An ensemble of 12 CMIP6 models was used to project future changes in summer extreme precipitation over eastern China during 2036-2055 under the SSP2-4.5 scenario. Extreme precipitation was quantified using the total precipitation from days exceeding the 95^th percentile of wet-day precipitation (R95pTOT). Large inter-model uncertainty is evident over the Huabei region, substantially reducing the reliability of the multi-model ensemble (MME) projection there. To address this inter-model uncertainty, a pattern-based clustering analysis was applied to the MME projections, yielding three distinct and equally likely patterns (Clusters 1-3) of summer extreme precipitation change. Clusters 1 and 3 project increases in extreme precipitation over Huabei for 24.8 mm and 12.7 mm, whereas Cluster 2 indicates a decrease for -1.2 mm. An atmospheric moisture budget analysis reveals that the inter-cluster differences in extreme precipitation changes are primarily driven by dynamic effect associated with contrasting circulations. In Cluster 1, a strengthened and westward-shifted western North Pacific subtropical high (WNPSH) enhances southerly moisture transport, which is associated with cold SSTA over the central tropical Pacific. Cluster 3 exhibits a circulation pattern similar to that of Cluster 1, but with weaker intensity. In contrast, Cluster 2 is characterized by a weakened and eastward-shifted WNPSH at lower level, together with a southward-displaced East Asian subtropical westerly jet at upper level, resulting in less southerly moisture transport. In addition to differences in summer-mean circulation, atmospheric stability conditions over Huabei were compared across these clusters. Clusters 1 and 3 exhibit higher frequency of cases with large convective potential energy (CAPE), whereas Cluster 2 indicates more frequent occurrence of cases with large convective inhibition (CIN).

extreme precipitation,CMIP6,inter-model uncertainty,clustering analysis,eastern China,multi-model ensemble (MME)