The extreme precipitation in the summer of 1931 caused severe flooding in East Asia, resulting in significant loss of life and economic damage. Our previous study, using century-scale reanalysis datasets, demonstrated that this event was driven by a combination of El Niño-related tropical Indian Ocean warming and extratropical wave trains over Eurasia. However, the limited availability of early 20th-century observational data introduces uncertainties in understanding the physical mechanisms derived from reanalysis datasets. Here, we employ century-long large ensemble hindcast experiment CSF-20C to examine the underlying mechanism of this event from a large-sample perspective, accounting for associated uncertainties. The results suggest that while tropical oceanic forcing contributed to a displaced 1931 rainband, the observed precipitation pattern emerged only with additional influence of mid- to high-latitude wave trains driven by North Atlantic dipolar sea surface temperature (SST) anomalies. The ensemble mean predictions from CSF-20C well capture the observed tropical SST anomalies and the westward extension of the western Pacific subtropical high, while the predicted rainbelt is displaced southward compared to the observation. Nonetheless, some ensemble members successfully replicate the observed pattern of precipitation anomalies, and ensemble sensitivity analysis identifies North Atlantic SST dipole-driven wave trains as crucial for precise prediction. The North Atlantic SST dipole-driven wave trains are also evident in the observations. Our study advances the understanding of the 1931 East Asian summer flood by showing that tropical oceanic forcing alone cannot explain its position and magnitude, revealing the crucial role of mid- to high-latitude wave trains in shaping the observed precipitation pattern and identifying the North Atlantic as their source.

seasonal forecast,large ensemble,flood,East Asian summer monsson