A new attempt for modeling erosion risks using remote sensing-based mapping and the index of land susceptibility to wind erosion

Abstract

Digital soil maps at proper scales can support sustainable land-use planning over large areas; however, applying this approach in erosion studies is still limited. Therefore, in this work, we adopted remote sensing-based mapping and fuzzy logic techniques to develop wind erosion risk maps with a spatial resolution of 30 m. These procedures were then conducted in a newly-reclaimed area in Matruh Governorate, the Egyptian western desert. Sixty surface soil samples (0–30 cm) were collected and analyzed for soil erodibility-related properties (sand, silt, clay, organic matter, and CaCO3). The relationships of these properties with the reflectance data of Landsat 8 multispectral bands were explored. Then, the spatial predictions were performed using the stepwise multiple linear regression models. The results showed that band reflectance in the shortwave infrared spectrum had higher correlations with soil properties than in the visible and near-infrared regions. The regression models achieved reliable precision and acceptable prediction abilities for all soil properties, except for the silt content. For validation datasets (30 % of the total data), the best-fitted models had coeffect of determination (R2) and residual prediction deviation of 0.61–0.92 and 1.42–2.67, respectively. The fuzzy memberships revealed various contributions of the five drivers (climate erosivity, soil erodibility, soil crust, vegetation, and terrain roughness) to potential soil loss. The success rate curves indicated the different performances of the applied fuzzy overlay operators (AND, OR, Sum, Product, and Gamma 0.9). The best-predictive model with an overall success rate of 85.7 % was obtained by overlaying the five fuzzified layers under the fuzzy algebraic Sum operator. This model delineated five hazard classes over the study area, including severe (79.5 %), high (12.6 %), moderate (5.7 %), slight (1.8 %), and very slight (0.4 %). The developed approach would improve the insight into large-scale monitoring of wind erosion status in desert ecosystems.