Predicting Pesticide Runoff from Agricultural Land: A Conceptual Model

Abstract A mathematical model of the dynamic, single‐rainfall‐event type is being developed to describe quantitatively pesticide runoff as a function of pesticide and soil properties, agricultural practices, watershed characteristics, and climatic factors. From the model, guidelines governing pesticide use will be derived to prevent or minimize water pollution resulting from runoff from agricultural land. The model described is based on a state‐of‐the‐art approach. Pesticide transport is “piggybacked” on existing hydrologic and soil‐loss models. Both direct mass‐balance and areal averaging‐statistical simulation approaches are being considered for the basic runoff and sediment transport descriptors. The conceptual structure of the overall model includes (i) source term effects (parameters that determine the initial concentration of pesticide available for runoff), e.g., pesticide formulation, physical‐chemical properties, and initial spatial distribution; (ii) loss of pesticide from the surface zone between rainfall events by various attenuation processes (volatilization, microbial, chemical, and photochemical degradation, and organism uptake); and (iii) loss of pesticide from the soil surface during runoff‐producing rainfall events because of mass transfer from soil surface into the moving runoff film and pickup of sediment containing pesticide particulates. The minimum output requirements for the model are the prediction of runoff rate and sediment loading plus the corresponding water and sediment phase pesticide concentrations at a designated boundary (normally a watershed confluence or drainage channel) as a function of time during a runoff event. This output is to be based upon input consisting of the type, formulation, and rate and method of application of the pesticide, and the areal extent, topography, type of soil, cropping characteristic, and rainfall characteristics of the watershed or basin. The model will internally generate and store the appropriate boundary condition changes required to predict sequential rainfall response using between‐event meteorological data as input. In addition, the model structure is to be suitable for scaleup for application to areas approaching basin size parametrically, by use of transfer functions, or by use of executive and routing subroutine additions.