A data-driven approach for assessing the wind-induced erodible fractions of soil

To develop an effective strategy for controlling wind erosion and soil degradation, it is necessary to identify the regions with the greatest wind erosion potential. In this regard, many wind erosion models are available that can be used to estimate the rate of wind erosion, allowing erosion control strategies to be assessed. A major factor in all wind erosion models is the inherent erodibility of soil. As it has been proven that the wind erodible fraction of soil (EF) is closely related to its erodibility, this parameter is of importance and used in many wind erosion models such as WEQ, RWEQ, EPIC, and APEX. To evaluate the effect of key soil parameters such as contents of sand, silt, clay, organic matter, and calcium carbonate on the EF, a dataset consisting of 293 samples was compiled from peer -reviewed studies. Initially, to evaluate the relationship between main soil parameters and the EF, Pearson correlation coefficients were calculated for the soil parameters. The results indicate that soil texture has a more significant impact on the EF than the contents of organic matter and calcium carbonate. Moreover, a total of six equations have been identified within the existing body of the literature for the purpose of predicting the soil EF. Through an evaluation of the performance of the existing equations, it was observed that their accuracy varies, often overestimating or underestimating the EF of soil, indicating their overall unsatisfactory performance. To address this issue, regression analysis was employed to propose an equation based on the compiled dataset, which provides satisfactory accuracy in EF prediction.