A Segmented Classification and Regression Machine Learning Approach for Correcting Precipitation Forecast at 4–6 h Leadtimes

Accurate and fine-scale short-term precipitation forecasting is crucial for disaster prevention, mitigation, and socioeconomic development. Currently, direct precipitation forecasts by numerical weather prediction models often face great challenges, and correction methods are still needed to further improve the forecast accuracy. By utilizing the 500-m resolution fusion precipitation data from the Rapid-refresh Integrated Seamless Ensemble (RISE) system in the Beijing–Tianjin–Hebei (BTH) region, this study proposes a new Segmented Classification and Regression machine learning model by extreme gradient boosting (XGBoost) algorithm, termed SCR-XGBoost, which can be applied to correct hourly precipitation forecast at dense-station areas for 4–6 h ahead. The performance of the model is evaluated by using six metrics including accuracy (AC), mean-absolute-error (MAE), root-mean-square-error (RMSE), correlation coefficient (CC), threat score (TS), and bias score (BS). The results indicate that although the XGBoost algorithm is almost ineffective for direct precipitation forecast, the SCR-XGBoost model can significantly improve the forecast performance compared with the original RISE forecast, and the segmented correction method for torrential rainfall (>= 20 mm h⁻¹) outperforms other precipitation grades, which can effectively improve the problem of false alarms from the RISE system for heavy rainfall (10 mm h⁻¹) and above. The optimization rates after the SCR-XGBoost model correction in precipitation forecast can be improved by 6.49%–23.21% in terms of RMSE and MAE reduction, and the CC and AC can be greatly improved by 35.38%–84.39%. Therefore, the SCR-XGBoost algorithm that introduces precipitation grades classification and multi-layer piecewise machine learning corrections, can significantly improve the 4–6 h precipitation forecast skill, especially for the heavy rainfall. The results of this paper not only provide a new clue for machine learning precipitation forecast, but also help improve rainfall forecast and the level of disaster prevention and reduction in the BTH.