ENSO-induced precipitation seasonality in the Chinese loess plateau during 1767-2050 CE.

Climate change has not only altered the total amount of rainfall but also significantly impacted its seasonal distribution. This alteration is particularly critical in monsoonal regions, where the availability and timing of water resources are essential for agricultural activities and ecological sustainability. However, the understanding of precipitation seasonality was largely constrained by the limitations of observational data. In this study, we present a reconstruction of seasonal precipitation patterns derived from the magnetic properties of a stalagmite collected from the Chinese Loess Plateau, which serves as a proxy for precipitation seasonality from 1767 to CE 2006. Our results indicate that magnetic-based precipitation seasonality exhibits significant fluctuations characterized by quasiperiodic oscillations with periodicities of 2-7 and 14 years. We find an upward trend in precipitation seasonality since the 1980s, suggesting that North China may be entering a new phase of increased rainfall seasonality, and by applying a convolutional-neural-network forecasting model, this upward trend could be expected to extend till CE 2050. We then observe a strong correlation between changes in this seasonality proxy and El Niño-Southern Oscillation (ENSO) conditions, inferring that El Niño events amplify precipitation seasonality in North China on decadal timescales over the past three centuries and sustaining at least to CE 2050. Therefore, our reconstruction demonstrates the importance of considering ENSO forcing when assessing precipitation seasonality and, in turn, provides alternative ways to document ENSO variability in the past.