Calibration, Validation, and Evaluation of the Water Erosion Prediction Project (WEPP) Model for Hillslopes with Natural Runoff Plot Data

Abstract

<b><sc>Abstract </sc></b> The Water Erosion Prediction Project (WEPP) model is a physically-based erosion computer simulation program, which has been used worldwide for runoff and soil loss estimation. Given the wide range of applications supported by the WEPP model, there is a continuing need to evaluate the performance of the latest versions to determine model prediction efficiency and provide confidence to users. The objectives of this study were to evaluate the latest WEPP model (version 2021.133) for runoff and soil loss predictions for hillslopes with natural runoff plot data. <fig><graphic xlink:href=23022_files/23022-01.jpg id=ID_3c0a3599-43ea-45e3-be15-0b891f9dc11d></graphic></fig> There were 1159 plot years of data from the USLE database utilized, which were obtained from 134 field experimental plots in the USA (Figure 1) with various climates, soils, topographies, and crops. WEPP model runs were conducted in both uncalibrated and calibrated modes. The runoff and soil loss predictions were compared to the observations. Four parameters were calibrated based on the least squares error criterion including baseline effective hydraulic conductivity, baseline critical shear stress, baseline rill erodibility, and baseline interrill erodibility. The baseline effective hydraulic conductivity was calibrated using all 1159 plot years of data to obtain the best runoff predictions. The other three parameters were calibrated and validated using 607 and 552 plot years, respectively. The results showed good predictions of runoff and soil loss were obtained with both the uncalibrated and calibrated WEPP model, with Nash-Sutcliffe model efficiencies (NSE) over 0.7 for average annual estimates. <fig><graphic xlink:href=23022_files/23022-00.jpg id=A9838FFE-8582-4FAD-A0F2-39D56E9DFA73></graphic></fig> The uncalibrated results are shown in Figure 2. The calibration of input baseline effective hydraulic conductivity, baseline critical shear stress, baseline rill erodibility, and baseline interrill erodibility improved WEPP model performance with the NSE values of 0.98 and 0.91 for average annual runoff and soil loss, respectively. The WEPP model tended to underestimate the runoff and soil loss for large events with runoff over 100 mm and soil loss over 120 t/ha. Good runoff predictions (NSE ≥ 0.4) were obtained for almost all crops considered. Good event predictions (NSE ≥ 0.4) of soil loss were obtained for tilled fallow, corn, wheat, and cotton crops. Overall, the WEPP model performed well for estimating runoff and soil loss, which could be improved through calibration. Model performance varied for different crops, but adequate predictions were obtained for the most common crops. This study illustrates the most recent WEPP model‘s performance for runoff and soil loss predictions, and it is the most comprehensive validation of the WEPP model to date.