Abstract
The value of the parameters in environmental fate model of chemicals has not been well examined. This paper discusses about that. The purpose of this paper is to evaluate the parameter uncertainty using Bayes theorem with the example of agricultural chemicals runoff in the test watershed. Bayeisian Monte Carlo technique is used to quantify errors in chemical fate model caused by uncertain parameters. Often there are a few site-specific data that describes the model parameters. The only information typically available for many parameters is a range of values obtained from related published studies. This technique combines Monte Carlo analysis with Bayesian inference specifying model input parameter distirbutions. The statistical likelihood function is employed to evaluate the ability of any given set of model parameters to describe the observed data on model state variables. Preliminary information on parameter distribution is combined with measurement of state variables to provide improved estimates of parameter distribution. This method was applied to a model developed to evaluate the pesticide runoff process on the Shibuta river, Kanagawa prefecture. The model used is one dimensional and purely advective and one state variable. Six parameters (partition coefficient, degradation rate constant in water, degradation rate constant in soil, soil density) were evaluated.
