Modifications to the SWAT code for modelling direct pesticide losses

• Published in: Environmental modelling & software : with environment data news 2008, Vol.23 (1), p.72-81
• Main Authors: Holvoet, K., van Griensven, A., Gevaert, V., Seuntjens, P., Vanrolleghem, P.A.
• Format: Article
• Language: English
• Subjects: Atrazine; Direct losses; Freshwater; Modelling pesticides; SWAT
• ISSN: 1364-8152
• DOI: 10.1016/j.envsoft.2007.05.002

In different river catchments in Europe, pesticide concentrations in surface waters frequently exceed the standards, possibly resulting in negative impacts on aquatic fauna and flora. Pesticides can enter river systems both immediately after application, i.e. as a direct loss, or with some time delay due to runoff or leaching. We define a direct loss as the sum of point losses and drift losses on an application day that will reach the river immediately after or during application. Point losses are due to the clean-up of spray equipment, leaking tools, waste water treatment plants etc. Different studies demonstrated the importance of direct losses. In small river systems, their contribution accounts for 30 to 90% of the pesticide load to surface water. As many studies and models only partly take into account these direct losses or even not at all, we attempted to model the dynamic occurrence of pesticides also coming from these sources. For this purpose, some modifications and extensions to the SWAT (Soil and Water Assessment Tool) model were made. Special attention was paid to closing mass balances and implementing an estimator for total direct losses, drift and point losses. To verify the modifications we focused on the use of the herbicide atrazine in the Nil, a small and hilly river basin in the centre of Belgium. The modified SWAT code resulted in a better correspondence between measured and simulated atrazine concentrations and loads, in particular for direct losses. For the year 1998, the Nash–Sutcliffe coefficient improved from a value of −2.63 to 0.66. In addition, the modelling results of the test case revealed that the contribution of drift losses to the total pesticide load in the river system is rather small: even without a ‘non spray zone’, they account for only 1% of the total load. Point sources, on the other hand, contribute for 22% up to 70% of the pesticide load and need to be considered in pesticide pollution management. The resulting model needs further testing for other pesticides and other catchments. In future, the model can be used for comparison of different measures that can be taken to minimise pesticide fluxes towards river systems and in performing realistic risk assessments.