A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering.

Thesis (M.Eng.) - Asian Institute of Technology, 2000

The present study aims at evaluating the Water Erosion Prediction Project (WEPP) model applicability and prediction accuracy for a small watershed in the East of Thailand. Three 5 by 22.1 m plots with 5% slope from seventeen runoff and erosion experimental plots established by the Land Development Department in the Rayong Research Station were selected for the study to simulate three different land use treatments. Runoff and soil loss data were collected over a period of 3 months (August-October 1999). Observed rainfall and observed runoff and soil loss data were analyzed for their relationships. The WEPP model was run in a continuous simulation mode for 4 months from July 1999 to October 1999 for each plot and the results were compared with the observed data. The USLE model was also applied in the study for soil loss prediction and the results were compared with the observed data and WEPP predicted results. The runoff amount and runoff coefficients from the traditional plot (Plot 1) were higher than the conservation plots (Plot 2 and Plot 3). WEPP produced slightly acceptable results for runoff prediction though it overpredicted total runoff for all plots. Considering the existing observed soil loss data, WEPP overpredicted total soil loss for all plots as well as USLE. From the regression results, USLE tended to predict soil loss less than WEPP in Plot 1 but it tended to predict soil loss more than WEPP in Plot 2 and 3. It is not explicitly addressed in this study that how the models gave the predictions for soil loss as a number of observed data were not enough. Application of WEPP to a sub-watershed of Dokkrai Watershed in the eastern part of Thailand showed the close prediction of runoff for one available observed event though it could not be concluded that how was the model ability in prediction for the watershed. The results of sensitivity analysis showed that rill erodibility, critical shear, rill cover, precipitation amount and duration, slope length and slope gradient are major factors in terms of model response to detachment.