【摘要】 地表径流的预测和模拟是水文地理研究的核心，对水资源管理和规划都具有重要意义。传统研究借助于实测地表径流进行水文模型率定和参数移植，对“无测站流域”进行径流预测，但在实测径流稀缺或受到强人类活动（如大坝调控）干扰时，参数移植能力受到限制、预测能力不足。而遥感数据具有时空连续的优势，且不受制于地表径流观测时间序列，为径流预测提供了新思路。本文采用了遥感数据率定水文模型的方法，在中国84个流域探究该方法采用遥感蒸散发、水储量和土壤水数据不同组合率定水文模型，评估各方案预测流域径流的潜力。结果显示基于偏差校正后的网格蒸散发数据约束水文模型的新方法，在中国干旱区和湿润区均具有较大的应用潜力；在格网上约束水文模型相比集总式约束具有较为明显的优越性；应采用多个水文模型提升该方法在不同流域的应用潜力。该方法的径流预测能力具有明显的空间分异性，因地制宜方可有效发挥遥感数据的优势。更多还原

【Abstract】 The prediction and estimation of surface runoff are core research topics in hydrology and geography, with important implications for water resource management and planning.Traditional research relies on measured surface runoff for model calibration and parameter transfer to predict runoff in ungauged basins. However, when measured runoff is scarce or disturbed by strong human activities(such as dam regulation), the ability of parameter transfer is limited, resulting in insufficient runoff prediction capacity. Remote sensing data, with its spatial and temporal continuity, and without being restricted by the observation time series of surface runoff, provides a new approach for runoff prediction. This article explores the method of calibrating hydrological models with remote sensing data in 84 basins in China. Different combinations of remote sensing evapotranspiration, water storage, and soil water data were used to evaluate the potential of predicting runoff by calibrating the hydrological models directly. The results showed that the new method of constraining hydrological models based on grid-based evapotranspiration data with bias correction has great potential for application in both arid and humid areas of China. Grid-based model constraint has a more obvious advantage over lumped model constraint, and multiple hydrological models should be used to enhance the application potential of this method in different basins. The runoff prediction capability and application potential of this method have spatial variability and should be tailored to local conditions to effectively leverage the advantages of remote sensing data.更多还原