Owing to their high vertical resolution, remote sensing data from meteorological satellite hyperspectral infrared sounders are well-suited for the identification, monitoring, and early warning of high-impact weather events. The effective utilization of full field-of-view (FOV) observations from satellite infrared sounders in high-impact weather applications remains a major research focus and technical challenge worldwide. This study proposes a generalized variational retrieval framework to estimate full FOV cloud fraction and precipitable water vapor (PWV) from observations of the Geostationary Interferometric Infrared Sounder (GIIRS) onboard the Fengyun-4A (FY-4A) satellite. Based on this method, experiments are performed using high-frequency FY-4A/GIIRS observations during the landfall periods of Typhoon Lekima (2019) and Typhoon Higos (2020). A three-step channel selection strategy based on information entropy is first designed for FY-4A/GIIRS. A constrained generalized variational retrieval method coupled with a cloud cost function is then established. Cloud parameters, including effective cloud fraction and cloud-top pressure, are initially retrieved using the Minimum Residual Method (MRM) and used as initial cloud information. These parameters are iteratively optimized through cost-function minimization, yielding full FOV cloud fields and atmospheric profiles. Full FOV brightness temperature simulations are conducted over cloudy regions to quantitatively evaluate the retrieved cloud fractions, and the derived PWV is further applied to the identification and analysis of hazardous weather events. Experimental results demonstrate that incorporating cloud parameters as auxiliary inputs to the radiative transfer model improves the simulation of FY-4A/GIIRS brightness temperature in cloud-covered areas and reduces brightness temperature biases. Compared with ERA5 Total Column Water Vapour (TCWV) data, the PWV derived from full FOV profiles containing cloud parameter information shows closer agreement and, at certain FOVs, more effectively indicates the occurrence of high-impact weather events. The simplified methodology proposed in this study provides a robust basis for the future assimilation and operational utilization of infrared data over cloud-affected regions in numerical weather prediction models.

The hyperspectral infrared observations of the Geostationary Interferometric Infrared Sounder (GIIRS) on the Fengyun-4A (FY-4A) satellite are an important data source for numerical weather prediction (NWP) assimilation. However, there are systematic differences between observed and simulated brightness temperatures (i.e., the observation increments contain predictable systematic bias components). To address the issue that traditional linear methods struggle to capture the nonlinear relationships between biases and forecast predictors, this study proposes an intelligent bias correction method that integrates ensemble learning and explainable artificial intelligence. First, the entropy reduction method is used to select 69 mid-wave channels. Then, Random Forest, XGBoost, LightGBM, Decision Tree, and Extra Tree are used as base learners to construct a weighted average ensemble model. Training and validation are conducted using high-frequency clear-sky observation data from FY-4A/GIIRS during Typhoon Lekima. The results show that: (1) the ensemble learning correction method outperforms single models and traditional offline methods, with root mean square errors of brightness temperature bias of less than 0.9209 K for the training set and 1.4447 K for the test set; (2) Shapley Additive Explanations (SHAP)-based interpretability analysis reveals the contribution and nonlinear influence mechanisms of factors such as longitude, atmospheric thickness, surface temperature, and total precipitable water on bias correction. This study provides an intelligent bias correction framework with both high precision and explainability, offering a reference for the bias correction and assimilation applications of hyperspectral satellite observations like GIIRS.