Long-Term Evaluation of AIRS Precipitable Water Vapor over China (2005–2025): Moisture-State-Dependent Bias and Temporal Consistency

Long-term satellite precipitable water vapor (PWV) records are widely used in climate studies, but excellent monthly agreement does not necessarily imply equivalent long-term climate-record consistency. We evaluated Atmospheric Infrared Sounder (AIRS) Version 7 Level-3 monthly PWV against Integrated Global Radiosonde Archive Version 2 (IGRA2) observations at 76 Chinese stations during 2005–2025, using 19,053 matched station-months. The overall agreement was high (R = 0.984, bias = −1.34 mm, RMSE = 3.36 mm). Nevertheless, station-by-calendar-month leave-one-year-out anomalies showed a robust moisture-state dependence of AIRS–IGRA differences (Spearman ρ = −0.722; 95% CI −0.741 to −0.704). The AIRS-on-IGRA OLS anomaly-response slope was 0.473; equal-variance Deming regression increased it to 0.560 (95% CI 0.519–0.596), still below unity. The ascending-minus-descending monthly PWV averaged +0.561 mm and was largest in summer. The network annual bias became more negative at −0.141 mm yr−1, compared with −0.138 mm yr−1 for a strict 73-station fixed network, while the annual R had no significant trend. The station-scale trend consistency was moderate (Pearson r = 0.545; sign agreement 65.8%). An independent BJFS GNSS comparison supported an AIRS contribution to local divergence, but did not establish nationwide attribution. These findings support monthly PWV climatology and broad variability applications over China while showing that small long-term trends require independent constraints, reference-record homogenization, and explicit attention to sampling stability.