Abstract Dust storms (DSs) have significant impacts on human health, air quality, and terrestrial ecosystems. In recent years, the increasing frequency of extreme DSs has underscored the urgent need to better understand their historical characteristics and dominant drivers. Here, we examined a 64‐year observational record (1961–2024) of springtime DS in China, emphasizing the spatiotemporal variations of DS across different intensities. A significant nationwide decrease in the frequency of DSs has been observed, primarily attributed to a reduction in lower‐intensity (ODS; ordinary DS) events. The proportion of severe and extreme dust storms (SEDS) has significantly increased since the early 2000s, suggesting a transition toward fewer but more intense occurrences. Using machine learning with interpretable methods, we show that ODS are primarily driven by synoptic‐scale dynamics, whereas SEDS are more sensitive to land‐surface deficits. SEDS frequency increases sharply when evaporation reaches extremely low values (<50 mm) and top‐layer soil temperature exhibits rapid warming (exceeding 290 K), These conditions indicate a precipitation‐limited surface state, with strong winds serving as the final trigger for SEDS. The increases in SEDS since 1990 in northwestern China coincide with intensifying spring drought and strengthening wind extremes, while in North China, despite only modest increases in wind speed, persistent surface drying has sustained elevated SEDS activity. In contrast, the eastern Tibetan Plateau exhibits a clear decline, consistent with progressive wetting and a notable weakening of near‐surface winds. Our findings identify early‐warning predictors that can improve the interpretation and development of future dust‐storm projections under climate change.
Bao et al. (Sat,) studied this question.