Stochastic Simulation of Pesticide Transport in Heterogeneous Unsaturated Fields

Abstract

AbstractCommonly used vadose‐zone models typically lack the capability to account for the spatial variability of soil properties that can greatly affect water flow and solute transport. In this study, stochastic techniques were applied to Groundwater Loading Effects of Agricultural Management Systems (GLEAMS), a deterministic vadose‐zone model, to simulate pesticide transport within experimental plots in southern Ohio subject to three different agricultural treatments. A hydrologic environment was discretized vertically into three layers based on the characteristics of the soil horizons. Selected hydrologic parameters in GLEAMS were described within each discretized layer by a random multivariate normal (MVN) vector. Simulations of transport and fate of three commonly used pesticides in the top 1.22‐m root zone were carried out at two levels: (i) single‐value output was obtained by using the mean values of the MVN vector components; and (ii) an empirical probability density function (pdf) was obtained by using the generated MVN vector realizations for each discretized layer. Statistical analyses were conducted to compare the simulation results and field observations, and examine the treatment effect on pesticide transport. Conclusions include: (i) soil spatial heterogeneity significantly affected pesticide movement and caused spatial variation of their distribution, and (ii) the Level 2 stochastic approach resulted in a better prediction of pesticide transport and fate.