A robust methodology to predict diffuse source pollution in the aquatic environment : A case study for the Colworth catchment, Bedfordshire

Abstract

Throughout the world, increased use of agrochemicals, warrant urgent measures to minimise their impacts on the aquatic environment. Monitoring is the usual practice to identify and quantify pollutants, however, it is an after-the-fact occurrence, and monitoring over large spatial scales is not feasible. Mathematical models provide an alternative to monitoring and can be used both to forecast future pollution and to investigate the impacts of potential control measures. This thesis is therefore focussed on catchment scale modelling of pesticide pollution of rivers. Data from a 142-ha agricultural catchment on the Unilever Colworth estate, have been used to model pesticide pollution at the catchment scale. After a review of available models SWAT-2000 was selected as the modelling tool. An acceptable performance in hydrological modelling, along with correct simulation of the processes driving the water balance were essential first requirements. After modification of the SWAT-2000 source code, hydrological simulation, crop growth and evapotranspiration patterns were realistic when compared with monitored data SOUTCGS. Hydrological modelling was carried out using both daily and sub-daily data with Hargreaves and Penman-Montieth methods of evapotranspiration and the NRCS- Curve Number and Green-Ampt methods of runoff generation. A sensitivity analysis identified GWQMN, AWC and ESCO as the most sensitive parameters. These control baseflow, water content of soil layers, and depth distribution of evaporation from the soil respectively. The best combination of runoff generation and evapotranspiration components was identified as Curve Number with Hargreaves. Modelling of sediment and pesticide loss was carried out for four pesticides. Various management scenarios were designed and tested with the objective of reducing pesticide loss to streams and a web-based tool was developed to give advice on pesticide application. With the modifications implemented, it has been demonstrated that SWAT is a useful tool for modelling pesticide behaviour at catchment scale in UK conditions.