A First Attempt at Reconstructing FengYun‐4B Stratified Precipitable Water Using GNSS

Abstract Layer Precipitable Water (LPW) characterizes the vertical structure of atmospheric moisture and is essential for accurate weather forecasts. China's FY‐4B satellite delivers near‐real‐time LPW products, but is constrained by large uncertainties. To address these limitations, we first integrated spherical cap harmonic analysis with extreme gradient boosting to enhance the Total Precipitable Water (TPW), and then calibrated the LPW by proposing a novel proportional allocation model considering spatiotemporal variability. Validation against radiosonde indicates that the root‐mean‐square errors of the FY‐4B LPW were reduced from 3.0 to 1.9 mm in the low layer, 4.2 to 2.3 mm in the mid layer, and 2.5 to 1.8 mm in the high layer, with the bias reduced from a maximum of −1.5 mm to near zero. Further evaluation with ERA5 demonstrates enhanced spatial consistency of the modified product. This work fills the gap of high‐accuracy LPW data, and supports more accurate forecasting and early‐warning applications.