Abstract：Under the background of global warming, Meiyu precipitation over the Yangtze River basin has increasingly deviated from its traditional characteristics of persistent light rainfall, exhibiting pronounced dual extremes of heavy rainfall and dry heat. Such complex changes are difficult to accurately characterize using a single variable. To comprehensively describe the evolving features of Meiyu, this study employs the Deviation Degree of Misty Rain (D2MR), a multidimensional index developed based on daily station observations, to characterize Meiyu variability from multiple perspectives, and its sub-indices are further used to identify the dominant Meiyu type in each year. Meanwhile, based on ERA5 reanalysis data and datasets from the Hadley Centre, physically meaningful potential predictors are systematically analyzed, and key preceding factors influencing D2MR and its sub-indices are identified. Prediction models are then constructed using the interannual increment approach. The results indicate that the proposed models exhibit significant and stable predictive performance for the Meiyu extremeness index and its sub-indices, with correlation coefficients exceeding 0.72. The models successfully capture the interannual variability and the significant upward trend of D2MR during 1963 – 2025, and effectively reproduce the characteristics of Meiyu in extreme years, accurately reflecting the increasing extremeness of Meiyu in recent decades. In particular, the models independently predict the occurrences of heavy-rainfall-dominated Meiyu in 2020 and 2024, dry-heat-dominated Meiyu in 2022 and 2025, and the alternating heat-rainfall Meiyu characteristics in 2023. As the frequency of extreme Meiyu events continues to increase, exerting greater socio-economic impacts across the Yangtze River Basin, the characterization and prediction of Meiyu extremity provide essential scientific support for disaster prevention, agricultural planning, and enhance our understanding of regional climate variability.