Seasonal Prediction of Pre-Summer Clustered Heavy Precipitation Days in South China

Clustered heavy precipitation (CHP) events can severely impact human society, infrastructure, and natural ecosystems. Consequently, short-term climate prediction of CHP events is vital for the prevention and mitigation of associated hazards. Employing year-to-year increment (DY) and multiple linear regression approaches, this study developed a seasonal prediction model for pre-summer (i.e., May and June) CHP frequency in South China (SC) during 1981– 2022. Three robust predictor factors were identified: March sea surface temperature in Southwestern Atlantic, early-winter snow depth in East Europe, and winter soil moisture in Central Asia. Three predictors exert substantial impacts on presummer precipitation in SC via modulation of an anomalous anticyclone (cyclone) over the (subtropical) western North Pacific. In leave-one-out cross-validation test during 1981–2022, the prediction model exhibited reasonable performance in predicting the interannual and interdecadal variations and trends of CHP days. The temporal correlation coefficient (TCC) was 0.66 between the observations and predictions. In the independent hindcast for 2013–2022, the TCC was as high as 0.85. Moreover, coherent covariations were observed between the frequency and the amounts of CHP, with a TCC of 0.99 for 1981–2022. Those three predictors show good performance in forecasting CHP amounts over SC, with a TCC of 0.68 between the predictions and observations in the cross-validation test during 1981–2022 and of 0.86 in the independent hindcasts during 2013–2022. Notably, the predictors also showed good predictive skill for years with high CHP occurrence (e.g., 1998 and 2019). The predicted high-incidence areas of heavy precipitation days were highly consistent with observations, with a pattern correlation coefficient of 0.44 (0.55) for 1998 (2019). This study provides valuable insights to improve seasonal prediction of pre-summer CHP frequency in SC.