AN EXPERT-SYSTEM TO DETERMINE THE PROBABILITY OF PESTICIDE LEACHING

This effort focused upon the development of an expert system capable o f determining the probabilities of pesticide leaching. Two levels of a nalysis are available to the user. The first accesses cumulative densi ty functions describing pesticide leaching prepared in the related eff ort described in a parallel article in this issue, while the second al lows the user to configure a Monte Carlo simulation for a specific sit e of interest. The first option is recommended as a screening tool use ful in defining general probabilities as the results represent interpo lations from a regional data base. The second option is for more speci fic analyses and applies the U.S. Environmental Protection Agency's (E PA) pesticide Root Zone Model (PRZM) within the expert system. Detaile d directions for the selection of site-specific parameters required by the EPA code are given by the expert system. Base maps, tables and fi gures are provided together with detailed discussions of the input val ues requested. The expert system prepares input data files for the PRZ M code, randomizes annual meteorological data, completes and subsequen tly pools individual simulations into cumulative density functions for the determination of the probabilities of pesticide leaching. Additio nally, the expert system plots the running mean and standard deviation of these individual simulations to identify the adequacy of the sampl e base. The expert system allows individuals with minimal computer ski lls access to the EPA code as well as to probabilistic simulations.