Peter Daldorph1, Kevin Hiscock2, Helen He2, Sam Taylor2, Andrew Wade3
1Atkins Global; 2School of Environmental Sciences, University of East Anglia; 3Department of Geography and Environmental Science, University of Reading.
Simon Eyre, Anglian Water; Zoe Frogbrook, Scottish Water; Claire Bell, Environment Agency.
How can I know if land management will be effective?
Where is the pollution coming from? And how do pollutants act together?
Using the water quality models, SAGIS-SIMCAT and SWAT, a reduction in the risk of exceeding drinking water quality standards at a water treatment works intake was shown for nutrients and metaldehyde as a result of introducing farm measures. The costs of these measures were then compared to the economic benefits of reduced risk.
Impact of agricultural measures on metaldehyde (number of days exceeding drinking water standard) and nitrate (99 percentile) at river intake. Scenarios – S1 = 6m buffer strips, S2 = Max application rate 0.16 kg ha-1, S3 = No application slope > 2%, S4 = No application on clay soils, BM = best measures, MM = maximum measures.
The water quality models SWAT (metaldehyde) and SAGIS-SIMCAT (phosphorus and nitrate) were set up and calibrated on the River Wensum (Norfolk). For SWAT, a daily timestep was used whereas SAGIS-SIMCAT was set up with statistical data (mean and standard deviations) for the period 2010 to 2012. Sequential Uncertainty Fitting was applied to calibrate SWAT whilst SAGIS-SIMCAT was calibrated by a calibration technique developed for UKWIR.
Model scenarios were then run to test the impact of catchment measures (e.g. substitution of metaldehyde by the alternative ferric phosphate and farm management to reduce nutrients defined using the agricultural management model, Farmscoper) on operational and compliance risk at Anglian Water’s Costessey Pits intake. Compliance risk for metaldehyde was assessed as the number of days exceeding the drinking water standard, whilst for nitrate and phosphorus mean, 95 and 99 percentile values were compared to the standard at the intake and associated reservoir.
The water quality models SWAT (metaldehyde) and SAGIS-SIMCAT (phosphorus and nitrate) were set up and calibrated on the River Wensum (Norfolk). SWAT includes parameters which define the physical and chemical properties of pesticides and control their transport and fate within the model (Arnold et al., 2012). Metaldehyde is not included in the SWAT pesticide database and so the values of these properties for metaldehyde were determined from the literature and manually added to the model database. Inputs to SAGIS are based on a combination of observed data (e.g. for point sources and output from other models; notably PSYCHIC and NEAP N for phosphorus and nitrogen respectively; Comber et al. 2013). For SWAT, a daily time-step was used whereas SAGIS-SIMCAT was set up with statistical data for the period 2010 to 2012. Sequential Uncertainty Fitting was applied to calibrate SWAT whilst SAGIS-SIMCAT was calibrated by a calibration technique developed for UKWIR.
For SWAT, the following scenarios were run (the percentage reduction in pesticide use is also shown).
For SAGIS, agricultural measures were applied to reduce the catchment inputs of both nitrogen and phosphorus from the catchment based on output from the model, Farmscoper, calculated at the waterbody scale. The following measures were applied:
Compliance risk for metaldehyde was assessed as the number of days exceeding the drinking water standard at the Costessey Pits intake and further downstream at the Heigham intake. For nitrate the mean, 95 percentile and 99 percentile values were generated with a view to assessing the risk of exceedance of the drinking water standard. For phosphorus, the key issue is the impact on nutrient loading and consequent enrichment of the bank side storage reservoir, Costessey Pits, that provides short-term storage of river water before treatment. Average phosphorus loads (kg/day) were compared as well as predicted mean and percentile values in the reservoir. In addition, SAGIS provides information on source apportionment which was reviewed to identify the key pollution sources.
In addition, Farmscoper provides indicative implementation costs for the measures whilst substitution costs for metaldehyde with ferric phosphate were calculated based in the Anglian Water’s incentive scheme ‘Slug It Out’. These costs were compared to the operational and compliance risk benefits.
S.D.W.Comber, R.Smith, P.W.G.Daldorph, M.J. Gardner, C.Constantino, and B.Ellor (2013) Development of a Chemical Source Apportionment Decision Support Framework for Catchment Management. Environ. Sci. Technol., 2013, 47 (17), pp 9824–9832
Arnold, J. G., Kiniry, J. R., Srinivasan, R., Williams, J. R., Haney, E. B. and Neitsch, S. L. (2012). Soil and Water Assessment Tool Input/Output File Documentation, Version 2012. Temple: Texas Water Resources Institute.
For metaldehyde measures considered included: no applications to high risk areas (high slope and clay soils), reduced maximum application rates, cultural measures to reduce need for application and buffer strips. Although no mitigation option resulted in nil-exceedance of the drinking water standard for metaldehyde at the intake, some measures resulted in a marked reduction in the risk of water quality non-compliance (percentage of time exceeding standard reduced from 15% to 6%). The findings suggest that the most effective approach to reduce metaldehyde concentrations at the raw water inlet sites involves targeting areas that are at a relatively high-risk of metaldehyde loss. For phosphorus, only a marginal reduction in reservoir nutrient loading was achieved as a result of best practice voluntary measures and maximum measures but for nitrate the 99 percentile was reduced further below the standard increasing headroom and therefore reducing risk.
River Wensum catchment showing location of Costessey Pits intake
Modelled metaldehyde concentrations at Costessey Pits intake (using SWAT)
Source apportionment of nitrate along the length of the River Wensum (arrow shows location of intake)
Drinking water standard for metaldehyde and percent exceedance curves for the Costessey Pits inlet
Figure 1 shows that River Wensum catchment with key features including the location of the Costessey Pits and Heigham intakes. Simulated river flow and metadehyde concentrations generated by SWAT are shown in Figures 2 and 3, respectively.
Figure 1 River Wensum catchment showing location of Costessey Pits intake
Figure 2 Hydrographs depicting observed (solid line) and simulated (dotted line) daily mean discharge for the flow gauges located at Costessey Mill during the calibration
Figure 3 Modelled metaldehyde concentrations at Costessey Pits (c) and Heigham intakes (d)
To assess the risk of metaldehyde concentrations exceeding the 0.1 µg/l drinking water standard, percent exceedance curves were developed which depict the proportion of time exceeding the standard (Table 1).
Table 1 Percentage of time exceeding drinking water model scenarios
No scenario resulted in nil-exceedance of the 0.1 µg/l standard limit at the Costessey Pits and Heigham intakes. Scenarios S3 and S4 were found to be the most effective mitigation options, reducing the number of exceedance days from 15 to 6-7% at the two intakes. Although metaldehyde use was prohibited for approximately 42% and 39% of the catchment area under scenarios S3 and S4, respectively, the percentage reductions in metaldehyde concentrations at the inlet sites was much greater, indicating that the targeted areas of clay soils and slopes greater than 2% are at a relatively higher risk of metaldehyde loss. These findings suggest reducing the risk from metaldehyde is difficutlt, bearing in mind the requirement for no exceedances. The most effective approach to reduce metaldehyde concentrations at the raw water inlet sites, however, involves targeting areas that are at a relatively high-risk of metaldehyde loss.
Figures 5 and 6 show source apportionment chainage plots along the length of the River Wensum for nitrate and phosphorus, respectively. Nitrate concentrations at the intake are dominated by arable farming and atmospheric sources with less of a contribution from sewage works. In constrast, sewage works, industrial discharges (food processing plants) and livestock provide a greater contribution of ortho-phosphate concentrations.
Figure 5 Source apportionment of nitrate along the length of the River Wensum (arrow shows location of intake)
Figure 6 Source apportionment of ortho-phosphate along the length of the River Wensum
Figure 7 shows the impact of the agricultural measures at the Costessey Pits Intake. Nitrate concentrations are below the nitrate drinking water standard of 11.3 mg/l N so the risk of exceedance is low but the results show that the headroom below the standard is increased by the measures. Reductions in ortho-P concentrations are relatively small (15% reduction for the maximum measures) so unlikely to have a significant impact on the trophic status of Costessey Pits (simulated within-reservoir total P (TP) concentrations reduce from 0.013 to 0.011 mg/l under the maximum measures).
Figure 7 Change in the nitrate 99%ile and mean ortho-P concentrations (mg/l) at the Costessey Pits Intake as a result of agricultural measures
The estimated cost incurred by the water company for product substitution (by ferric phosphate) across the Wensum catchment are £52K and £48K for the most effective metaldehyde scenarios S3 and S4, respectively. This does not include costs to employ advisors and agronomists or costs to farmers related to crop yields. The costs of the agricultural measures to control nutrients are £900K for best measures and £7500K for the maximum measures.
Metaldehyde is not removed effectively by conventional pesticide treatment such as GAC and ozone. The need for catchment management to work is based on the principles of avoiding the significant additional CAPEX and OPEX to install and operate advanced ozonation and UV processes or a regulatory ban on the product.
Applying more than one model allowed consideration of several pollutants which is necessary when considering water treatment and operational issues at a water treatment works as compliance is required for all chemicals. Coupling models (Farmscoper and SAGIS) allowed linking of the impact of farm measures to within-river concentrations.
Engagement with the stakeholders was important to learn from previous modelling work and past analysis of the impact of measures which helped to develop meaningful scenarios. Identifying key outputs was also important to make these relevant to the operators and regulators.
Each model requires specific formats and, consequently, the pre-processing work mainly involved formatting of raw data to create the required inputs. Post-processing was required to create the most useful output statistics, for example the percentage of time exceeding the standard.
Observed river flow data, observed effluent quality and flow data, observed river quality data, observed rainfall, land use, agricultural census data, topography and geology, soil chemistry, river network and catchment GIS data.
East Anglia, England
Improved access to and integration among data and models to address key questions in catchment management for water quality and wider ecosystem services, providing a more holistic view to inform scientific understanding and policy development.