Main authors: | Berit Hasler, Fiona Nicholson, John Williams, Rachel Cassidy, Linda Tendler, Peter Lendertsee, Marije Hoogendoorn, Rikke Krogshave Laursen, Doan Nainngolan, Ingrid Nesheim |
FAIRWAYiS Editor: | Jane Brandt |
Source document: | »Hasler, B. et al. (2019) Assessment of costs and benefits for farmers, water companies and society from using Decision Support Tools. FAIRWAY Project Deliverable 5.3 49 pp |
The following decision support tools for use at catchment and national farm level were evaluated.
Contents table |
1. Farmscoper, UK |
2. TargetEconN, Denmark |
3. Summary and discussion on catchment scale DSTs |
1. Farmscoper, UK
The Farmscoper DST and its use at the farm scale have been described in Section 3.6 above. In addition, the DST can also be used at a catchment or national level using the Upscale tool. The Upscale tool is prepopulated with catchment level census data for Water Framework Directive waterbodies up to river basin scales (in England only). This has been used to inform water quality management strategies and has been applied as a policy tool in a number of studies to date (e.g. Micha et al., 2018; Collins et al., 2016; Gooday et al., 2014; Zhang et al., 2012).
Modelling Cost-Effective Measures using Farmscoper
The current version of Farmscoper contains over 100 mitigation measures that can be applied to a given farm or farm type. The options are based primarily on the Defra Mitigation Method Guide and includes those relating to Cross Compliance, Catchment Sensitive Farming and the Countryside Stewardship Schemes in England and Wales. Each option has a full cost and contaminant loss estimate associated with it, and they are classified depending on whether their impacts relate to nutrients, livestock, soil, delivery or pesticides either singly or in combination. Measures applicable to arable crops, for example, include cultivation of compacted soils, establishing buffer strips, management of field corners, wild bird cover, uncropped margins and leaving residual levels of non-aggressive weeds in crops. For dairy farms, measures include increased scraping frequency in dairy cow cubicle housing and washing down of dairy cow collecting yards.
Inclusion of Measures in the Farmscoper DST
Measures for implementation are selected from lists in the Farmscoper Evaluate tool. For selected measures it is possible to estimate, both individually and in combination, the effects on losses to the environment and alterations in economic outputs in terms of increased/decreased yields, changes in inputs, time spent and increased/decreased uncertainty of the outcomes. It provides both the environmental impacts and the cost implications of single and sets of mitigation measures. It is possible to select measures based on their effect on particular pollutants (e.g. pesticides or nitrate) and optimise these automatically within the tool. The output from such an optimisation process is a graph of cost plotted against the % reduction in a specific pollutant allowing the user to identify the point at which cost-benefits are maximised. Reporting features provide tabulated and graphical outputs to facilitate comparison of options.
The initial scenario evaluated considers the impact of introducing mitigation methods that correspond to Cross Compliance Good Agricultural and Environmental Conditions, namely:
- 1 (Establishment of Buffer Strips along Water Courses),
- 4 (Providing Minimum Soil Cover) and
- 5 (Minimising Soil Erosion).
These methods (“X-C GAEC” in the model) are given in Table 6 below. In a second scenario, 4 additional measures were added covering potential sources and pathways of sediment and nutrients to water courses (Table 7). A 100% implementation of each measure on the farm was assumed, but it is possible to also consider a partial implementation (such as where a crop rotation is implemented and a measure is only active on a sub-set of fields in each year). A comparison of the outputs from the model is given in Table 8. The baseline scenario refers to no implementation of measures on the farm.
Farm total production changes little from £81,212 (€95,455) £44,949 (€52,832) gross margin after costs of £36,263 (€42,623)) for no measures to £81,173 (€95,410) for all measures (both scenarios) as little land is removed from production and stocking rates would not alter. The costs of implementing the measures however need to be included; as fixed costs covering the capital investment, labour and machinery associated with the measure and variable costs covering the change in gross margin for stock or cropping based on implementation of the measure. This is an estimated total of £8,618 (€10,129) for Scenario 1 or £10,211 (€12,002) for Scenario 2.
The environmental benefit value (£912 (€1,072) for Scenario 1; £1972 (€2,318) for scenario 2) is an estimate of the monetary value of the pollutant emission reductions, converting reductions in methane and nitrous oxide to CO2 equivalents, reductions in ammonia to a value representing the damage costs to society and reductions in nitrate, phosphorus and sediment to values representative of the impacts of water pollution on water quality. Given that the mitigation measures selected for this example target overland pathways for contaminant losses to waterbodies, the greatest improvements in pollutant outputs are in sediment, with a reduction in load of 17.4% for Scenario 1 and 30.2% for scenario 2, and phosphorus, with a reduction in load of 16.1% (1,404kg) for Scenario 1 and 21.7% (2,441kg) for Scenario 2. To maximise cost-benefits Farmscoper allows model optimisations to be run for single or multiple parameters, identifying the investment which maximises the reduction of phosphorus and sediment losses.
Table 6: Mitigation measures relating to Cross Compliance Good Agricultural and Environmental Conditions (“X-C GAEC” in the model).
Method IDs: Set 1 | Description |
5 | Early harvesting and establishment of crops in the autumn |
8 | Cultivate compacted tillage soils |
9 | Cultivate and drill across the slope |
10 | Leave autumn seedbeds rough |
11 | Manage over-winter tramlines |
14 | Establish riparian buffer strips |
37 | Reduce field stocking rates when soils are wet |
76 | Fence off rivers and streams from livestock |
115 | Leave over winter stubbles |
Table 7: Mitigation measures relating to Cross Compliance Good Agricultural and Environmental Conditions with 4 additional mitigation measures (marked with bold text).
Method IDs: Set 1 | Description |
5 | Early harvesting and establishment of crops in the autumn |
8 | Cultivate compacted tillage soils |
9 | Cultivate and drill across the slope |
10 | Leave autumn seedbeds rough |
11 | Manage over-winter tramlines |
14 | Establish riparian buffer strips |
37 | Reduce field stocking rates when soils are wet |
38 | Move feeders at regular intervals |
39 | Construct troughs with concrete base |
60 | Site solid manure heaps away from watercourses/field drains |
76 | Fence off rivers and streams from livestock |
79 | Farm track management |
115 | Leave over winter stubbles |
Table 8: Comparison of output costs (£) and benefits for implementation of 2 mitigation option sets on the Lowland Grazing Farm.
Spatial scale – Upscaling to Catchment or National Scales
The Farmscoper Upscale tool allows pollutant losses to air and water to be made at catchment up to national scales for England (data flows and relationships among the tools are summarised in Figure 2). For each catchment multiple farm models are generated representing those typical within the catchment and calculating losses from those catchments under the different climatic and soil characteristics present in the catchment. Up to ten farm types covering the main categories present in England can be represented in the upscaling model; each customised according to the likely number, size and stocking rates and land use within a particular catchment. For small catchments with records for individual farms can be used or for larger catchments the farms can be generated using Census data (from Defra’s 2015 June Agricultural Survey for England) included in the tool covering 4091 WFD waterbodies, 336 Operational Catchments,90 water management catchments and 10 river basin districts in England. The spatial definition of farms has been improved within the 2015 database used in the current DST. Instead of a single geographical location for a farm to which a climate and soil type was defined, the current database uses field boundary data and extracts the proportions of each farm within each WFD waterbody, their individual climate and soil types and represents them within the model.
Figure 2
Ability to model risk and uncertainty
Uncertainties in the model both at single farm and small to large catchment scales have been considered by the developer and are included with the DST documentation.
The model is based on a database of pre-calculated contaminant export coefficients to generate a baseline contaminant load. These calculations are based on a set of assumptions from national level practices (such as the standard times at which fertiliser or slurry is applied) and what is typical across a farm type rather than data from an individual farm. Climate and soil data were aggregated to 3 soil types and 6 climatic zones. In each application of the model these assumptions need to be considered and baseline outputs validated against available monitoring data for the catchments where possible.
There is considerable flexibility within the model for customisation and adaptation. Baseline pollutant estimates and reductions associated with each mitigation measure can be assigned uncertainty bounds within the tool. A set of sensitivity controls within the Evaluate Tool allows a range of variation to be set for each contaminant, with a default value of 25%, within a Confidence Ranges worksheet (Table 9). Evaluation of measures with sensitivity included provides outputs which are robust within the defined uncertainty limits in the relative contributions of different contaminant source and pathway.
Table 9: Tabulated display of confidence limit settings (set to 25% default) for Nitrate, Phosphorus and Sediment within Farmscoper
Use of the model by decision makers
Several published studies have used the Farmscoper DST at catchment scales to estimate the impact of a range of mitigation measures. Collins et al. (2016), for example, surveyed farmers across England to identify the mitigation measures more likely to be adopted by farmers and then applied the model to identify the potential reductions in emissions to air and water relative to business as usual. Business as usual emissions and uncertainties were based on comparisons with available monitoring data for England and Wales but acknowledging the limitations of available data in terms of low sampling frequencies and difficulties in disaggregating non agricultural sources. Across a range of farm types they identified 29 measures which, due primarily to low cost of implementation, were appealing to farmers and likely to be adopted. They then evaluated these measures for the major farm types in 99 water management catchments across England and Wales and, assuming a 95% uptake level, established what the technically feasible impact on agricultural emissions to air and water would be. Projected emission reductions across the catchments were estimated to range between 8-37% for sediment, 12-24% for ammonia, 6-29% for Total Phosphorus, 4-16% for nitrate/methane and 5-10% for nitrous oxide. This information provides guidance and evidence for policy makers in the development of agri-environmental schemes, the likely costs and their efficacy across a range of catchment typologies and farm types.
2. TargetEconN, Denmark
The Target Econ N model for decision support The TargetEconN model is an integrated economic and biophysical social planner DST, set up for Danish catchments/watersheds. The model minimizes the costs for society, of meeting a nutrient load reduction target in a specific water body, from the catchment loading to this water body.
The DST was designed for the assessment of cost-effective implementation of nutrient load reductions, as required in the Water Framework Directive to achieve good ecological status. The model is now calibrated for the whole of Denmark and is set up for modelling reductions to coastal water and lakes. The model will be set up to include groundwater as soon as the data sources for this modelling are available. The first version of the model was designed for the Limfjorden catchment, where Aalborg, one of the Danish case studies in FAIRWAY, is situated (Konrad et al., 2017; Hasler et al., 2019). A fact sheet describing the model concept can be found at http://dnmark.org/wp-content/uploads/2017/03/Fact-sheet-TargetEconN-modelling-framework_Final.pdf.
The model has been developed by Aarhus University as part of several research projects and by funding from the Danish Economic Councils and by the Ministry of Environment and Food. The model has been used for advising the Danish Ministry of Environment and Food, as well as the Danish Economic Councils.
Method to elicit cost-effective solutions
The method used in TargetEconN is to minimise the costs for reaching the N reduction target by implementing abatement measures in the fields of the catchment. Only one measure can be implemented for each field in order to avoid infeasible solutions, such as implementation of reduced nitrogen application and wetland at the same time. The costs and effects of the measures are modelled using information on the crops grown at field scale in the catchments. The effects of the measures are measured as the leaching in kg per ha from the root zone, using empirical leaching functions. The transport from the root zone to the coast is, being modelled by retention coefficients and the transport includes retention in soil, surface and groundwater. The retention reduces the nitrogen load to coastal areas between 0 and 90% of the initial leaching. The capacity for implementing each of the N abatement measures in the catchment is an assessment of the hydrological and land use specific potential, subtracting the area where measures have been implemented before.
The optimization routine enables calculation of the optimal spatial location of nutrient abatement measures, taking into account the spatial differences in costs, the potential of implementing the different measures at field level as well as the nitrogen leaching reduction effects. The model is developed for both nitrogen and phosphorus, but because of poor data the model is not calibrated for phosphorus yet.
Data overview
The modelling at field scale level is enabled by detailed data at field level from the General Agricultural Register and the Danish Husbandry register, collected by the Ministry of Environment and Food, and used for establishing the dataset Basemap (Levin et al, 2017). These registers include information on crops grown at each field, and time series data are used. The data set also includes information at crop- and field level on manure and fertiliser application. Prices for the crops included in the model were averages from 2011-2013 in the former Limfjord application (Hasler et al., 2019), and for the most recent version of the model these costs are updated to an average for the period 2013-2018. Cost data are collated from Farmtal Online, administered by the Danish Agricultural Advisory service, SEGES (2018).
Abatement measures
The model includes in total 24 different N abatement measures for sand and clay soils. The abatement measures include technological, land management and set aside options, and all measures are implemented on agricultural land, except for constructed wetland. This measure is expected to be implemented in areas close to the fields, and the wetlands reduce the nitrogen loads from the fields to the coast.
TargetEconN's ability to model the spatial distribution between farms or locations
TargetEconN is not set up to model farms, but is configured to model the cost-effective allocation of measures at agricultural field scale. The map in Figure 3 shows the spatial allocation of measures in the Limfjord catchment from implementing nitrogen load reductions of 4165 tons N. The map shows the distribution of the measures implemented to achieve the cost-effectve solution (Hasler et al., 2019, page 915).
Figure 3
TargetEconN has also been used to compare a general regulation with more targeted regulations, which is more cost-effective as the targetting included optimisation of the spatial localisation of the measures to both costs and nitrogen load reductions. This comparison showed that the savings for society and agriculture can be substantial: While the general, non-targeted regulation could achieve the load reduction target of the fjord at a cost of 58 DKK/kg N (approx. 7.76 €/kg N), the targeted and cost-effective solution fulfilled the same load reduction target at a cost of just 13 DKK/kg N (approx. 1.74 €/kg N) (Hasler et al., 2015). This assessment was done for a rather low load reduction target which was the maximum achievable load reduction when using genereal load reductions of the nitrogen appliation to crops in the catchment. The results are not comparable to results from Hasler et al. (2019), illustrated in the maps, as different load reduction targets are achieved, i.e. the costs per kg N are necessarily higher to achieve the high load reduction.
The ability to model risk and uncertainty in TargetEconN
TargetEconN is suitable for sensitivity analyses of the assumptions and data inputs to the model, and the uncertainty inherited in such assumptions. Examples are the assumptions made on effects of measures to reduce the nitrogen leaching and load, the abatement costs or the retention of nitrogen in soil and water before it reaches the target water body. Hasler et al. (2019) tested the effects on the cost-effective solutions from uncertainty on the retention in the catchment. They investigated both the level of retention and the distribution of the retention within the field blocks in the catchment.
Figure 4
The map in Figure 4 shows the spatial distribution of the cost-effective mix of measures, when differences in retention are not taken into acount, but modelled as an average retention in the whole catchment. This application of constant retention on 69% in the whole catchment can be compared to the solution shown in Figure 3, which is built on variations in the retention, ranging from 0 to 90% retention. It can be observed that the measures are more uniformly distributed in Figure 4, compared to Figure 3. Furthermore the costs of achievning the load reductions are 218 Million DKK/year (approx. 29.8 million. €/year) when variation in the retention is assumed, while the costs increase to 273 DKK/year (36.5 €/year) when the retention is averaged thorougout the catchment. The cost-effectiveness of the achievement in the solution with variation in retention is 52 DKK/kg N (6.96 €/kg N) and with constant retention the cost-effectiveness is 66 DKK/kg N (8.83 €/kg N); i.e. taking retention into account reduces the cost signifiantly. Since retention is modelled with uncertainty this assessment is an example of how uncertain assumptions can be modelled and measured (Hasler et al., 2019).
Use of the model by decision makers
The results from the model have been used by different decision makers, such as the Ministry of Environment and Food as well as the Danish Economic Councils. In 2019 the model is being used to advise the Ministry of Environment and Food on cost-effective implementation of the Water Framework Directive, and different sentitivity analyses are being made in order to explore the effect of e.g. capacity of the implementation of measures in the catchments on the cost-effective solution. In the mentioned study the model results are being compared to the results from other models. The usefulness for cost-assessments for waterworks has been discussed with Aalborg Water, who found the facilities of this model attractive because of the detailed data sources and the field scale model results. Because of the field scale data inputs the model can be run at other spatial scales than the predefined catchments used now – the model can e.g. be calibrated to a drinking water protection area. The costs of using the model correspond to the costs of the time researchers use for modelling, and these costs have been covered by the users. The development costs of the model have been covered by research grants.
3. Summary and discussion on catchment scale DSTs
Both Farmscoper and TargetEconN are examples of catchment scale tools that model cost-effective solutions, that can be used for policy advice.
The TargetEconN model is a model tool which is designed to model cost-effective solutions to nutrient policies, and the model has been used to advise decision makers in Denmark. The model is complicated to run, and requires specific software, so the model is only run by researchers. The spatial distribution of the results is considered valuable for decision support, and this is also being commentedsupported by feedback from the Aalborg Water’s point of view. The Ministry of Environment and Food in Denmark have also indicated that the model results provide good information on the allocation of the measures and their distribution. In Hasler et al. (2019) it was concluded that the model is well suited for policy advice, and for assessment of the sensitivity of the assumptions used. TargetEconN has been used to compare a general regulation compared to a more targeted regulation, which is more cost-effective as the targetting included optimisation of the spatial localisation of the measures to both costs and nitrogen load reductions. This comparison showed that the savings for society and agriculture can be substantial. The scenario assessment illustrates that using a DST with detailed hydrologial parameters, as well as cost information, is important for assessment of cost-effectiveness and uncertainty on data inputs and assumptions. This is an important feature of integrated economic and ecological DSTs. The model TargetEconN is well suited to do theise types of sensitivity analyses, and to indicate what the effects are for both resulting load reductions and costs.
Farmscoper has been used as a policy tool in a number of studies to estimate the impact of mitigation measures on water quality. It allows a policy user to test the potential reduction in pollutant loads (includinge nitrate and pesticides) loads that could be achieved by implementing one or more diffuse or point pollution mitigation measures; it also quantifies the costs of such measures and the potential benefits for biodiversity. This approach was designed to allow a more holistic assessment of the mitigation of diffuse agricultural pollution given the different policy targets (e.g. Water Framework Directive, Climate Change Act, and the Gothenburg Protocol) and to identify the mitigation methods that provide multiple benefits. The explicit calculation of agricultural production allows for the identification of mitigation methods that can help to achieve target pollutant reductions whilst not reducing food production or adversely affecting farm profitability.
The Farmscoper tool has been used by a variety of government organisations, research institutes, consultancies, levy bodies and other agricultural organisations for more complex assessments of the impacts of policy scenarios and agri-environment schemes, through to prioritising catchment management plans, and assessing pollutant footprints and mitigation potential of individual farms or groups of farms.
Note: For full references to papers quoted in this article see