علی آذره

Soil erosion is a significant threat to global food production, reducing the productivity of natural ecosystems and agricultural lands. In this study, we examine the loss of crop water productivity in the Halil River agricultural watershed, Iran. Areas within this watershed that are prone to soil erosion were delineated using two machine learning algorithms viz. SupportVector Machines (SVM) and Multivariate DiscriminantAnalysis (MDA), and 11 geo-environmental factors including elevation, lithology, land use, hydrologic soil group, soil depth, drainage density, soil available water capacity (SAWC), population density, slope (degrees), R factor, and distance to road. Finally, the loss of blue, green, and total crop water productivity of some main cultivated crops (wheat, dates, citrus, and tomatoes) in the study watershed was assessed under pessimistic (20%), optimistic (10%), and normal (15%) scenarios of crop water productivity loss. The results indicate that hydrologic soil group, elevation, and land use are the most important factors for soil erosion susceptibility. In addition, validation results of machine learning algorithms showed that the SVM model (AUC=94%,TSS=0.85) outperformed MDA (AUC=92.3%,TSS=0.81), and was therefore selected for further analysis.According to the SVM model, 14.3% of the watershed falls within the very high erosion susceptibility class.Agricultural lands are mostly located in areas of moderate to very high erosion risk. Production of wheat, citrus, dates, and tomatoes in moderate, high, and very high areas of erosion susceptibility map are estimated to be 2.4, 5.3, 13.3, and 10.7 million tons, respectively. In the optimistic scenario, total productivity losses per unit of water consumed water by wheat, dates, citrus, and tomatoes are 0.22, 1.24, 2.15, and 2.35 kg m−3. and total economic loss will be 14,371, 93,675, 83,247 and 51,893(×104) US$. In the more realistic scenario, these losses are, respectively, 0.32, 1.86, 3.23, and 3.53 kg per unit of consumed water and the economic loss are 21,557, 140,517, 124,872 and 77,841(×104) US$.The total productivity losses per unit of water consumed in the pessimistic scenario are, respectively, 0.43, 2.47, 4.30, and 4.70 kg m−3 and total economic loss are 28,743, 187,353, 166,494 and 103,782 (×104) US$. The results underscore the urgent need for site-specific erosion mitigation strategies to safeguard agricultural productivity and water efficiency.