<p>This study presents a scalable framework for regional wind-erosion assessment by integrating a harmonized global wind-tunnel database with satellite-derived soil and climatic predictors. The approach is designed to support wind-erosion mapping in data-scarce drylands where extensive field measurements are difficult to obtain. By combining experimental erosion measurements with globally available gridded datasets, the framework enables spatial identification of erosion-prone areas and supports prioritization of field monitoring and soil-conservation planning. To ensure robust generalizability, our model, unlike previous localized studies, was trained on a comprehensive empirical dataset of 465 physical wind tunnel experiments (both laboratory and portable field measurements) covering a wide range of diverse soil types and climatic conditions. This empirical data was integrated with environmental variables extracted from global satellite imagery and open-source platforms, including NASA-GLDAS for meteorological data and OpenLandMap for soil properties, to create a unified database for rapid analysis and root-cause determination of wind erosion. The results demonstrated that the Random Forest (RF) model is the most suitable predictive tool, achieving 86% accuracy and superior computational performance compared to Support Vector Regression (SVR). Feature importance analysis conclusively identified Sand Content and Wind Velocity as the most influential factors governing erosion susceptibility on a large scale. Successful implementation of this framework in the Sistan and Baluchestan Province (Iran) over a decade (2015–2024) validated the method’s capacity for determining the wind erosion magnitude in the shortest possible time and accurately predicting erosion dynamics under specific climate change conditions (e.g., during prolonged drought periods).</p>