Abstract A variational quality control (VarQC) method based on a non‐Gaussian error model plays a crucial role in enhancing the assimilation of satellite radiance data by dynamically adjusting observation weights according to observation quality. Here, we adapted a VarQC scheme within the China Meteorological Administration‐mesoscale model (CMA‐MESO) to improve the efficiency and effectiveness of assimilating clear‐sky water vapor brightness temperatures (WV‐BTs) from the infrared channels of the Advanced Geosynchronous Radiation Imager (AGRI) aboard Fengyun‐4A (FY‐4A). It was first validated through idealized single‐point experiments using both real and perturbed WV‐BT observations under clear‐sky conditions. After applying preprocessing, including thinning, quality control, and bias correction, sensitivity experiments were conducted to determine the appropriate prior probability of gross errors, thereby establishing a reasonable non‐Gaussian error model for clear‐sky infrared WV‐BTs in VarQC. Two typical heavy rainfall cases in South China were selected for 10‐day cycling assimilation experiments to evaluate the impact of the VarQC scheme on precipitation forecasts. The results show that the VarQC scheme adaptively adjusts the clear‐sky WV‐BT weights based on the innovation magnitude, preserving valuable information while limiting the impact of low‐quality observations, thus balancing observation quality and analysis accuracy. Compared with those of the control experiments, VarQC yielded boundary‐layer geopotential height, specific humidity, and wind fields more consistent with sounding observations and ERA5 reanalysis, particularly over oceanic and coastal–inland transition zones. Furthermore, it significantly improved the 12–24 h rainfall forecasting skills in South China. We provide a foundation for future VarQC applications for all‐sky radiance assimilation from the FY satellite series.
He et al. (Tue,) studied this question.