Two different modeling approaches to simulate pesticide fate and transport in soil were investigated in this study. First, a process-based mathematical model, DRAINMOD-P, was developed by combining the attractive features of DRAINMOD and PESTFADE. While DRAINMOD formed the main component to perform hydrological predictions, PESTFADE's pesticide sub-model was used to simulate pesticide fate. The new model was validated against three years of independently collected field data from southern Ontario. Several statistical parameters were calculated to evaluate model performance. / Second, an implicit model, Multivariate Adaptive Regression Splines, MARS, which is also a novel data mining tool, was used to assess pesticide transport. MARS was first validated against the field data on three herbicides, namely, atrazine, metribuzin and metolachlor. DRAINMOD-P and MARS simulations, though impressive, need further validations before they can be recommended for actual real-world use.* / *This dissertation is a compound document (contains both a paper copy and a CD as part of the dissertation). The CD requires the following system requirements: Microsoft Office; Adobe Acrobat.

The work presented in this thesis represents a contribution to the area of modeling of the transport and fate of herbicides applied to cropped fields, and was part of a larger research effort geared towards better management of herbicides. The main objective of this thesis was to develop a graphical user interface (GUI) for PESTFADE, a process-based mathematical model of pesticide transport and degradation, and to provide documentation for the execution of PESTFADE. The model simulates changes in pesticide concentration at different depths in the soil, based on relevant physical, chemical, biological and meteorological factors. PESTFADE is considered to be one of the most comprehensive models of its kind. However, it was, until now, difficult to implement due to absence of a user manual and graphical interface suitable for exploitation in a Windows environment. The author developed the GUI in Visual Basic, created macros to facilitate certain calculations, rewrote the original FORTRAN 77 code and then validated the updated version against field data obtained from an experimental site (Eugene Whelan Farm, Woodslea, Ontario). A preliminary development of an artificial neural network (ANN) to perform the same simulation implicitly, with fewer input parameters and less computational time, was also done. / The thesis describes PESTFADE and the GUI, gives guidelines for implementing the package, and presents the results of the field validation of the revised version. During this work, the author discovered that there were problems in the parts of the code dealing with sorption phenomena. This can be solved by conventional kinetics or by Gamble kinetics.

No description available.

No description available.