Sensitivity experiments were conducted on Typhoon Mujigae, which occurred in 2015, wherein the Weather Research and Forecasting Advanced Research (WRF-ARW) model was used to select two local and two nonlocal planetary boundary layer (PBL) parameterization schemes: the quasi-normal scale elimination (QNSE) and Mellor–Yamada–Janjic (MYJ) schemes, and the Yonsei University (YSU) and medium-range forecast (MRF) schemes, respectively. The differences in rainfall response in the typhoon’s inner core and outer region were evaluated by comparing the anomaly rainfall distribution, heat transmission, and mixing processes in the boundary layer among the PBL schemes. The results show that the simulated rainfall in typhoon Mujigae has large uncertainty among the PBL schemes and a significant difference between the inner and outer regions. Compared with the observation, the simulated rainfall was significantly higher in the inner core and slightly lower in the outer region. All PBL schemes accurately identified the rainfall location, although the amounts differed between the schemes. The rainfall levels in the MRF scheme were closest to the observation, followed by those in the YSU and MYJ schemes; the QNSE scheme showed the largest deviation. In general, rainfall simulation using a nonlocal boundary layer scheme such as MRF had the best results for both the inner core and the outer region.

Variational data assimilation theoretically assumes Gaussian-distributed observational errors, yet actual data often deviate from this assumption. Traditional quality control methods have limitations when dealing with nonlinear and non-Gaussian-distributed data. To address this issue, our study innovatively applies two advanced machine learning (ML)-based quality control (QC) methods, Minimum Covariance Determinant (MCD) and Isolation Forest, to process precipitable water (PW) data derived from satellite FengYun-2E (FY2E). We assimilated the ML QC-processed TPW data using the Gridpoint Statistical Interpolation (GSI) system and evaluated its impact on heavy precipitation forecasts with the Weather Research and Forecasting (WRF) v4.2 model. Both methods notably enhanced data quality, leading to more Gaussian-like distributions and marked improvements in the model’s simulation of precipitation intensity, spatial distribution, and large-scale circulation structures. During key precipitation phases, the Fraction Skill Score (FSS) for moderate to heavy rainfall generally increased to above 0.4. Quantitative analysis showed that both methods substantially reduced Root Mean Square Error (RMSE) and bias in precipitation forecasting, with the MCD method achieving RMSE reductions of up to 58% in early forecast hours. Notably, the MCD method improved forecasts of heavy and extremely heavy rainfall, whereas the Isolation Forest method demonstrated a superior performance in predicting moderate to heavy rainfall intensities. This research not only provides a basis for method selection in forecasting various precipitation intensities but also offers an innovative solution for enhancing the accuracy of extreme weather event predictions.

Variational data assimilation theoretically assumes Gaussian-distributed observational errors, yet actual data often deviates from this assumption. Traditional quality control methods have limitations when dealing with nonlinear and non-Gaussian-distributed data. To address this issue, our study innovatively applies two advanced machine learning (ML) based quality control (QC) methods, Minimum Covariance Determinant (MCD), and Isolation Forest to process precipitable water (PW) data derived from satellite FengYun-2E (FY2E). We assimilated the ML QC-processed TPW data using the Gridpoint Statistical Interpolation (GSI) system and evaluated its impact on heavy precipitation forecasts with the Weather Research and Forecasting (WRF) v4.2 model. Both methods notably enhanced data quality, leading to more Gaussian-like distributions and marked improvements in the model’s simulation of precipitation intensity, spatial distribution, and large-scale circulation structures. During key precipitation phases, the Fraction Skill Score (FSS) for moderate to heavy rainfall generally increased to above 0.4. Quantitative analysis showed that both methods substantially reduced Root Mean Square Error (RMSE) and bias in precipitation forecasting, with the MCD method achieving RMSE reductions of up to 58% in early forecast hours. Notably, the MCD method improved forecasts of heavy and extremely heavy rainfall, whereas the Isolation Forest method demonstrated superior performance in predicting moderate to heavy rainfall intensities. This research not only provides a basis for method selection in forecasting various precipitation intensities, but also offers an innovative solution for enhancing the accuracy of extreme weather event predictions.