Summer Extreme Rainfall Over the Middle and Lower Reaches of Yangtze River: Role of Synoptic Patterns in Historical Changes and Future Projections

It is one of the major challenges in climate science to project future changes in extreme rainfall. To overcome this challenge, in this study, four typical synoptic patterns (SPs) triggering summer extreme rainfall over the middle and lower reaches of Yangtze River (MLYR) are identified through hierarchical clustering. These typical SPs share common characteristics of intensified Mei-yu trough and Western Pacific Subtropical High but differ in terms of mid-latitudes disturbances, such as an intensified ridge (Cluster 1, Cluster 2) or trough (Cluster 3, Cluster 4) near Lake Baikal (Cluster 1, Cluster 3) or Northeast China (Cluster 2, Cluster 4). The linkage between extreme rainfall and typical SPs is verified at various time scales. The typical SPs associated with extreme rainfall are substantially different from the circulation patterns found on ordinary days, and their frequency is significantly correlated with that of extreme rainfall across the interannual scales. Furthermore, the distinct changes in different typical SPs serve as a "bridge" for understanding the long-term impact of circulation changes on local extreme rainfall, even though the two do not appear to be connected at first sight. Specifically, the circulation changes imply more (less) frequent SP-Cluster 1 (SP-Cluster 3), which tends to produce more extreme rainfall to the south (north) of the Yangtze River within MLYR. To project future changes in extreme rainfall, we utilize a weighting method for the multi-model ensemble based on each model's capability to capture the observed typical SPs. This method effectively narrows the inter-model spread. Extreme rainfall has a huge impact on human society and ecosystem, but the large uncertainty of extreme rainfall simulation hinders the use of climate models in future risk assessment. To reduce the uncertainty, we identify critical physical mechanisms for extreme rainfall formation in the real world and then utilize them as metrics to assess climate models and give more weight to models with better performances. We first identified four types of weather configurations as the critical physical mechanisms for extreme rainfall formation, which are further verified to be closely connected with extreme rainfall at different time scales (i.e., daily, interannual, and several decades). Then, we assess how well each model captures these weather configurations and then generate weighted multi-model ensembles based on the model performances. This method effectively reduces the uncertainty in future projections. This assessment and correction procedure has great potential to provide reliable projections of extreme events in other regions and seasons. The typical synoptic patterns (SPs) triggering summer extreme rainfall (Rx) are differed by mid-latitudes disturbancesThe connection between long-term changes in circulation and Rx emerges when considering various trends of different typical SPsBased on the model's capability to capture the typical SPs, a weighting method could narrow the inter-model spread of future projections