Summer precipitation contributes nearly half of the annual rainfall in eastern China, playing a crucial role in socio-economic development. Its accurate near-term prediction has long been sought by policymakers, stakeholders, and the climate science community to enhance climate risk management. This study evaluates the decadal predictive capabilities of 9 initialized models from the Decadal Climate Prediction Project (DCPP) of the Coupled Model Intercomparison Project Phase 6 (CMIP6). Results indicate significant but modest skills for precipitation over Southern China (SC) during 1963–2019 and over North China (NC) starting in the late 1990s, while predictive skill over Jianghuai (JH) remains low. Sources of predictability are investigated to enhance the imperfect forecasts. The North Atlantic Subtropical Sea Surface Temperature (SST) emerges as a key predictability source for SC precipitation, facilitating the “Silk Road” teleconnection pattern. Similarly, the Subpolar Gyre SST is closely associated with NC precipitation through the stimulation of the circumglobal teleconnection (CGT) pattern. These SST sources can significantly improve the decadal skill of precipitation over eastern China after applying machine learning methods to build the forescast models. Specifically, the anomalous correlation coefficients (ACC) for SC, NC and JH precipitation are 0.8, 0.79 and 0.65, and mean Squared Skill Scores (MSSS) skills can be enhanced from 0.19, 0.33 and -0.17 in the multi-model ensemble (MME) mean to 0.54, 0.59 and 0.36, respectively. Overall, our study offers valuable references for understanding physical mechanisms influencing summer precipitation over eastern China and advancing decadal prediction skills.
 

暂无