Spatiotemporal Evolution Characteristics of Atmospheric Humidity over Xinjiang during 1991–2020 Based on Radiosonde Data and Its Dominant Factors

Understanding global warming’s impact on atmospheric specific humidity (SH) in arid regions is crucial for hydrological prediction and water sustainability. Based on homogenized radiosonde data from 14 Xinjiang, China, stations and ERA5 reanalysis (1991–2020), this study investigates SH’s spatiotemporal evolution and governing mechanisms. Key findings are as follows: (1) the mid-to-lower troposphere (850–500 hPa) shows significant drying, with summer 850 hPa drying being the most severe (−1.04 g·kg–1); 700 hPa interannual variability is ENSO-modulated (El Niño: +0.32 g·kg–1, La Niña: −0.25 g·kg–1), amplifying by 30% post-2010. (2) A spatial “wet south–dry north” pattern exists, with Minfeng Station showing the strongest moistening (+0.1864 g·kg–1/decade) due to glacial melt and topography. (3) Iranian High is the dominant moisture source (66.6% contribution), while thermodynamic effects (68%, r = 0.73) are lower than Clausius–Clapeyron expectations. (4) A “wind-speed increase–humidity decrease” trend in southwesterly transport indicates aridification risks. This study provides critical insights for regional climate projections and water resource management in arid Central Asia.