Dust storms are among the most critical atmospheric hazards in arid and semi-arid regions, and Isfahan Province is highly vulnerable due to climate change, declining precipitation, and recurrent droughts. This study aimed to model and zone dust hazard susceptibility in Isfahan Province during 1995–2021 by integrating geomorphological features and employing Support Vector Machine (SVM) and Generalized Linear Model with Polynomial terms (GLM-POLY) algorithms. Hourly dust data were extracted from synoptic stations, and climatic variables (precipitation, temperature, relative humidity, maximum wind speed) alongside terrestrial layers (geology, soil, land use, NDVI, TWI) were prepared using GIS techniques. Multicollinearity assessment via Variance Inflation Factor (VIF < 5) confirmed variable independence. Variable importance analysis using Mean Decrease in Accuracy identified maximum wind speed (27.5%), precipitation (26.2%), and relative humidity (25.4%) as the most influential factors. Model performance evaluation using the ROC curve and AUC index revealed that SVM (AUC = 0.712) demonstrated "good" predictive capability, significantly outperforming GLM-POLY (AUC = 0.621) with "moderate" performance. The final hazard zonation map using the superior SVM model, classified into four categories, indicated that the eastern, southeastern, and central parts of the province exhibit the highest susceptibility, consistent with barren lands, saline soils, and poor rangelands, while western areas show lower risk due to higher precipitation and vegetation cover. SVM's superiority is attributed to its capacity to handle complex non-linear interactions. This research provides a practical scientific tool for environmental planning and land-use management in Isfahan Province and recommends that future studies explore deep learning approaches and incorporate real-time data for dynamic forecasting.