Application of a Bayesian belief network model for assessing the risk of wind erosion: A test with data from wind tunnel experiments

Abstract

Abstract The complexity of the interactions between drivers in wind erosion processes and the absence of adequate and reliable data are major constraints to achieving a quantitative assessment of wind erosion. Bayesian Belief Networks (BBNs) provide a useful approach to address real-world problems, where available data and knowledge are disparate, limited or uncertain. We investigated the potential use of BBNs to assess soil erosion risk in a typical arid region that experiences severe wind erosion. The developed framework was based on a standard risk assessment procedure, where the risk of wind erosion was quantified by incorporating assessments of consequence and vulnerability. Performance of the constructed model was evaluated using scenario testing, sensitivity analysis, and wind-tunnel measurements. The model provided reasonable estimates of the soil vulnerability, consequence, and risk to/of wind erosion. The results showed that weather and management factors were the most important parameters affecting wind erosion risk. Based on the fitted regression lines, there were positive (R2\u202f=\u202f0.82) and negative (R2\u202f=\u202f0.72) relationships between the measured wind erosion rates and the predicted probabilities to ‘high’ and ‘low’ vulnerability classes, respectively.