The Jing River is a secondary tributary of the Yellow River, which flows through the middle of the Loess Plateau in China. Severe water scarcity and soil erosion in the basin have threatened sustainable social and economic development. To assess and solve the region’s water resource problems, it is important to understand its historical hydrological climate change. Accordingly, we used five machine learning models and simple linear regression to reconstruct the January-June streamflow of the Jing River based on the tree ring width of Pinus tabulaeformis and Pinus armandii. By combining six models into an ensemble streamflow reconstruction, we obtained a more accurate reconstruction and streamflow variability information than with a single model. Over the past nearly four centuries, the Jing River has experienced seven high streamflow periods and ten low streamflow periods. The main atmospheric forcing factors driving the streamflow variability are the Pacific Decadal Oscillation and the El Niño-Southern Oscillation, which regulate the climate and hydrology of the region by affecting water vapor fluxes and the Asian monsoon. The different climate scenarios revealed the continued reduction in the future Jing River streamflow and a worsening water resource situation. This new streamflow reconstruction can serve as a valuable reference for analyzing regional hydrology and informing water resource management and policy formulations.

Tree rings,Streamflow reconstruction,Machine learning,Jing River