| Predicting the fate of agrochemicals in the environment is subject to imprecision in model input data. Although several methods are available to address uncertainty in the predicted fate of agrochemicals in the soil, no method has been accepted in a decision framework. This study focused on the development and application of a method to quantify uncertainty due to imprecision in input data for pesticide fate predictions in the soil profile.;The postulated method to quantify uncertainty relies on the development of probability functions for predicted and true pesticide mass emissions and travel times. The difference in probability between the two functions is used as a measure of uncertainty. Entropy and fuzzy logic methods are applied to quantify the uncertainty in predicted pesticide mass emissions and travel times to specified control depths in the soil profile.;Application of the entropy approach to compute the uncertainty in the predicted mass emissions and travel time revealed considerable spatial variability at the test site. Based on the hexazinone and simazine simulations, uncertainty increases with depth within the first 2.0m of the soil profile. Total site uncertainty in the hexazinone mass emission was 0.33, 0.38, and 0.46 for the 1.0-m, argillic horizon, and 2.0 m depths, respectively. For hexazinone travel time the total uncertainty for the site was 0.30, 0.40, and 0.47 for the 1.0m, argillic horizon, and 2.0m depths, respectively. The soil map and organic carbon content in the surface horizon were identified as the most uncertain parameters in predictive modeling, whereas the curve number and precipitation were the least sensitive parameters.;Classification of uncertainty using fuzzy set theory revealed little uncertainty in the predictions for mass emissions and travel time. Also, it was concluded that fuzzy set classification is not suitable in the format used for quantifying uncertainty in the mass emissions and travel times because of the excess number of fuzzy maps required and the ease with which bias can be introduced in the classification.;This study demonstrated that predictive pesticide modeling is uncertain when site-specific data are not taken into account, and it remains uncertain even if site-specific data are used, though the uncertainty will be less. |
