In a paper published in National Science Review, a team of Chinese scientists develop an AI-powered framework designed to achieve real-time, seamless retrieval of PM₁₀ concentrations. This breakthrough addresses the challenges of spatial gaps and nighttime observation deficiencies in current satellite-based PM10 data. It extends daily data to high-resolution, real-time hourly insights, providing strong support for precise dust storm monitoring.

The findings reveal that, although the ensemble mean of the ECMWF model has limited forecasting ability for extreme cold events after two weeks, some ensemble members exhibit significantly high forecasting skill. The members with high forecasting skill can accurately predict the rapid change of surface air temperature and the intensity of the minimum temperature during an extreme cold event. This mainly depends on the accurate prediction of the atmospheric circulation situation in Eurasia (sea level pressure and 500-hPa geopotential height).