Evaluation of an open-source nutrient delivery model for estimating pesticide loads in river catchments

Quantifying pesticide runoff hazard in catchments is necessary to predict the impacts and target mitigation. Achieving this at scale through regular, long-term water quality monitoring at multiple sites is time- and resource-intensive. Ideally, such monitoring should be supplemented by models that can estimate pesticide loads in a quicker, less costly manner, especially for unmonitored catchments. We developed a novel modelling method combining the Integrated Valuation of Ecosystem Services and Tradeoffs Nutrient Delivery Ratio (InVEST® NDR) model and the UKCEH Land Cover® plus: Pesticides maps to estimate pesticide load across England. The InVEST NDR model is a widely used, open-source pollutant runoff model, but has not yet been evaluated for use with pesticides. We compared our modelled approach with a measurement-based ("measured") approach. This measured approach used pesticide concentration data from the Environment Agency and river flow data using Qube (a water resource estimation tool) for catchments upstream of the sampling sites: 54 for bentazone and 21 for chlorotoluron. The significant positive relationships between measured and modelled pesticide loads were stronger when the measured approach accounted for the proportionate area of arable land cover, presumably because the modelled approach only incorporated pesticides from arable sources, explaining up to 90 % of the variation in relative hazard between catchments. Thus, our modelled method forms a flexible approach to mapping relative pesticide runoff hazard over large spatial extents, especially where monitoring is limited. It could also be used to rapidly prioritise catchments for more complex analysis to produce accurate measures of absolute loads.