?
Do you assess, that the selected predictors are the right predictors
in order to developed statistical models in the project?
Could there have been chosen fewer predictors without influencing
the project statistical models and the project result of linear
relationship between modeled and observed data?
6.3
To what extent can it have influenced on the statistical models, that
all data series have not been analyzed for outliers individually
(page 32)?
Do you agree, that in order to balance the two aspect of the
predictor variables described (page 33) it is correct to specify, that
the predictors variables should not start earlier than the year before
the responding variable? And do you find, that the rules for
predictor variables are sufficient (se also figure 6.3 at page 34)?
Do you agree that the additional analyses used to identify the most
likely variable in those cases, where different sets of predictor
variables described the selected responding variable almost
equally, is sufficient (see also page 38)?
8
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0027.png
We notice, that it is referred, that a closer autocorrelation
analysis revealed, that the historical signal for TN have
different effect to different water bodies, but due to the
relative short time series available.
6.4
Do you access, that there could have been done quantification of
autocorrelation in order to improve the models based on time series
available?
Please comment the two last sections at page 43. Do you agree in
the arguments and the assessments in these two sections?
We notice, that the aim of the project is to provide a
model-based management tools for estimation Maximum
allowable loadings (MAI) for each of the 119 marine water
bodies covered by the WFD in Denmark (page 46).
6.5
Do you agree that there is overwhelming evidence in the scientific
literature, that nutrient loadings do have an impact on selected
response variables (page 46)?
General
7. Mechanistic model development
Chapter/section
Comments
Questions
7.1
7.2
We notice, that the modelling work, where the focus was
on the inner Danish waters did not experience any
systematic errors and therefore it could be concluded that
the official data on loadings were valid for the purpose of
the modeling page 59).
7.3
Referring to, that the project found, that the specific acceptance
criteria were lower for the coastal areas and enclosed water bodies
as specific bathymetric details and local conditions become
increasingly important, do you find, that there are scientific
evidence for this (page 62)?
9
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0028.png
7.4
General
10
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0029.png
8. Model application
Chapter/section
Comments
Questions
8.1
8.2
We notice, that there is referred to the principle
‘one-out-
While nutrient loadings are a major pressure factor do you agree
all-out’
in the WFD and
the project considers one pressure that the set up of the project using several indicators to describe the
factor (nutrient loadings) (page 91).
effect of this pressure factor is reasonable and correct? And do you
agree that though not taken the principle ‘one out all out’ into
consideration MAI estimated in project is sturdy?
Do you agree in the assumption, that a weighted average approach
provides a more correct estimate of the maximum allowable load
and making it less susceptible to random variation in the data
parameters (page 91)?
8.3
Do you agree in the use of each of six indicators and arguments for
the modifications and values of the constant involved (an overview
is given in table 8.7 at page 100
–
101))?
Do you agree in the approach to handle the described off-sets and
thus the assumption, that it is a valid approach as the overall
calibration seems strong (page 103)?
Do you find that the percentage chosen for ‘Categorized in case of
time Lag’ are correct in order to the estimated
GES (see table 8.7
at page 100
–
101)?
Do you find that the methodology described is sturdy, and
combined with the reference values from section 8.1 can be used to
estimate the part of the individual indicator that can be regulated
from Danish land-based N loadings alone (page 102)?
8.4
8.5
Do you agree, that even though the nature of the model types
differs pronouncedly, the slopes are very similar, and thus support
both the use of models for defining MAI and the application of water
body types (page 119)?
8.6
11
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0030.png
8.7
8.8
It should be mentioned, that the year 1900 is chosen as
the historical reference conditions in Denmark mainly
founded on historical observations documenting eelgrass
depth distribution and light penetration at that time. See
also section 8.1.
This decision is provided in the models.
General
It should be mentioned, too that the historical observation
is not used directly, even though there are observations
for several Danish water bodies. Instead, it is decided to
use the 90 pct-percentil of the historical observations.
Furthermore it is decided, that the reference for GES is
defined as 25-30 pct. deviation from the reference.
This means there has been used a principle saying that
despite the historical data shows otherwise it is assumed
that GES for Danish water bodies can be estimated at a
lower level.
Can the decisions of how to use the historical observation together
with the handling of the model uncertainty and sensitivity result in
an underestimated nutrient reductions requirement in one or more
of the 119 Danish WFD water bodies to fulfill GES according to the
WFD?
9. Discussion
Chapter/section
Comments
Questions
9.1
9.2
12
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0031.png
9.3
9.4
General
13
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0032.png
10.-12 Conclusion, Epiloque, and References
Chapter/section
Comments
Questions
Conclusion
Epilogue
Do you find the references used in the project are sufficient (page
144
–
163)?
Do you find the references support the tool development and
application, the specific use for setting chlorophyll-a targets and
calculating the load reduction requirements from Danish
catchments in the project?
References
14
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0033.png
Chapter/section
Comments
Questions
Appendix A
Appendix B
Appendix C
General
15
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0034.png
Dansk Akvakultur
Ramme for kommentarer og spørgsmål til evalueringsrapporten om de danske
kvælstofmodeller
Formålet
Spørgerammen skal sikre overskuelighed i kommentarer og spørgsmål fra deltagere i Blåt
Fremdriftsforum til evalueringspanelet, samt i panelets efterfølgende håndtering af disse
kommentarer og spørgsmål. Evalueringsrapporten fra AU og DHI vil sammen med
kommentarer og spørgsmål fra Blåt Fremdriftsforum udgøre det samlede
evalueringsgrundlag, som evalueringspanelet skal forhold sig til i evalueringen.
Guidelines for formulering og fremsendelse af kommentarer og spørgsmål
For at sikre evalueringspanelet de bedst mulige arbejdsbetingelser i forhold til en systematisk
og grundig håndtering af kommentarer og spørgsmål fra Blåt Fremdriftsforum opstilles der en
række guidelines for fremsendelse af bemærkninger til evalueringsrapporten. Hver
interesseorganisation kan i udgangspunktet fremsende sine samlede spørgsmål og
kommentarer én gang i denne skabelon.
Den udfyldte skabelon skal udfyldes og indsendes elektronisk til Implement indenfor den
fastsatte tidsfrist. Kommentarer og spørgsmål skal formuleres på engelsk og i videst muligt
omfang opfylde følgende.
•
Være så kortfattede og så præcise som muligt.
•
Relatere sig til kvælstofmodellernes anvendelse i vandområdeplanerne.
•
Kommentarer og spørgsmål til evalueringsrapporten bør indeholde sidehenvisninger.
•
Baggrundsmateriale kan vedlægges som dokumentation, og der skal i få fald
henvises konkret og specifikt til materialet i denne spørgeramme.
•
Dokumentationsmaterialer kan ikke stå alene som kommentarer og spørgsmål.
Tidsfrist for fremsendelse af kommentarer og spørgsmål
Evalueringsrapporten blev fremsendt til interessenterne i Blåt Fremdriftsforum den 6. juni
2017.
Fristen for fremsendelse af kommentarer og spørgsmål er 4. juli 2017. De skal sendes til
Eske Benn Thomsen,
[email protected],
og de modtages gerne inden fristen.
Implement sikrer i videst muligt omfang, at kommentarer og spørgsmål til panelet lever op til
rammerne, og tager evt. en dialog med enkelte interessenter, hvis der skulle vise sig behov
for tilpasninger i formen.
I kan kontakte Eske Benn Thomsen, hvis I har spørgsmål til udfyldningen af spørgerammen.
På de følgende sider fremgår den tabelramme som I bedes anvende til at indtaste jeres
kommentarer og spørgsmål. Den indeholder indledningsvis felter til generelle kommentarer,
og derefter felter til kommentarer rettet mod de enkelte kapitler/sektioner i
evalueringsgrundlaget. Interessenternes kommentarer og spørgsmål vil blive vedlagt
evalueringsrapporten og dermed udgøre en del af den endelige dokumentation for
evalueringen af kvælstofmodellerne.
1
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0035.png
Dansk Akvakultur
General comments to the evaluation of the models
Comments
Questions
General
General comments to the scientific documentation
Comments
Questions
General
1. Prologue
Comments
Questions
General
2
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0036.png
Dansk Akvakultur
2. Introduction
Chapter/section
Comm
ents
Questions
2.1
2.2
Per 1: The eel grass tool ”However,
though the best availably tool at that
time, it…”
Ell grass tool was not the best tool at
that time. There were models (DHI)
which were much better developed.”
General
3
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0037.png
Dansk Akvakultur
3. Danish marine waters
Chapter/section
Comments
Questions
3.1
3.2
4
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0038.png
Dansk Akvakultur
General
4. Danish monitoring data DNAMAP (NOVANA)
Chapter/section
Comments
Questions
4.1
An aquaculture plant in Smålandshavet
is mentioned as increasing the load
here. The discharge of nutrients is very
low compared to other sources.
Therefore it is incorrect and misleading
and should be removed. There hasn’t
been any new aquaculture farm here in
many years.
4.2
Sentence: ” Despite the efforts to
reduce the diffuse loads, Danish
agriculture remains the major source of
both N (80%) and P (50%) in Danish
streams, lakes and coastal waters
(Kronvang et al. 2005).”
is not correct for coastal waters, as
external sources are far more important
here.
5
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0039.png
Dansk Akvakultur
4.3
General
5. Overview of WFD tool development in a Danish context
Chapter/section
Comments
Questions
5.1
6
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0040.png
Dansk Akvakultur
p. 26 par 2: Is it a lack that biodiversity
not is included.
5.2
General
7
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0041.png
Dansk Akvakultur
6. Statistical model development
Chapter/section
Comments
p.27 par 1: Statistical linear models with
multiple predictors (MLR, mixed
models, PLS
etc.) have previously been applied in
several studies of marine
eutrophication
published in international peer-
reviewed journals (Conley et al.
2007;….”
These models are as far as we know
not pre-reviewed but only used in
reports.
There should be a clear discussion of
the advantages, as well as the
disadvantages of using the statistical
models.
Questions
Is it right that these models are not pre-
reviewed?
The models should be pre-reviewed if they are
used, and it should be clear that the models
are not pre-rewired.
6.1
6.2
6.3
6.4
8
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0042.png
Dansk Akvakultur
6.5
General
7. Mechanistic model development
Chapter/section
Comments
Questions
7.1
7.2
Following sentence is very important:
“As can be seen, there is a strong
correlation between especially the Danish
and the German N loads, but also a rather
strong correlation between the Danish and
the Swedish loads.”
It verifies that models only calculating
Danish loads are misleading.
7.3
9
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0043.png
Dansk Akvakultur
7.4
General
10
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0044.png
Dansk Akvakultur
8. Model application
Chapter/section
Comments
p.73. 8.1.4: Important to discuss the
reasonableness of using 1900 as
historical reference year in relation to
data and natural changeability and
fluctuation.
Questions
Is it optimal to choose 1900 as historical
reference your, or was it better to use an other
periode?
8.1
8.2
8.3
8.4
8.5
It is important to underline that the
statistical models (vs. mechanistic
models) overestimate the Danish
contribution to the eutrophication in the
marine waters.
8.6
8.7
8.8
11
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0045.png
Dansk Akvakultur
General
9. Discussion
Chapter/section
Comments
Questions
9.1
9.2
9.3
9.4
General
12
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0046.png
Dansk Akvakultur
10.-12 Conclusion, Epiloque, and References
Chapter/section
Comments
Questions
Conclusion
Epilogue
References
13
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
Dansk Akvakultur
14
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0048.png
Dansk Akvakultur
Chapter/section
Comments
Questions
Appendix A
Appendix B
Appendix C
General
15
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0049.png
Dansk Sportsfiskerforbund
Ramme for kommentarer og spørgsmål til
evalueringsrapporten om de danske kvælstofmodeller
Formålet
Spørgerammen skal sikre overskuelighed i kommentarer og spørgsmål fra deltagere i Blåt
Fremdriftsforum til evalueringspanelet, samt i panelets efterfølgende håndtering af disse kommentarer
og spørgsmål. Evalueringsrapporten fra AU og DHI vil sammen med kommentarer og spørgsmål fra
Blåt Fremdriftsforum udgøre det samlede evalueringsgrundlag, som evalueringspanelet skal forhold
sig til i evalueringen.
Guidelines for formulering og fremsendelse af kommentarer og spørgsmål
For at sikre evalueringspanelet de bedst mulige arbejdsbetingelser i forhold til en systematisk og
grundig håndtering af kommentarer og spørgsmål fra Blåt Fremdriftsforum opstilles der en række
guidelines for fremsendelse af bemærkninger til evalueringsrapporten. Hver interesseorganisation kan
i udgangspunktet fremsende sine samlede spørgsmål og kommentarer én gang i denne skabelon.
Den udfyldte skabelon skal udfyldes og indsendes elektronisk til Implement indenfor den fastsatte
tidsfrist. Kommentarer og spørgsmål skal formuleres på engelsk og i videst muligt omfang opfylde
følgende.
•
Være så kortfattede og så præcise som muligt.
•
Relatere sig til kvælstofmodellernes anvendelse i vandområdeplanerne.
•
Kommentarer og spørgsmål til evalueringsrapporten bør indeholde sidehenvisninger.
•
Baggrundsmateriale kan vedlægges som dokumentation, og der skal i få fald henvises
konkret og specifikt til materialet i denne spørgeramme.
•
Dokumentationsmaterialer kan ikke stå alene som kommentarer og spørgsmål.
Tidsfrist for fremsendelse af kommentarer og spørgsmål
Evalueringsrapporten blev fremsendt til interessenterne i Blåt Fremdriftsforum den 6. juni 2017.
Fristen for fremsendelse af kommentarer og spørgsmål er 4. juli 2017. De skal sendes til Eske Benn
Thomsen,
[email protected],
og de modtages gerne inden fristen.
Implement sikrer i videst muligt omfang, at kommentarer og spørgsmål til panelet lever op til
rammerne, og tager evt. en dialog med enkelte interessenter, hvis der skulle vise sig behov for
tilpasninger i formen.
I kan kontakte Eske Benn Thomsen, hvis I har spørgsmål til udfyldningen af spørgerammen.
På de følgende sider fremgår den tabelramme som I bedes anvende til at indtaste jeres kommentarer
og spørgsmål. Den indeholder indledningsvis felter til generelle kommentarer, og derefter felter til
kommentarer rettet mod de enkelte kapitler/sektioner i evalueringsgrundlaget. Interessenternes
kommentarer og spørgsmål vil blive vedlagt evalueringsrapporten og dermed udgøre en del af den
endelige dokumentation for evalueringen af kvælstofmodellerne.
1
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0050.png
Dansk Sportsfiskerforbund
General comments to the evaluation of the models
Comments
The work with the models started late in the process.
The WFD was signed in 2000.
This first Danish plan covered the period from 2010-2015
Questions
Do you find that the Danish surveillance is sufficient and is this
data good enough to support the models?
Do you find that there had been the necessary finance and time
for the development of the models?
Is there the necessary continuity in the model work?
General
General comments to the scientific documentation
Comments
Questions
General
1. Prologue
Comments
Questions
General
2
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0051.png
Dansk Sportsfiskerforbund
2.
Introduction
Chapter/section
Comments
Questions
2.1
In Denmark the required reduction of nutrients inputs Section 2.2.1
is political defined. And the reduction have been
It is possible to classiefi the ecological status in the Danish
changed in the period – and prosponed.
plan period 2015-21, according to the three indicators
mentioned?
We find that The Water Framework Directive (WFD),
demands that all surface waters shall achieve at least Could the plan have been classified according to other
good ecological and chemical status.
indicators?
2.2
Is the three indicators representative?
Could that have had influence of the result for the maximum
allowable nutrient input due to the models for calculation?
The development of the marine model tools was founded
on recommendations of the ‘Eelgrass Working Group II’.
Are these tools qualified?
Are they sufficient?
General
3
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0052.png
Dansk Sportsfiskerforbund
3.
Danish marine waters
Chapter/section
Comments
Questions
3.1
3.2
General
4
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0053.png
Dansk Sportsfiskerforbund
4.
Danish monitoring data DNAMAP (NOVANA)
Chapter/section
Comments
Questions
4.1
4.2
4.3
In the chapter it is mentioned, that originally The Danish
Do you agree?
National Aquatic Monitoring and Assessment Programme The Danish National Aquatic Monitoring and Assessment
(DNAMAP) in the start was the best programme.
Programme (DNAMAP) probably no longer is the best programme
in the world?
Do you agree?
Is the numbers of stations and monitoring land-based loadings of
N and of P in Denmark sufficient?
General
5
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0054.png
Dansk Sportsfiskerforbund
5.
Overview of WFD tool development in a Danish context
Chapter/section
Comments
Questions
Development of four mechanistic biogeochemical models Do you find that the necessary money and time was given to the
and statistical models had a budget and time schedule,
development of the models?
that set the frame.
Are the models and methods used to support the establishment of
Danish River Basin Management Plans generally scientifically
accepted?
Do you find that the Danish water bodies is sufficient covered, with
the used of the mechanistic biogeochemical models and statistical
models?
Should there have been developed more models calculating
nutrient reduction requirement and corresponding MAI to obtain
GES?
5.1
In the Danish plan period 2015-21, ecological status
is classified to three indicators.
Section 5.2:
Do you find it was necessary to make the adjustments as
described?
What is your assessment of the adjustment described?
Will you comment the influence on the result of linear
relationship between model and observed data with these
adjustments?
5.2
General
6
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0055.png
Dansk Sportsfiskerforbund
6.
Statistical model development
Chapter/section
Comments
Questions
6.1
6.2
We notice, that the predictors was selected due to their
known ability to act as forcing factors on the indicators.
Please comment figure 6.1 according to the numbers of official
stations compared to the Danish waterbodies and their individual
characteristics.
Do you find it correct, that only monitoring stations within the zone
of WRD and data series with at least 15 years of data during the
period 1990 to 2012 with a minimum of one bimonthly observation,
has been used?
Do you assess, that the selected predictors are the right predictors
in order to developed statistical models in the project?
Could there have been chosen fewer predictors without influencing
the project statistical models and the project result of linear
relationship between modelled and observed data?
To what extent can it have influenced on the statistical models,
that all data series have not been analyzed for outliers
individually?
Do you agree, that in order to balance the two aspect of the
predictor variables described (p. 33) it is correct to specify, that the
predictors variables should not start earlier than the year before
the responding variable? And do you find, that the rules for
predictor variables are sufficient (se also figure 6.3)?
Do you agree that the additional analyses used to identify the most
likely variable in those cases, where different sets of predictor
variables described the selected responding variable almost
equally, is sufficient?
6.3
We notice, that it is referred, that a closer autocorrelation
analysis revealed, that the historical signal for TN have
different effect to different water bodies, but due to the
relative short time series available.
6.4
Do you access, that there could have been done quantification of
autocorrelation in order to improve the models based on time
series available?
7
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0056.png
Dansk Sportsfiskerforbund
The aim is to provide a model-based management tools
for estimation Maximum allowable loadings (MAI) for
each of the 119 marine water bodies covered by the
WFD in Denmark.
6.5
Is there evidence, that nutrient loadings do have an impact on
selected response variables?
General
7.
Mechanistic model development
Chapter/section
Comments
Questions
7.1
7.2
7.3
7.4
8
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0057.png
Dansk Sportsfiskerforbund
General
9
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0058.png
Dansk Sportsfiskerforbund
8.
Model application
Chapter/section
Comments
Questions
8.1
8.2
8.3
It is well documented that hypoxia or anoxia in the bottom
water will accelerate the negative effects of
eutrophication, such as loss of macro vegetation, release
of both nitrogen and phosphorus from the sediment, fish
kills and, ultimately, direct release of hydrogen sulphide
to the atmosphere.
Do you agree, that if the low oxygen concentrations are restricted
to a deep hole in an estuary, it may not have a significant impact
on the estuary as a whole, whereas comprehensive hypoxia
covering a large-sized area will most likely result in notable
derived negative effects.
8.4
8.5
8.6
8.7
8.8
General
The year 1900 is chosen as the historical reference
conditions in Denmark founded on historical observations
documenting eelgrass depth distribution and light
penetration at that time.
The historical observation is not used directly, even
though Denmark have the data. It was decided to use the
90 pct percentil of the historical observations. The
Do you agree, that the use of the historical data together with the
handling of the model uncertainty, result in underestimating the
requirement of nutrient reductions in more of the 119 Danish WFD
water bodies, just to fulfill GES according to the WFD?
10
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0059.png
Dansk Sportsfiskerforbund
reference for GES was defined as 25-30 pct. deviation
from the reference. Thus you have the data it was
decided to assumed that GES for Danish waterbodies
can be estimated at a lower level.
9.
Discussion
Chapter/section
Comments
Questions
9.1
9.2
9.3
9.4
General
11
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0060.png
Dansk Sportsfiskerforbund
10.-12 Conclusion, Epiloque, and References
Chapter/section
Comments
To obtain more certain MAI estimates, it is important to
continuously monitor the ecosystems as they approach
GES and to evaluate, update and improve the models
and methods accordingly based on new knowledge.
Thus, themodel tools and methods developed in this
project should be regarded as part of an ongoing process
towards better understanding and improved predictability
of the behaviour of marine ecosystems in a changing
world.
Conclusion
Questions
Do you find that the Danish surveillance is sufficient and is this
data good enough to support the models?
Do you find that there had been the necessary finance and time
for the development of the models?
Is there the necessary continuity in the model work?
Epilogue
References
12
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0061.png
Dansk Sportsfiskerforbund
Chapter/section
Comments
Questions
Appendix A
Appendix B
Appendix C
General
13
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0062.png
Landbrug & Fødevarer
Ramme for kommentarer og spørgsmål til
evalueringsrapporten om de danske kvælstofmodeller
Formålet
Spørgerammen skal sikre overskuelighed i kommentarer og spørgsmål fra deltagere i Blåt
Fremdriftsforum til evalueringspanelet, samt i panelets efterfølgende håndtering af disse kommentarer
og spørgsmål. Evalueringsrapporten fra AU og DHI vil sammen med kommentarer og spørgsmål fra
Blåt Fremdriftsforum udgøre det samlede evalueringsgrundlag, som evalueringspanelet skal forhold
sig til i evalueringen.
Guidelines for formulering og fremsendelse af kommentarer og spørgsmål
For at sikre evalueringspanelet de bedst mulige arbejdsbetingelser i forhold til en systematisk og
grundig håndtering af kommentarer og spørgsmål fra Blåt Fremdriftsforum opstilles der en række
guidelines for fremsendelse af bemærkninger til evalueringsrapporten. Hver interesseorganisation kan
i udgangspunktet fremsende sine samlede spørgsmål og kommentarer én gang i denne skabelon.
Den udfyldte skabelon skal udfyldes og indsendes elektronisk til Implement indenfor den fastsatte
tidsfrist. Kommentarer og spørgsmål skal formuleres på engelsk og i videst muligt omfang opfylde
følgende.
•
Være så kortfattede og så præcise som muligt.
•
Relatere sig til kvælstofmodellernes anvendelse i vandområdeplanerne.
•
Kommentarer og spørgsmål til evalueringsrapporten bør indeholde sidehenvisninger.
•
Baggrundsmateriale kan vedlægges som dokumentation, og der skal i få fald henvises
konkret og specifikt til materialet i denne spørgeramme.
•
Dokumentationsmaterialer kan ikke stå alene som kommentarer og spørgsmål.
Tidsfrist for fremsendelse af kommentarer og spørgsmål
Evalueringsrapporten blev fremsendt til interessenterne i Blåt Fremdriftsforum den 6. juni 2017.
Fristen for fremsendelse af kommentarer og spørgsmål er 4. juli 2017. De skal sendes til Eske Benn
Thomsen,
[email protected],
og de modtages gerne inden fristen.
Implement sikrer i videst muligt omfang, at kommentarer og spørgsmål til panelet lever op til
rammerne, og tager evt. en dialog med enkelte interessenter, hvis der skulle vise sig behov for
tilpasninger i formen.
I kan kontakte Eske Benn Thomsen, hvis I har spørgsmål til udfyldningen af spørgerammen.
På de følgende sider fremgår den tabelramme som I bedes anvende til at indtaste jeres kommentarer
og spørgsmål. Den indeholder indledningsvis felter til generelle kommentarer, og derefter felter til
kommentarer rettet mod de enkelte kapitler/sektioner i evalueringsgrundlaget. Interessenternes
kommentarer og spørgsmål vil blive vedlagt evalueringsrapporten og dermed udgøre en del af den
endelige dokumentation for evalueringen af kvælstofmodellerne.
1
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0063.png
Landbrug & Fødevarer
General comments to the evaluation of the models
Comments
Questions
The EU Water Framework Directive (WFD)
prescribes all water bodies to attain “good
ecological status”. In Denmark, River Basin
Management Plans (RBMP) are developed to
ensure that this goal is achieved. However,
considering coastal waters and the framework
in which they are used, the models forming the
basis for these plans contain a number of
flaws and unsupported assumptions. This may
cause one of three highly undesirable
situations:
1. Doing too little
2. Doing the wrong thing
3. Doing too much
Doing too little or doing the wrong thing both
means non-compliance with the WFD. Doing
the wrong thing or doing too much will also be
at a high cost. The RBMP have a unilateral
focus on agricultural nitrogen leaching.
Omitting all other stress factors may hamper
the process towards good ecological status,
and unnecessary expenses are likely to occur.
This practice hardly lives up to the
requirement of the WFD of “identifying
the
cost-effective and proportionate level and
combination of controls”,
and it may result in
enormous economic challenges for Danish
agriculture.
Already in 2012, the EU Commission
responded to the first RBMP recommending
that “Appropriate
methods for assessing all
potential pressures need to be developed”
(SWD, 2012). This is still not addressed, and
the RBMP still focus on nitrogen as the only
pressure for coastal waters despite
overwhelming scientific evidence that
numerous other stress factors are important.
The purpose of the present comments is to
emphasize the major problems in the
modeling work forming the basis for the
RBMP, hopefully leading the way for
management plans comprehensively targeting
all relevant stress factors directly. The overall
aim must be to live up to the WFD
requirements through RBMP based on solid
science, thus ensuring that only
necessary
and
effective
measures are imposed.
General
2
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0064.png
Landbrug & Fødevarer
General comments to the scientific documentation
Comments
Questions
The report from Aarhus University (AU) and
DHI ”Development
of models and methods to
support the establishment of Danish River
Basin Management Plans, Scientific
documentation”
is a summary of at least nine
different reports (in Danish) from the two
institutions describing the methods used in this
model work.
When evaluating scientific work, peer-
reviewed papers should be the main focus of
the evaluation and only if no such are
available, un-reviewed, published reports may
be consulted instead, considering the possible
reasons for the lack of peer-reviewed
publications.
In the case of the statistical models,
no peer-
reviewed, scientific papers have been
published
based on this work. The present
comments will in detail describe the numerous
problems in the modeling approach, thereby
creating an understanding of why publishing of
the work most likely has not been possible.
In several central points, the English report,
developed for the international evaluation,
does not include all relevant information. In
these cases we will refer to the original reports
in Danish, enclosing relevant figures or tables
in appendix.
Should eco-system models supporting RBMP build on
scientific documentation, i.e. peer-reviewed articles?
Can grey literature (reports) provide an adequate
basis for the ongoing review?
The panel is strongly encouraged to request
additional information regarding peer-reviewed
articles describing the modeling approach from both
DHI and AU.
General
1. Prologue
Comments
Questions
General
3
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0065.png
Landbrug & Fødevarer
2. Introduction
Chapter/section
Comments
Questions
The text in section 2.1 does not touch the central issue in the
WFD – returning to good ecological status. Already in 2009,
Duarte
et al.
(2009) described the problems with returning to
an earlier status. It is obvious that the knowledge of shifting
baselines is not used in this work. Marine ecological systems
are very complex and the model work and hence the RBMP
end up addressing just one stress factor; nitrogen. The
complexity as described by Duarte
et al.
(2009) is neglected.
Duarte
et al.
(2009) use an example from Odense Fjord in
Denmark to illustrate the point with shifting baselines:
It is clearly demonstrated that returning
to good ecological status is not merely
a question of reducing nitrogen loads
to previous levels.
Does the panel agree that several
stress factors must be taken into
account?
Does the panel agree that
understanding feedback mechanisms
is important in order to implement the
right measures for achieving good
ecological status?
2.1
Sample trajectory of annual means of chlorophyll
a
concentrations, as a proxy of ecosystem status, versus total nitrogen
loading of Odense Fjord that experienced significant eutrophication
followed by significant oligotrophication after management actions.
The full black symbols show the annual average values and the red
line follows the trajectory of a 5-year moving average. Initial
and final years of the time series are indicated. Insert shows the time
series and 5-year running average of total nitrogen inputs to the
ecosystems.
The figure demonstrates that following elevated nitrogen input
levels, a reduction does not result in a complete reversion of
chlorophyll
a
levels. Instead, chlorophyll
a
levels remain
elevated, even though nitrogen input has decreased
significantly.
A very important learning from this is that the ecosystem has
changed fundamentally and entered a new steady state.
Models addressing nitrogen as the only stress factor, thus, are
unable to describe the ecosystem well.
The Danish RBMP contain no answers to this challenge, as
no other stress factors than nitrogen are addressed.
This critique is consistent with feedback from the EU
Commission (SWD, 2012) presenting the following
recommendation to the first RBMP:
“Appropriate methods for assessing
all
potential pressures
need to be developed”.
The statistical models and meta models do not address other
stress factors. The more advanced mechanistic models are
4
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0066.png
Landbrug & Fødevarer
able to do so, but it has been chosen to focus on nitrogen
alone. See further remarks concerning this in section 7.
Figure 2.3, page 10 (status of chlorophyll
a)
demonstrates that
the targets for Chlorophyll
a
are scattered and random.
The reference values for chlorophyll
a
are determined based
on model calculations using a very coarse typologization. The
problems with this approach will be explained in more detail in
sections 3.2 and 8.1.
Examples of curious status levels are found e.g. in some
fjords, where the inner and outer parts are assigned different
status. This is seen for instance in Horsens and Kolding:
Abrupt changes in ecological status
between neighboring water bodies
frequently occur, as demonstrated in
the comments. Does the panel agree
that biologically it does not make sense
to see such changes?
Do the abrupt changes indicate
problems for instance with the
typologization being too coarse?
2.2
It is also demonstrated that there is a poor correlation
between the two quality elements eelgrass depth limit and
chlorophyll
a,
as can be seen e.g. by comparing Figures 2.3
and 2.4 (eelgrass depth limit) for the Northern East coast of
Jutland (see below). Eelgrass depth limit is assigned poor
status, but in the same area chlorophyll
a
is in high status.
This clearly demonstrates that other factors than chlorophyll
a
impact eelgrass depth limit.
5
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0067.png
Landbrug & Fødevarer
General
6
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0068.png
Landbrug & Fødevarer
3. Danish marine waters
Chapter/section
Comments
Questions
3.1
Many of the 119 Danish water bodies have
estuary character and therefore should have
been designated “transitional waters” according
to the WFD.
The EU commission report (SWD, 2012)
st
evaluating the 1 Danish RBMP has the
following comments on this subject:
“Transitional waters are not designated, and no
justification is given as to why this water
category has not been used. Denmark should
review its designation of at least some coastal
waters, notably those referred to as inner
coastal fjords water, and consider transitional
water designation, considering physical and
chemical factors that determine the
characteristics of transitional waters and hence
the biological population structure and
composition.”
Concerning hydromorphology in general, we
find that this issue has not been sufficiently
described. This is in line with the EU
Commission response to the first RBMP
recommending that “Denmark
needs to extend
its classification system for lakes and coastal
waters to address hydromorphological QEs
“
(SWD, 2012)
Has hydromorphology for Danish coastal waters
been sufficiently described?
Why have no Danish water bodies been designated
“transitional water”, given the description in the
WFD?
Would it be relevant to re-consider the designation of
certain water bodies – in particular the inner, coastal
fjords, as suggested by the EU commission?
3.2
The typology is central for the classification of
reference conditions and ecological status.
What is important to understand is that the
original typology description (Dahl
et al.,
2005)
is modified in this work to include only very few.
Out of all 119 Danish water bodies, 48 are
classified as one type (estuarine type 2) and 23
as another type (estuarine type 3). This means
that approximately 60 % of all Danish water
bodies are classified as only two different types.
This is a major problem because the water
bodies do have characteristics covering a much
broader spectrum.
Already in 2012, the commission (SWD, 2012)
made this recommendation to the first RBMP:
“Denmark needs to further develop water
typologies which are tested against biological
data, and develop and provide further
information on reference conditions for all water
types.”
Since 2012, the applied typology has been
simplified to a highly problematic point.
Examples of this are:
Odense Fjord at Funen (Fyn): At the figure
below, it is seen that the original typology is
named M4 for inner Odense Fjord (darkest blue
7
The European Commission has requested that
Denmark further develops water typologies.
Is it acceptable to simplify typologization to a degree
where highly different water bodies must live up to
similar environmental threshold values?
Physical modifications, such as dams and bridges,
are not taken into account in the typologization.
Does the panel agree that dams and, to some extent,
bridges may impact the exchange of water?
Only in two cases are fjords with a sluice designated
the “sluice fjord” typology.
Does the panel agree that as a basic premise, the
presence of a sluice should require an individual
assessment of the impact of the modification, and if
necessary specific threshold values for the given
fjord?
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0069.png
Landbrug & Fødevarer
color), P3 for outer Odense Fjord and P4 for the
small bay (Dalby Bugt) just east of the fjord.
These names refer to the typology as defined
by Dahl
et al.
(2005) (see Table 3.1, page 16).
However, with the current typologization, the
three water bodies belong to the same type,
type 3, and thereby the same model complex
determining chlorophyll
a
threshold values. This
is shown in Table 3.1.
The inner part of Odense Fjord receives
freshwater from a catchment half the size of the
island of Funen (Fyn) and is, thus, naturally
exposed to high nutrient levels. Dalby Bugt, on
the other hand, receives freshwater from only a
small catchment and given the large opening
towards open water, water exchange is rapid.
Because of the simplification, all three water
bodies, however, share the same chlorophyll
a
threshold value of 3.6 µg / l
This of course results in big differences in
status of the water bodies, ranging between
high and poor status as shown in the figure
below.
8
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0070.png
Landbrug & Fødevarer
Same goal
(chlorophyll
a:
3.6 µg / l)
Nyborg Fjord at the east cost of Funen (Fyn):
As seen below, the original typologies are M3
and P3, but in the simplified version used in the
RBMP, both are classified as type 3 (Table
3.1).
The figure illustrates how big the difference of
the water bodies is, as the M3 water body is
almost closed behind a dam, while the P3 water
body has an open boarder to Great Belt, an
area with strong currents and, hence, rapid
water exchange. Again, the threshold between
good and moderate status is 3.6 µg / l for both.
This means that status changes from bad in the
inner part to high in the outer, open, part of the
fjord, as seen below.
9
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0071.png
Landbrug & Fødevarer
The dam is clearly seen in the photo below, and
the existence of the dam likely explains much of
the problems using the same, type 3, typology:
In addition to the simplification of typologies,
specific choices during typologization are
difficult to understand. Most obviously, when a
sluice, type 5, is present, then why is the water
body in question not classified as type 5?
An example of this is Norsminde Fjord, which is
classified as type 3. However, a sluice is
obviously present, as seen in the photo below.
General
10
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0072.png
Landbrug & Fødevarer
4. Danish monitoring data DNAMAP (NOVANA)
Chapter/section
Comments
Questions
The monitoring carried out in Danish marine
waters does not cover all water bodies with
specific nitrogen reduction targets. Only very
few typologies are applied to the 119 Danish
marine water bodies (see section 3.2).
Considering the extensive use of models, does the
panel find the ongoing monitoring program sufficient?
4.1
Meta models are used when modeling data is
insufficient. As meta models are developed in
different water bodies than where applied, they often
It would be of interest to the panel to be
presented with an overview of data categories produce result of high uncertainty. Should the
monitoring program be extended in order to reduce
and the extent of monitoring to compare with
the original typology (Dahl, 2005) as described the use of meta models?
in section 3.2
In addition, the monitoring of seagrasses has
not been presented. Being a key indicator of
the WFD, eelgrass monitoring is of high
importance. Traditionally, monitoring has
mainly been focusing on depth limit instead of
area cover. It could, therefore, be of interest to
the panel to be presented with the actual
seagrass monitoring program.
Nitrogen loading, on an annual basis, is the
See questions regarding this point in section 9.1
target of action in the Danish RBMP.
A new analysis carried out in Karrebæk Fjord
indicates that this approach is problematic for
water bodies with low residence time. The new
analysis demonstrates that winter and spring
land-based loading of nitrogen only play a
minor role for the summer (May - September)
chlorophyll
a
concentration due to the low
residence time for water in Karrebæk Fjord. In
winter, 90 % of the water has been exchanged
within 10-16 days (see Appendix 1).
These results are of great importance to all
water bodies with little or medium residence
time. Instead of focusing on annual land-
based nitrogen loads, focusing on loads at
relevant points in time may have a dramatic
impact on the required load reductions.
Figure 4.4, where annual loads (kg N / water
body ha / year) are presented, will then not
give a relevant picture. Generally, water
bodies with high loads have little or medium
residence time because it is small water areas
compared with the catchment.
4.2
4.3
11
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0073.png
Landbrug & Fødevarer
General
The monitoring program has been decreased
at the same time as the government
introduced different reduction targets for each
water body. In total, Denmark has 119 water
bodies more or less affected by Danish
nutrient loads. However, the monitoring
program does not cover all water bodies, and
even very ambitious reduction targets are
introduced in water bodies with no monitoring
program.
This is highly problematic, because without
data there would be no models. This problem
is solved by introducing meta models;
however, as will be elaborated in section 8.6,
this approach introduces other problems.
12
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0074.png
Landbrug & Fødevarer
5. Overview of WFD tool development in a Danish context
Chapter/section
Comments
Questions
5.1
It is explained that meta models are used for
Have scientific criteria for identification of ‘too small’
“too small” water bodies and when data
water bodies been established / provided?
availability is limited.
As described in the original reports in Danish,
also a third way leads to meta models: When
nitrogen load was not selected by the MLR-
method as input variable for a given statistical
model. In these cases, the actual model was
discarded and replaced by a meta model. (See
Appendix 3, Table 6, comment marked with *).
This will be further commented in section 8.6.
The WFD operates with three “biological
quality elements”, of which angiosperm
distribution is one. In the Danish context,
eelgrass depth limit is used to describe the
angiosperm distribution and is intercalibrated
as such. However, eelgrass is not the only
angiosperm to be found in Danish, marine
waters, and the substitution is therefore
unacceptable.
Shallow areas with a healthy angiosperm
distribution fulfilling the requirements of “good
ecological status” may be assigned with an
inferior ecological status, if eelgrass is not the
dominating species. An example of this is
Ringkøbing Fiord, which has a healthy
population of other angiosperms than eelgrass
due to brackish water.
However, using Kd as an indicator for eelgrass
depth limit, a nitrogen reduction demand of 75
% has been assigned this area. According to
the procedure described in section 8.3, and
our comments to the same, this refers to a
calculated reduction demand above 200 %.
This is in contrast to the fact that other
species, such as spiral tasselweed (Ruppia
cirrhosa)
and sago pondweed (Potamogeton
pectinatus),
already not only cover huge areas
of shallow water but also reach the depth limit
set for good ecological status.
Another example is found in a specific area in
Limfjorden, where approximately 80 % of the
area is covered with eelgrass. However, the
eelgrass does not reach a specific depth in the
narrow channel through the area. Because of
this, the area is designated poor ecological
status in spite of the widespread distribution of
eelgrass.
The example is illustrated in the figures below.
Is it acceptable to disregard species of angiosperms
other than eelgrass, e.g. spiral tasselweed, even
though specific areas have abundant populations of
these?
Is it reasonable to assign poor ecological status,
concerning “other aquatic flora”, to areas with a
widespread eelgrass population, but where eelgrass
is not found at the bottom of e.g. an artificial channel
in the fjord?
5.2
13
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0075.png
Landbrug & Fødevarer
Kd is a physico-chemical quality element;
basically the transparency of the water.
Even though light is indispensable for growth
of eelgrass, lack of light is very far from being
the only reason that eelgrass populations may
not increase. The Danish research projects
REELGRASS and NOVAGRASS (ongoing)
and a large amount of peer-reviewed papers
have for the last 15 years pointed at very
specific stress factors, which have to be
addressed for eelgrass to recover.
In the Danish context, important stress factors
in addition to nutrient load are:
-
-
-
Organic matter
High organic content in sediment
(inner fjords)
European green crab. International
literature from the last five years
describes how these crabs prevent
recovery or even cause decline in
seagrass cover
Sand transport has been a great
stress factor in many Danish waters
preventing eelgrass recovery since a
massive decline in the 1930ies,
caused by disease.
Numerous factors influence the eelgrass depth limit.
Is it reasonable to focus exclusively on Kd as a
proxy?
Is it possible, maybe even likely, that eelgrass will not,
even after a time lag, spread to the required depth
limit if only one stress factor is addressed?
-
Eelgrass in itself is also a measure to increase
water transparency and thereby increase
depth limit for eelgrass or other seagrasses.
14
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0076.png
Landbrug & Fødevarer
At the east coast of the USA, a recovery of
eelgrass beds of more than 2000 ha have
been made by harvesting and sowing eelgrass
seeds. This recovery has had a positive impact
on the ecosystem and has lowered the
turbidity dramatically (see figure below) along
with the nutrient concentration in water.
Relation between turbidity and area of
eelgrass (ha) recovered by human restauration
efforts. From Orth
et al.
(2012).
The way to recover eelgrass and to have
better ecosystem functioning is therefore not
only a question of reducing nitrogen but to
address the interaction between land loads
and recovery in coastal waters.
General
15
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0077.png
Landbrug & Fødevarer
6. Statistical model development
Chapter/section
Comments
Questions
The cited ”earlier work on MLR” (Markager
et
al.
2006, 2008) are non-peer reviewed reports.
Thus, both previous work and the present
reports have not passed a scientific peer-
review.
6.1
Peer-reviewed papers are the only acceptable
reference for any scientific work. For the
statistical models in the present report, and
similar models in previous reports, no peer-
reviewed papers have been published.
It must be questioned why no publishing has
taken place.
According to this section, PLS models were
developed “with
the main purpose of
quantifying the relationship between nutrient
loadings and the selected response variables”.
Thereby presuming that there is a direct
relationship between them, unaffected by
other factors. This approach is repeated in
section 6.3.
Does the panel agree that selecting input variables in
advance is a problematic approach, which is
unnecessary given the many advantages of PLS
regression?
6.2
As commented in Appendix 2, the main
problem of selecting factors in advance is that
some factors might be overlooked and thus
not included in the model and, hence, in the
conclusions.
A PLS model will highlight the specific factors
that are important to the model, thereby
allowing a quantification of their relationship.
Selecting factors in advance means that the
model will be biased towards the selected
factors.
“Four responding variables (…) were chosen
as environmental indicators due to their well-
documented response to nutrient enrichment”.
The panel is kindly reminded that the WFD
aims at improving the ecological status of
water bodies, not at reducing nutrient inputs.
Is choosing responding variables based on which
factors they respond to an acceptable method in
accordance with scientific standards?
Should non-Danish contributions to the total nutrient
load in Danish marine waters be taken into account
when developing regression models describing the
ecosystems in these waters?
6.3
Selection of predictor variables is described in
section 6.3.1 (page 29), and it is stated that
the purpose of the regression models is to
“quantify
the relationship between the
responding variable and the predictor
variables especially the nutrient loading which
can be managed”.
At no point do the authors address the fact
that only part of the nutrients in Danish coastal
waters derives from Danish land-based
nutrient load. This is a serious point of critique,
as not accounting for all nutrient input, along
16
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0078.png
Landbrug & Fødevarer
with other stress factors, will reduce the
model’s ability to explain the response
variable.
Should the trend in climate change be included in the
According to equation 6.2, detrending was
model work?
used for preprocessing data. The exact
settings for the detrending are not stated in the
paper, why the strength of the detrending is
unknown to the reader.
However, as the climate has changed
significantly since the reference year (1900),
as described later in section 9.1.3, it is curious
that a “normal climate” is used for calculations.
As commented in Appendix 2, it is unusual to
use MLR for variable selection, followed by
PLS for the actual modeling. MLR is unable to
handle correlated variables, as it is actually
noted on page 36. PLS on the other hand is
perfectly able to handle correlated variables,
and it is therefore curious why MLR is used at
any point.
Imagine two (or more) correlated predictor
variables, which describe a given factor in a
quantitative manner through their correlation,
and only through their correlation. By
excluding certain factors beforehand, such a
correlation and the relationship to the
responding variable will not be found.
Several variable selection methods are
available for PLS, easily accessible in the
specific PLS program package from
Eigenvector® that was used in the modeling.
Specific comment to page 36: “we
experienced that the parameters (PLS
coefficients) were still sensitive to small
variations in the data set when highly
intercorrelated predictors (r > 0.9) were used,
making use of highly correlated data sets
problematic even in PLS regressions”.
As already described, PLS is able to handle
intercorrelated predictors very well. Excluding
correlated variables therefore means that
important information could be lost. It is
unclear in which way highly correlated data
sets turned out “problematic”.
This is elaborated in Appendix 2.
It is noted that the Kd models do not describe
data very well. This is explained by influence
of light absorption by dissolved organic matter
and detritus as well as scattering of light by
particles. This clearly shows that the wanted
Kd level cannot be obtained by acting on
nitrogen load alone.
When the ecological status of Kd is determined by
several factors in addition to N loading, is it then
scientifically correct to investigate and address only N
loading?
Is good ecological status obtainable when other
variables of significance are not addressed?
Does the panel agree that omitting intercorrelated
variables, which are very well handled by PLS
regression, might mean that important information is
lost?
Has variable selection been carried out in a
satisfactory way?
Could important information potentially be lost through
the applied procedures, specifically the use of MLR
for variable selection before PLS modeling?
6.4
17
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0079.png
Landbrug & Fødevarer
When assessing the target nitrogen loads, and Is it problematic to extrapolate a correlation far out of
thereby the required nutrient reduction, the
its defined range, as is done in the statistical model
developed models are extrapolated down to
approach?
the desired nitrogen load level. This
extrapolation is highly questionable as in most
cases it will reach far out of the defined range
of the correlation.
It is argued that due to substantial year-to-year
variations, the lowest values in the
extrapolations are nearly encompassed by the
models.
However, this point is made based on national
data. Extrapolation of correlations to
determine required nutrient reductions are
carried out for each model, and as required
reductions frequently exceed 100 % (Appendix
3, Tables 6 and 7 from the original report); this
is hardly within the defined range of the
correlation.
6.5
General
It is unfortunately necessary to inform the
panel that the described approach of modeling
seems to have changed from the original
reports, in Danish, to the English report
forming the basis for the present evaluation.
A phrasing such as
“…we have assumed that nutrient loadings do
have an impact on the selected response
variables and we therefore designed the
method to provide the
most likely coefficient
for this response” (section 6.5)
is in contrast to page 7 of the original report
no. 3 (translated):
“The main principle in developing the
statistical models is selecting the explanatory
variables (nutrient loading, climate etc.), which
best describe a given indicator (i.e. chlorophyll
a,
Kd, TN and TP).”
The panel is kindly requested to reflect and comment
on the evaluation report differing from the original,
Danish reports, which form the basis for RBMP.
What is the value of an international evaluation, if the
background reports have been altered at critical
points?
18
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0080.png
Landbrug & Fødevarer
7. Mechanistic model development
Chapter/section
Comments
Questions
7.1
7.2
7.3
7.4
The calibration and validation of mechanistic
models, like that of statistical, is crucial.
If the models do not react like expected it is
often because the ecosystem is not fully
understood and action should be taken to fully
understand the processes of the ecosystem to
highlight the important pressures.
It does not seem that the mechanistic models are
used for studying all relevant aspects of the
ecosystem. Would it have been relevant to use
mechanistic models for analyzing other scenarios
than reducing nitrogen and phosphorus?
General
Different aspects of the various water bodies’
ecosystems could have been investigated by
using the models. One example is oxygen
depletion, which is a major problem for the
central Limfjord. Oxygen depletion is linked
very closely to stratification and warm summer
days with no wind. Solutions for this could
have been highlighted with different scenarios
but have not been done. Only reducing
nitrogen from land has been proposed.
As already described, eelgrass cover has
important feedback mechanisms in the
ecosystem, including increased nutrients
uptake and decreased resuspension.
Measures such as eelgrass restauration and
change in sluice practice could therefore be
included in scenarios using the mechanistic
models, and knowledge on the effect of this
could be obtained.
Why are only scenarios of nitrogen and phosphorus
load reductions included in the modeling work?
Alternative scenarios, focusing on different stress
factors, would support the work towards finding the
most promising solutions.
19
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0081.png
Landbrug & Fødevarer
8. Model application
Chapter/secti
on
Comments
Questions
The chlorophyll
a
reference value is central in the
modeling work. Finding a value for that can be
approached in several ways. As stated in the
report, no data was available for the reference year,
1900, and hence, a model approach was chosen,
using models correlating nitrogen and chlorophyll
a.
Using a model describing the relationship between
nitrogen load and chlorophyll
a
and using this
relation to determine the reference value means
that determining the threshold value of chlorophyll
a
(moderate to good status) becomes solely
dependent of nitrogen. Thus; all other stress factors
are deemed irrelevant with no reference to the
biological complexity.
In this work, water body typology plays a central
role. It is stated on page 71 that:
“In order to reduce (some of) the uncertainties, we
have applied a typological approach where site-
specific model results were used to establish robust
type-specific reference and target values
transferable to Danish water bodies.“
What is stated is not the case. The typology is far
too simple, as explained in section 3.2. Ending up
with only two different type specific reference
values covering 71 water bodies out of 119 is a
major problem.
Threshold values for open waters are presented in
Table 8.6. It is seen that the “type-specific GM
target value” varies between 1.5 and 1.9 µg / l.
Two issues should be noted here:
1)
Only very minor changes in the chlorophyll
a
target value make a big difference in the
nitrogen reduction target because the
response curves are almost flat. The figure
below shows the correlation between
Danish nitrogen loads (X-axis) and
chlorophyll
a
(Y-axis). It is clear that
changing the target value from 1.5 to 1.9
would mean significant changes in required
nitrogen reductions. The weak response in
chlorophyll level to reductions in nitrogen
load indicates that chlorophyll
concentrations are likely to depend more
significantly on other factors than on
Danish land-based nitrogen loads
.
Does the panel agree that the almost flat response
curves describing the correlation between nitrogen
load and chlorophyll
a
result in large uncertainties
on the estimated nitrogen load reductions?
Does the panel find that the certainty of the
reference load in 1900 has been satisfactorily
accounted for?
And does the panel find that it falls within an
acceptable range?
8.1
20
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0082.png
Landbrug & Fødevarer
The correlation between chlorophyll concentration
(µg / l) and nitrogen load (%) in Lillebælt. Figures
covering all of Denmark are available at:
http://vandplan.dhi.dk
2) In some cases the “type-specific GM target
value” seems to decrease from open water
towards land. This is opposite of what
would be expected. For example, the
threshold value is 1.6 µg / l at the islands in
the middle of Kattegat (Læsø, Anholt) while
it is 1.5 µg / l closer to the shore and
thereby closer to Danish land-based loads.
Se threshold values indicated on the map
below.
21
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0083.png
Landbrug & Fødevarer
8.2
Quoting the report page 90: “…this implies a
restriction to indicators for which a reference
condition and an EQR value for good-moderate
status have been established.”
It is important to notice that the
mechanistic
models
are restricted exactly to such indicators, namely
chlorophyll
a
and Kd. The statistical models, on the
other hand, include the four “supporting indicators”
as described. None of these four indicators have a
defined reference condition or EQR value, and
there are no scientific references indicating any
previous usage of this approach anywhere.
Can the two model approaches be compared
directly, given that the statistical modeling
approach requires the inclusion of four supporting
indicators, whereas the mechanistic approach does
not?
The one-out all-out principle means that the
Does the panel find that using a weighted average
ecological status of a water body is governed by the is in acceptable compliance with the WFD?
biological quality element
of lowest status.
Comparing the one-out all-out principle with the
weighting approach used for the statistical models
therefore makes little sense.
In addition, the supporting indicators are used in a
highly questionable way, which will be elaborated
below.
The indicator “chlorophyll
a-concentration”
It is reported that 17 out of 28 chlorophyll
a
models
have a significant nitrogen coefficient. In Table 6.3,
which presents the various models, only 24
chlorophyll
a
models are presented, of which 16
include nitrogen load as a predictor variable, six
include phosphorus (one has both N and P) and
three have neither nitrogen nor phosphorus as a
predictor variable. Thus, only 67 % of the presented
models have nitrogen as the first selected predictor
variable, why focusing exclusively on reducing
nitrogen loads to obtain good ecological status
seems like a curious choice.
88 % of the models have either nitrogen or
phosphorus as a predictor variable – not 93 % as
noted in section 8.3.
It is noted that percent load reductions (PLR) range
between zero and 134 % for the chlorophyll
a
indicator, and that values above 100 % are most
frequently found in open areas of Kattegat and the
Belt Sea, where statistical models are not used.
This is, however, not the case. Enclosed, as
Appendix 3, are Table 6 and Table 7 from the
original, Danish reports showing the occurrence of
values above 100 %.
Values above 100 % frequently occur, definitely
also in closed, coastal water bodies, such as Stege
Nor and Holckenhavn Fjord, to mention just two
examples. It occurs most frequently and with the
highest numbers, but not only, for water bodies
where meta models have been used for assessing
22
When only 67 % of the developed models have
nitrogen as a predictor variable, is it reasonable to
focus exclusively on nitrogen regulation?
Or should other factors be taken into account?
Percent load reductions above 100 % frequently
occur for models on the chlorophyll
a
indicator. Not
only in open waters, also indeed in closed fjords.
Numbers as high as 135 % (Haderslev Fjord) are
included in the weighted average to give the final
PLR.
Is including unrealistic model results in further
calculations acceptable, scientific practice?
Or should it be considered that maybe the model is
not optimal if yielding unrealistic results?
8.3
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0084.png
Landbrug & Fødevarer
PLR.
Is it reasonable to link nitrogen load and
angiosperm distribution directly, without
The problem with eelgrass being the only considering other stress factors?
angiosperm included in the Danish RBMP has been
elaborated in section 5.2, but it is likewise relevant
Is it acceptable scientific practice to replace clearly
when discussing the indicator “light attenuation”.
erroneous results with values that are
chosen,
A direct correlation between nitrogen load and Kd is based on no scientific evidence or calculations?
assumed. The correlation is, however, not that
simple. The most important elements determining
light attenuation are particles, dissolved matter and
phytoplankton, of which only the latter is to some
extent related to nitrogen load.
The indicator “light attenuation”
From 1989 to 2013, nitrogen load in coastal areas in
Denmark was reduced by 50 %; however, Secchi
depths (and thereby Kd) remained unaltered
(Hansen, 2015), demonstrating a lack of correlation.
It is also stated in the modeling report that eelgrass
will not necessarily reach the depth limit even if
nitrogen load is reduced as requested, as re-
inhabitation is delayed. When the eelgrass is lost,
the physical conditions at the bottom may change
and thereby prevent that re-inhabitation takes place
at all; the habitat is lost (Flindt
et al.
2011).
Complementary actions are not addressed nor
included in the RBMP.
A “transformation” of data is described, “to overcome
the effects of this time lag on the estimated load
reductions”. However, what is actually done, as seen
in Table 8.7 on page 101, is that calculated values
are changed into arbitrarily selected values.
If PLR is between 25 and 100 %, it is changed into
25 %. 100 – 200 % is changed into 50 %, and
calculated values above 200 % are presented as 75
%. By consulting the tables in Appendix 3, it is clear
that calculated PLR values above 200 % are very
frequent!
In other words: The developed Kd models calculate
nitrogen load reductions of up to more than twice the
present nitrogen load. To make up for these
impossible results,
calculated
values are replaced by
chosen
values.
Occurrence of Hypoxia / Ecological Signs of
Hypoxia
Occurrence of low oxygen conditions, or ecological
signs of the same, is directly translated into a
demand of 25 % reduction of total nitrogen (TN),
which is again translated into nitrogen reduction
demand by using a TN model (page 96 of DHI/DCE
report). No calculations of any form are carried out
to assess the 25 %, and there is no argumentation
or any references for this choice. In the Danish
reports, the choice is explained by saying that “it
has to be sufficiently high to move the system”
23
Is it common, scientific practice to simply choose a
nitrogen reduction demand?
Is it acceptable to base regulation on numbers
chosen without any scientific basis, calculations or
references?
Could the TN reduction demand just as well have
been 20 %? 30 %? Or 15 %?
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0085.png
Landbrug & Fødevarer
(report no. 3, “Statistiske modeller og metoder til
bestemmelse af indsatsbehov”, DCE 2015, page
18), with no reference to locations where different
levels of nitrogen reduction have been tested.
Nitrogen limitation of phytoplankton growth
Again, please be advised that this indicator is not
used in the calculations using mechanistic models.
As for Kd models, the calculated values are
changed according Table 8.7, though the change is
less dramatic than for Kd.
Weights – as noted in Table 8.7
According to the table, chlorophyll and Kd model
results are each given the weight 2, “occurrence of
hypoxia” and “N limitation” are each given the
weight 1, and the two “ecological signs of hypoxia”-
indicators each have the weight 0.5.
For most water bodies, chlorophyll and Kd model
results, thus, make up 4/7 = 57 % of the final result.
When using meta models, only chlorophyll, Kd and
“occurrence of hypoxia” are included, using the
same weights. This means that for these water
bodies, chlorophyll and Kd make up 4/5 = 80 % of
the final result.
The table below demonstrates how the supporting
indicators affect the final results of the statistical
model approach by making up part of the weighted
average. The table shows weighted averages with
and without the supporting indicators.
Is it acceptable to include supporting indicators
which, in almost all cases, lead to lower required
reductions – in the statistical and not in the
mechanistic models?
Is it acceptable to include four supporting indicators
which, in almost all cases, lead to lower required
reductions in the designated statistical models -
and including only one supporting indicator in the
meta models?
24
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0086.png
Landbrug & Fødevarer
It is obvious that including the supporting indicators
in most cases reduces the final PLR substantially.
8.4
8.5
It is noted in this section that “…statistical models
Is it acceptable to describe a modeling approach as
are “black-box” models with a direct link to
a “black-box” approach, when in fact input
observations but without any descriptions of causal variables to some extent are selected in advance?
links”
This, unfortunately, is not consistent with the actual
modeling approach. Discarding models when
nitrogen is not selected as an input variable (see
Appendix 3, comments to Table 7 marked with *) is
a choice, not something that happens in a “black
box”.
Likewise, the “black box” – approach also seems
inconsistent with the statement in section 6.2 that
the PLS models were developed “with
the main
purpose of quantifying the relationship between
nutrient loadings and the selected response
variables”
In other words, focusing exclusively on nutrient
loadings was a choice made in advance of the
actual modeling
25
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0087.png
Landbrug & Fødevarer
Meta models are used for water bodies where no
mechanistic or statistical model has been developed,
for various reasons. However, for the statistical
models, meta models are also used, if nitrogen load
was not selected as an input variable. This is not
clearly described in the English text, but by
consulting Appendix 3, it is seen that for nine out of
the 28 water bodies (32 %) presented in the pictured
Table 6 from the original report, nitrogen load
reduction based on chlorophyll was calculated using
a meta model, because nitrogen load was not
selected as an input variable in the original model.
The same is the case for eight of the Kd models, still
out of 28 (equal to 29 %).
Is it acceptable scientific procedure to omit results
that differ from the expected? In this case meaning
when nitrogen load is not selected as an input
variable.
The idea in meta models is to apply models from Would a more differentiated typologization possibly
different water bodies to a water body of the same improve the applicability of meta models?
type. A great part of the problem with meta models
is, thus, the rough typologization, assigning highly
different water bodies to the same type, as explained
in section
Fejl! Henvisningskilde ikke fundet..
As discussed in section 8.3, nitrogen load reduction
demands frequently surpass 100 %. This problem is
especially relevant when using meta models, as can
be seen in the pictured Table 7 in Appendix 3.
This supports the arguments that the criteria for
using meta models are not acceptable and that meta
models do not describe the water bodies, where they
are applied, sufficiently well.
8.6
One specific example of the implications of
problematic use of meta models is Stege Nor, a
small water body with very limited opening towards
open water. A satellite image of Stege Nor is
presented in Appendix 4.
A final nitrogen load reduction demand of 77 % is
calculated for Stege Nor using meta models, without
addressing the 122 % reduction based on
chlorophyll as part of the average. In a Natura 2000
report (Naturstyrelsen, 2013), Stege Nor is
described as a water body with a healthy vegetation
and beds with eelgrass in the deepest parts. Large
specimens of the pollution sensitive stoneworts
(Charophyceae) are found in smaller beds around
the cove, and the fauna in and around the cove is
described as “well developed”. In other words,
nitrogen load reductions of 77 % are demanded in
the catchment area, even though the water body is
described as healthy and with a thriving flora and
fauna.
A most likely explanation for the special case of
Stege Nor is a deviating value in reference
chlorophyll concentrations. As seen in Appendix 4,
summer chlorophyll concentrations are generally
below 5 µg/l. However, in 2011 as much as 30 µg/l
Is it reasonable to include a measured value so
clearly deviating from the general level?
Should it be expected that input data for models of
this type are comprehensively screened for outliers?
26
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0088.png
Landbrug & Fødevarer
was reported.
8.7
8.8
As a comment to the calculations of model
uncertainty, it is important to know that the
calculations presented here, to the international
panel, are completely different from the original
calculations which were presented to Danish
politicians and the public. The original calculations
are erroneous and have been strongly criticized.
Researchers from the Technical University of
Denmark (DTU) have made a thorough description
of these problems. This has been translated into
English and is enclosed with these comments as
Appendix 5.
The DTU researchers suggested an alternative
calculation, which has been presented to Aarhus
University, DCE. This calculation, also found in
Appendix 5, is what is shown in the present report
for the international evaluation. The DTU
researchers have not been quoted for the
suggestion.
The panel is kindly advised to take note of this and
include in the final report that the original
calculations are erroneous and based on
problematic assumptions, whereas the calculations
presented here derive from the specific DTU paper
presented in Appendix 5.
The analysis of variance results in a minimum
confidence interval of ± 13.3 %-points. Thus, for
three out of the 11 water bodies in question, no
required load reduction has been demonstrated, as
the mean reduction is less than 13.3 %.
It is important to note that the ± 13.3 %-points are a
minimum
uncertainty, and no maximum uncertainty
can be calculated.
On page 130 it is stated that “the
assumption of
independency might not be fulfilled. Especially the
neighbouring water bodies (water body 156 and
157) might be correlated”.
It should be very clear that neighboring water
bodies in all cases will be correlated, and the non-
independence of water bodies is the reason why no
maximum uncertainty can be determined.
Given that neighboring water bodies are definitely
correlated, does the panel find that a confidence
interval of ± 13.3 %-points
based on an assumption
of independence
provides useful information on the
actual uncertainty?
See further questions in section 8, General
27
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0089.png
Landbrug & Fødevarer
Quantification of model uncertainty
The presented analysis of variance shows, by a
very narrow margin (P = 0.06), no significant
difference between required nitrogen load
reductions calculated by mechanistic and statistical
models.
However, as previously discussed, the final result
using mechanistic models is an average between
results based on a chlorophyll model and on a Kd
model. For statistical modeling, one or three further
indicators are taken into account in the average.
Instead of comparing final results, it is therefore
relevant to compare results based on chlorophyll
and Kd models, respectively. The calculations are
presented in detail in Appendix 6.
In brief, percent required load reduction based on
chlorophyll models were compared using a two-way
analysis of variance (carried out in Excel). The
outcome was a P-value of 0.038 for the effect of
model type, i.e. model results differ significantly
between mechanistic and statistical models.
For Kd, a similar calculation resulted in P = 0.05,
meaning the models are just significantly different.
However, if inserting the lowest value in the interval
of
actual
model results, i.e. the results before
assignment of new values (as described in section
8.3), the result is P = 0.026.
Thus, results based on Kd models also differ
significantly between mechanistic and statistical
models.
Based on these results, the two-way analysis of
variance cannot be reduced to a one-way analysis
of variance as in the presented report.
All calculations of uncertainty are based on the final
results concerning required nitrogen load
reductions. It is shown that the available data are
insufficient for determination of the maximum
confidence interval.
Furthermore, no investigations are carried out
concerning uncertainty of model input. Throughout
both statistical and mechanistic modeling
procedures, assumptions, choices and calculations
add uncertainty to the final results. This is further
elaborated in Appendix 5.
No attempt is made at quantifying these
uncertainties at any point.
In effect this means that the actual uncertainty of
the final model results is largely unknown.
The panel is kindly requested to comment on the
statistically significant differences between model
results using the mechanistic and the statistical
approaches, respectively.
No attempt is made to quantify the uncertainties
arising from model input and through modeling
procedures, using both mechanistic and statistical
approaches.
Does the panel agree that a solid assessment of
uncertainties of the models is missing?
General
28
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0090.png
Landbrug & Fødevarer
9. Discussion
Chapter/section
Comments
Questions
The WFD requests that each member state
ensures “a review of the impact of human
activity on the status of surface waters” (article
5 (1)). In the Danish RBMP, no thorough
review of all relevant stress factors was
performed, and N is the only stress factor
addressed.
However, a thorough cumulative impact
assessment of stressors caused by human
activity in Danish marine waters has recently
been carried out by NIVA Denmark. This
analysis will be published before September
2017, and it is attached here in full as
Appendix 7. A scientific paper based on the
report has been drafted and will be submitted
to Journal of Marine Systems before
st
September 1 , 2017, for consideration, peer-
review and publication.
The analysis is based on a data set including
35 human stressors and their impact on 47
ecosystem components.
The human stressors are pooled in eight
groups for easier interpretation. In the figure
below (on page 86 of Appendix 7) an overview
is provided for stressor contributions in
percent of total cumulative impact in the WFD
area. Effects of climate change have not been
included in this figure.
Locally, the distribution of stressors may differ
widely, as elaborated in the report. Stressor
impact distribution along transects going from
open water to the bottom of fjords is presented
in the report for more detailed insight.
The goal of the RBMP is to obtain good ecological
status, as stated in the WFD, not to reduce nutrient
loads. Should other stress factors than nitrogen load
therefore be taken into account in the RBMP?
Is it realistic that acting solely on a single stress factor
will be the best way to attain good ecological status
for all required elements?
Is it possible that if acting only on a single stress
factor, the need to reduce impact from this will be
higher than by using a combined effort on several
stress factors?
The WFD has a requirement of applying a cost-
effective approach. When leaving out clearly relevant
stress factors from the modeling, can it be claimed
that the RBMP live up to this requirement?
9.1
Figure A6.2 in Appendix 7
The figure shows that N is responsible for
approximately 30 % of the total pressure on
Danish coastal waters. It is, thus, aggravating
that 100 % of the measures are directed
towards this single stress factor.
Non-indigenous species and their impact on
marine ecology so far have not been
considered at all in Danish RBMP. A species
such as the round goby (Neogobius
29
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0091.png
Landbrug & Fødevarer
melanostomus)
is already established in
several Danish water bodies. It feeds,
amongst other things, on mussels. Areas with
dense populations of the round goby may
experience a sharp decline in mussel
population leading directly to high levels of
chlorophyll in the water, as filtration will be at a
minimum. Thus, the effect of non-indigenous
species may be misinterpreted as an effect of
land-based nitrogen loads.
With the present modeling approach, non-
indigenous species will directly affect both Kd
and chlorophyll models. But a great part of the
actual problem will not be addressed.
The stressor groups shipping, noise, fishery
and physical modifications are of lower
importance; however, locally they may have
significant influence on the marine
environment (Appendix 7, page 26-27). The
enclosed report (Appendix 7) therefore
provides substantial evidence that focusing
exclusively on nitrogen as a stress factor and
directing all measures towards nitrogen is an
insufficient approach to obtaining good
environmental status in Danish coastal waters.
The impact of future climate changes is briefly
Does the panel find that climate change can be
discussed, and climate changes in the form of
omitted when estimating which ecological status can
increased temperature and precipitation since
be obtained in Danish coastal waters?
1875 are mentioned. It is noted that these
changes have not been taken into account in
the modeling work.
Please be advised that effects of climate
changes since the reference year (1900) have
been actively removed from the statistical
models, as discussed in comments to section
6.2.
The thorough analysis carried out in Appendix
7 includes climate change on an overall level,
but not in the local analyses along the
described transects.
It is shown (Figure 3.6, page 23) that climate
change makes up as much as 15 % of the
total stress of the Danish marine environment.
The effects of climate change, hence, cannot
be overlooked in RBMP, and actively
removing the effect will seriously deteriorate
the quality of the models.
30
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0092.png
Landbrug & Fødevarer
The hydraulic residence time in a water body
is of great significance to the biological effect
of nutrients released into the water. This is not
taken into account in the Danish RBMP.
Timing of nutrient release is yet another factor
that is not included. The growing season for
algae is from spring to fall (March –
September). The main part of nitrogen loss
from fields takes place during winter months.
No attention has been given to time periods of
nitrogen loads in mechanistic nor in statistical
models of the RBMP. The mechanistic models
simply use the annual nitrogen load. For the
statistical models, various time periods were
tested. However, it was not investigated which
time periods ended up being included in the
models.
Apparently, a list of selected time periods was
made recently; however, this has not been
released. Hence, the reader is not informed
whether the selected time periods appear
meaningful in a biological sense.
Karrebæk Fiord is a shallow and relatively
closed lagoon with a hydraulic residence time
of about two weeks. Most of the nutrients
entering the fjord during winter months are,
thus, likely to have left again before the onset
of the growth season for algae and
phytoplankton. This notion was investigated
thoroughly by DHI using mechanistic
modeling. The report describing the analysis is
attached as Appendix 1.
Four scenarios were tested. One scenario
reducing N load with the amount required in
the RBMP, reductions distributed evenly over
the year. This scenario was compared to three
other scenarios with reductions concentrated
in winter months, late winter and in summer
months. The conclusion is that the impact of
winter reductions on summer chlorophyll and
Kd is negligible. Reductions in late winter
likewise have minor impact on summer
chlorophyll and Kd.
Reducing nitrogen load in summer months, on
the other hand, directly lead to improvements
in chlorophyll and Kd. Reducing nitrogen load
at the correct time also means that the
required reductions are much smaller than
when reductions are distributed over the year.
Hence, taking period of nutrient loss and time
of algal growth season into account is crucial
in order to obtain the optimal effect of nutrient
load reductions.
It is known that the nitrogen lost during winter months
in many water bodies with short residence time will be
gone (washed to sea) before the onset of the algal
growing season. Based on this, should timing of
nitrogen reductions be included in the modeling work?
Various time periods are selected for input variables
in the statistical models, but the periods are not
included in the public reports. Should this information
be included in order to evaluate the models better
from a biological perspective?
31
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0093.png
Landbrug & Fødevarer
Chlorophyll
a
targets
It is described how reference values are
determined according to type of water bodies,
instead of based on the individual water
bodies.
As discussed in section 3.2, the applied
typology is very coarse, with the majority of
Danish coastal water bodies classified in only
two types, and hence, given only two different
chlorophyll
a
targets. This leads to a situation
where type-specific reference values are
hardly more precise than site-specific values
9.2
Do more data lead to more accurately determined
reference values for a specific water body, if the data
derive from widely different water bodies assigned to
the same type?
Chlorophyll
a
as indicator
The discussion mentions high grazing
pressure and high density of benthic filter
feeders as cases where chlorophyll
a
levels do
not increase in spite of high nutrient loads.
It is curious that the authors do not mention
the opposite cases, e.g. when benthic filter
feeders are almost absent because of
stressors different from land-based nutrient
loads. This is discussed in more detail in
section 9.1 and in the report attached as
Appendix 7.
The statistical model approach is again
described as built solely on monitoring data
“without including any process descriptions or
mechanisms”. The panel is kindly reminded
that in all cases where nitrogen load was not
selected as an input variable, the model has
been discarded and replaced by a meta
model.
For the statistical models it is repeated that “a
suite of ecological[ly] relevant indicators […]
was introduced in order to obtain a more
holistic approach”.
Again, the panel is kindly reminded that no
suite of ecological indicators was introduced to
support model results from the mechanistic
modeling approach. Comparing the final
results from the two approaches directly is
therefore a questionable procedure.
The comparison of results from the two
modeling approaches “revealed
an overall
satisfactory agreement between the two model
approaches”
according to the presented
report. This, unfortunately, is not true.
In section 8.8.1, page 129, it is described that
a two-way analysis of variance gives P = 0.06.
The estimated percent load reductions are,
thus, very close to being significantly different
between the two modeling approaches.
As described further in our comments to
section 8.8, comparing model results on
32
9.3
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0094.png
Landbrug & Fødevarer
chlorophyll
a
and Kd models, respectively,
clearly shows that the two modeling
approaches yield significantly different results.
Tables 8.3 and 8.4 (on page 79) are further
mentioned as proof that results only deviate
slightly between approaches. However, this is
not the case. In the table below, differences,
numerical and in percent, between the
chlorophyll
a
reference values calculated by
mechanistic and statistical models as
presented in Table 8.3 are shown:
WB no.
165
1
102
147
2
92
123
156
1
1
1
2.5
1.3
1.4
1.8
1.1
1.3
2.2
4.1
1.7
2.4
Mech.
Stat.
2
Diff.
-
-
0.1
0.3
-0.3
2.8
0.3
0.6
Diff (%)
-
-
10%
30%
-12%
215%
21%
33%
It is clear that large and very large deviation
between results from mechanistic and
statistical models are common.
Regime shifts are mentioned and briefly
discussed. Such shifts are central to the
critique of extrapolating correlations between
chlorophyll
a
and nitrogen load far beyond the
defined range.
For instance the “eelgrass disease” in the
1930ies, which killed a vast part of the Danish
eelgrass population, very likely caused a
regime shift. The same could be the case of
stone fishing, significantly altering bottom
conditions in many larger water bodies; to
mention just two examples.
It is mentioned that model development should
be based on “state-of-the-art knowledge”.
The panel is, once again, advised to pay
attention to the lack of peer-reviewed
publishing of the statistical models. A report
alone cannot be accepted as scientific
documentation!
9.4
General
33
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0095.png
Landbrug & Fødevarer
10.-12 Conclusion, Epiloque, and References
Chapter/section
Comments
Questions
Conclusion
Epilogue
References
Chapter/section
Appendix A
Comments
Questions
Appendix B
Appendix C
General
34
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
1803667_0096.png
Landbrug & Fødevarer
References
Dahl, K. (ed.), Andersen, J.H. (ed.), Riemann, B. (ed.), Carstensen, J., Christiansen,
T., Krause-Jensen, D., Josefson, A.B., Larsen, M.M., Petersen, J.K., Rasmussen,
M.B. & Stand, J., 2005. Redskaber til vurdering af miljø- og naturkvalitet i de danske
farvande. Typeinddeling, udvalgte indikatorer og eksempler på klassifikation.
Danmarks Miljøundersøgelser. 158 pp. Scientific report from DMU nr. 535. (In Danish)
http://www2.dmu.dk/1_viden/2_Publikationer/3_fagrapporter/rapporter/FR535.PDF.
Duarte, C.M., Conley, D.J., Carstensen, J. & Sanchez-Camacho M., 2009. Return to
Neverland: Shifting Baselines Affect Eutrophication Restoration Targets. Estuaries and Coasts 32, 29-
36.
Flindt, M.R., Kristensen, E. & Valdemarsen, T., 2011. Svigtende reetablering af ålegræs i fjorde. Vand
& Jord 18 (1), 17-20.
Hansen, J.W. (ed.), 2015. Marine områder 2015, NOVANA. Aarhus University, Department of
Bioscience. Scientific report no. 123. (In Danish).
http://dce2.au.dk/pub/SR123.pdf.
Naturstyrelsen, 2013. Natura 2000 basisanalyse 2016 – 2021, Stege Nor. (In Danish).
http://naturstyrelsen.dk/media/nst/90509/N2000_N180_Stege_Nor.pdf.
Orth, R.J., Moore K.A., Marion, S.R., Wilcox, D.J., Parrish, D.B., 2012. Seed addition facilitates
eelgrass recovery in a coastal bay system. Marine Ecology Progress Series 448, 177–195.
SWD, 2012. Commission staff working document. Member state: Denmark. Report from the
commission to the European Parliament and the Council on the implementation of the Water
Framework Directive (2000/60/EC).
http://ec.europa.eu/environment/water/water-
framework/pdf/3rd_report/CWD-2012-379_EN-Vol3_DK.pdf.
35
 MOF, Alm.del - 2017-18 - Bilag 35: Rapporten fra det internationale ekspertpanel om evaluering af de danske marine modeller
Landbrug & Fødevarer
List of Appendices
Appendix 1: Dannisøe, J.G. & Erichsen, A., 2017. Optimisation of the Nitrogen Loadings to Karrebæk
Fjord. Seasonal Effects from Nitrogen Reductions. DHI.
Appendix 2: Skov, T., 2017. Comments to data analysis part in the report “Development of models and
methods to support the establishment of Danish River Basin Management Plans”.Chemometrics and
Analytical Technology, University of Copenhagen.
Appendix 3: Tables 6 and 7 from the original reports in Danish.
Appendix 4: Additional information about Stege Nor.
Appendix 5: Møller, J.K. & Christiansen, L.E., 2016. Memorandum regarding “Modeller for danske
fjorde og kystnære havområder”. DTU Compute, Technical University of Denmark.
Appendix 6: Alternative Analysis of Variance.
Appendix 7: Andersen, J.H., Harvey, T., Kallenbach, E., Murray, C., Al-Hamdani, Z. & Stock, A., 2017.
Under the surface: A gradient study of human impacts in Danish marine waters. NIVA Denmark Water
Research.
36