SENSITIVITY ANALYSIS OF SOIL AND PM10 LOSS IN WEPS USING THE LHS-OAT METHOD

Abstract

The Wind Erosion Prediction System (WEPS) was developed for the specific application of simulating erosionprocesses from agricultural lands. WEPS is a physically based model, with a moderate to large number of input parameters.Knowledge about model sensitivity is essential to both model developers and users in ascertaining those parameters mostinfluential to modeled object functions. A combined method of Latin hypercube sampling (LHS) and one-factor-at-a-time(OAT) parameter examination was used to assess the sensitivity of parameters in the WEPS erosion submodel in simulatingtotal soil loss, creep/saltation, suspension, and PM10 emission. The ranges of the parameters considered in this analysis wereobtained from the WEPS Users Manual and determined for the Columbia Plateau region of the U.S. Overall, the analysisindicated that the model was most sensitive to changes in biomass flat cover, near-surface soil water content, ridge height,wind speed, rock volume, soil wilting-point water content, field length and width, crust cover, aggregate and crust stability,and random roughness. The model was least sensitive to changes in bulk density, silt content, and aggregate and crust density.For the Columbia Plateau, erosion processes were more sensitive to surface soil water content and random roughness inspring than in autumn and more sensitive to residue cover and aggregate mean diameter in autumn than in spring. Thissensitivity analysis suggests that residue management, surface soil moisture conservation, aggregation, and field size caneffectively influence soil loss and PM10 emission.