The Channel‐Correlated Error Covariance Matrix for FY‐4B GIIRS and Its Application in the CMA‐BJ v2.0 Assimilation and Forecast System

Abstract The Fengyun‐4B (FY‐4B) satellite's Geostationary Interferometric Infrared Sounder (GIIRS), featuring thousands of hyperspectral channels and a persistent viewing geometry, induces stronger inter‐channel observational error correlations (IOEC). This necessitates, explicit characterization of IOEC to optimize its impact on data assimilation. This study presents the first comprehensive assessment of GIIRS IOECs, demonstrating their potential to improve the assimilation of GIIRS observations for numerical weather prediction (NWP). Results indicate that a more appropriate characterization of GIIRS error covariances enables IOEC to assign more optimal observational weights, leading to improved atmospheric state estimates. Cycling assimilation and forecasting experiments based on the CMA‐BJ operational NWP system further demonstrate that IOEC facilitates more accurate temperature and water vapor analysis by adjusting the magnitude and structure of the analysis increments. This, in turn, leads to an improvement of up to 2% in forecast skill for key variables and enhances 24‐hr precipitation prediction scores.