Abstract: Improving initial moisture characterization is crucial for predicting tropical cyclone evolution and related extreme precipitation, particularly as sparse terrestrial observations are often insufficient to capture complete moisture structures. In this study, Precipitable Water Vapor (PWV) data from the AMSR2 aboard the Global Change Observation Mission 1st-Water (GCOM-W1) satellite sensor and the Global Navigation Satellite System (GNSS) ground stations were assimilated into the WRF model using a multi-scale three-dimensional variational (3D-Var) framework. To bridge the scale discrepancy between maritime footprints and terrestrial point observations, differentiated horizontal influence radii were implemented. The impact of this strategy was examined using the example of Typhoon Haikui (2023). Results indicated that the joint assimilation effectively optimized the tropical cyclone’s thermodynamic structure by correcting terrestrial moist biases and maritime dry biases, leading to a more realistic land-ocean moisture gradient. Validation against Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG) satellite data showed that the multi-scale joint assimilation outperformed the control and single-source experiments. At the 100 mm threshold for 24-h accumulated rainfall, the joint scheme improved Probability of Detection (POD), Critical Success Index (CSI), and Equitable Threat Score (ETS) by 23.4%, 12.2%, and 12.1%, respectively, while effectively limiting the False Alarm Ratio (FAR) increase to only 5.4%. Independent verification against China Meteorological Administration (CMA) automatic weather station observations showed consistent improvements in rainfall forecast skill. This research demonstrates that the proposed multi-source, multi-scale framework is effective for capturing critical moisture transport and enhancing convection-permitting forecasts of landfalling typhoons.
 

Precipitable water vapor (PWV); Satellite microwave observation; Multi-scale 3D-Var; Typhoon Haikui (2023); Extreme precipitation.