Assessing uncertainties in WEPP's soil erosion predictions on rangelands

ABSTRACT:

An assessment of uncertainties in predictions was performed for the WEPP model by identifying the contribution of parameter variance and model bias on the Mean Square Error of three model response variables. A Monte Carlo simulation scheme with correlated variable generation was assembled into WEPP to produce a large number of model responses to be compared to observed data of a semi-arid rangeland watershed. The behavior of errors and prediction intervals illustrate the model's ability to simulate the system. This analysis answered relevant questions related to model uncertainty and to identify noisy components in the model.