ABSTRACT
The use of simulation models to address environmental problems is now a wide spread practice (van der Heijde and Elnawawy, 1993). Yet, these models are limited by the assumptions inherent in their development, the availability and uncertainty of required input data, and the consequent reliability of output predictions. These issues are both compounded and confounded by the fact that spatial and temporal scales of variability dictate the availability and quality of data and therefore the appropriate modeling choices. As the constraints of available data at a particular scale are rec ognized, there must also follow a consideration of the degree of determinism and mechanism that can be invoked in the modeling process. One unfortunate conse quence is a continuing use of simulation models derived for use at one scale to predict effects at larger scales. This process is a form of extrapolation of limited information on natural processes beyond the limits of scientific understanding. This type of misuse of models is important and widespread in vadose zone simulation modeling. Partic ularly important examples are the numerous pesticide leaching models which are now used to estimate pesticide transport and transformation at spatial scales from the soil profile to the regional, and for temporal scales from seconds to decades.
