Reconstructing China's Natural Streamflow at 1‐km Resolution

Climate change and intensifying human activities are placing unprecedented pressure on China's water resources, necessitating high‐resolution, naturalized streamflow records for effective management. We develop a 1‐km gridded fully coupled land‐surface‐hydrologic‐hydrodynamic modeling system for China based on the CLHMSv2.0 model. To refine grid‐level runoff simulations, we propose a machine learning based gridscale calibration approach using observed streamflow and climatic and physiographic attributes from thousands of Chinese catchments. To represent China's complex river‐lake networks at fine scales, we introduce an optimized hybrid 1D/2D diffusion wave scheme to resolve backwater effects and bidirectional exchanges between rivers and lakes. The modeling system is extensively validated at daily and monthly scales against 1,225 flow gauges and 5 lake water level gauges across the nation. Streamflow evaluations show that the median daily (monthly) Nash‐Sutcliffe efficiency is 0.57 (0.77) and the daily (monthly) Kling‐Gupta Efficiency is 0.65 (0.73); water level simulations for major freshwater lakes are also satisfactory, with correlation coefficients mostly above 0.80. ... mehrApplying the system over 1962–2024, we generate the national 1‐km gridded daily natural streamflow estimates named CHASE v1.0, which is conditionally available at https://hydrodata.cn/chase.