Heterogeneous Observation-Based Threshold Velocity Dataset for Wind Erosion and Its Implementation in the iDust Prediction System

Wind speed is one of the most sensitive meteorological factors influencing dust storm simulations. The dust emission threshold wind speed (ut) represents the critical point, beyond which dust emission increases notably as wind speed intensifies. However, ut exhibits spatial and temporal variability, lacks direct measurement compared to conventional meteorological parameters, and is challenging to estimate. This study integrates in-situ observations, satellite data, and reanalysis datasets to develop a global dust emission threshold wind speed dataset. Site-specific ut values are determined using in-situ observations by defining and optimizing a dust emission threshold score (DuTS). Based on these site-level ut values, along with satellite dust optical depth (DOD) and wind speed reanalysis data, a global ut distribution dataset is created. This dataset is implemented and validated in a NWP-grade global dust-weather integrated model, iDust. Evaluating iDust against particulate matter (PM)  concentration and DOD observations demonstrates that incorporating this dataset significantly improves the seasonal variation of dust simulations and enhances PM concentration simulations across multiple regions.