Uncertainty and Sensitivity Analyses of the Modified Wind Erosion Equation for Application in Canada

The wind erosion equation (WEQ) was used for several decades for predicting soil loss by wind erosion, but few systematic studies on the uncertainty of predicting wind erosion have been conducted. Several researchers found that WEQ is not accurately representing soil erosion and under‐represents soil erodibility that consequently results in WEQ underestimations. WEQ was modified under Canadian conditions for the National Agri‐Environmental Health Analysis and Reporting Program, referred as (WEQ‐NAHARP). The model, input, and parameter uncertainties that control model efficiency were used to analyze the local and universal uncertainties for WEQ‐NAHARP. One and ninety‐nine percentiles were used as lower and upper boundaries of uncertainty bound when using general likelihood uncertainty estimation for estimating the uncertainty of WEQ‐NAHARP's prediction. The soil erodibility (I ), climate factor (C ), and soil surface roughness factor (K ) were found as the three most sensitive factors in predicting wind erosion in WEQ‐NAHARP. The vegetation cover factor (V ) was discovered not sensitive to the prediction model as it is less than 1,000 kg ha⁻¹ and became very sensitive as V ‐value is greater than 5,000 kg ha⁻¹ . Field length along the prevailing wind erosion direction (L ) and V have lower local sensitivity indexes than the other three factors. WEQ‐NAHARP underestimated wind erosion rate of Pampas, Argentina, and overestimated at Washington State, USA. This probably reflected the nature of WEQ‐NAHARP's behavior, which had a great uncertainty of its prediction. The model appears to underestimate total annual soil loss for coarse soil and overestimate annual soil loss for finer soil. Copyright © 2017 John Wiley & Sons, Ltd.