S.Z. Cohen, R.D. Wauchope, A.W. Klein, C.V. Eadsforth and R.Graney
Pure & Applied Chemistry, 67(12):2109-2148, International Union of Pure and Applied Chemistry, 1995.
The process of modeling the leaching and runoff of pesticides is simple in concept but complex in execution. Models are physical, conceptual, or mathematical representations of reality. Screening-level models are an appropriate first step for examining pesticide leachate and runoff potential, as long as conservative input assumptions are used. They may consist of comparisons of certain mobility and persistence properties with numerical criteria, or they may require pencil, paper, and a hand calculator. At a higher level of sophistication, a wide variety of computer models are available that can quantitatively simulate pesticide leaching and runoff in the aqueous phase. It is important to pick a model that has been validated in more than one study, has good user support, requires an amount of data input appropriate for the application, and has a history of producing results acceptable to scientists and regulatory authorities. Considering these various criteria for acceptability, EPA's PRZM2 model and the German modification, PELMO, would be appropriate for evaluating leaching potential. The GLEAMS, LEACHM, and CALF models are also scientifically acceptable, but have not been as widely used. The GLEAMS model is appropriate for quantifying runoff potential in simple, field-scale drainage patterns. The more complex SWRRBWQ model is more appropriate for watershed-scale assessments. The most appropriate use of these computer simulation models is to rank the contamination potential of a particular pesticide at several sites or rank several pesticides at one site. Another excellent application of these models is to calibrate them to fit the results of an intensive field study at one site, and extrapolate to other points in time and space for the same pesticide. One should always recognize the variability in natural processes and field conditions, and use probabilistic (stochastic) analysis whenever possible. More model validation and calibration is needed in tropical climates and in special situations such as turf, forests and orchards.