Mineral dust significantly affects air quality, visibility, and Earth's radiation balance. Dust storms frequently occur in arid, semi-arid lands, flat regions with erodible soils, where drought and land-use changes have increased their occurrence, harming agriculture and communities. Central Eurasia, particularly the Middle East, is a major dust source region. This study employed machine learning to evaluate dust emission susceptibility in South Khorasan, Iran, by analyzing environmental factors and enhancing existing dust prediction models. Researchers used land use/land cover (2004 and 2019) maps, lithology, elevation, and climate variables from ACCESS-CM2 and CANESM5 models under IPCC6's SSP5-8.5 scenario to predict dust source susceptibility. Among SVM, CART, and Linear Regression algorithms, Random Forest performed best for LULC classification and wind speed prediction. The study combined CA-Markov for LULC prediction with Maximum Entropy modeling to calculate the Dust Source Susceptibility Index (DSSI). Results showed CANESM5 projected higher dust susceptibility than ACCESS-CM2, with over 10,340 km² falling into the highest-risk DSSI category. Wind plays a determining role in starting dust storms. The research demonstrates that integrating multiple modeling approaches and validation metrics (Kappa, AUC, R²) provides an effective framework for investigating dust source susceptibility, offering improved predictive capability for dust storm management and mitigation strategies.
Ebrahimi, A. and Ahmadizadeh, S. (2025). Predicting Dust Sources Susceptibility Using Machine Learning Techniques in the Future (Case Study: South Khorasan, Iran).. Desert, 30(1), 184-203. doi: 10.22059/jdesert.2025.104762
MLA
Ebrahimi, A. , and Ahmadizadeh, S. . "Predicting Dust Sources Susceptibility Using Machine Learning Techniques in the Future (Case Study: South Khorasan, Iran).", Desert, 30, 1, 2025, 184-203. doi: 10.22059/jdesert.2025.104762
HARVARD
Ebrahimi, A., Ahmadizadeh, S. (2025). 'Predicting Dust Sources Susceptibility Using Machine Learning Techniques in the Future (Case Study: South Khorasan, Iran).', Desert, 30(1), pp. 184-203. doi: 10.22059/jdesert.2025.104762
CHICAGO
A. Ebrahimi and S. Ahmadizadeh, "Predicting Dust Sources Susceptibility Using Machine Learning Techniques in the Future (Case Study: South Khorasan, Iran).," Desert, 30 1 (2025): 184-203, doi: 10.22059/jdesert.2025.104762
VANCOUVER
Ebrahimi, A., Ahmadizadeh, S. Predicting Dust Sources Susceptibility Using Machine Learning Techniques in the Future (Case Study: South Khorasan, Iran).. Desert, 2025; 30(1): 184-203. doi: 10.22059/jdesert.2025.104762
)