全球智库信息集成服务系统

Land cover (LC) has a significant impact on rainfall-runoff dynamics of a watershed, so LC maps are often incorporated into hydrological models to simulate how changes in climate or land management will affect water quantity/water quality within the watershed. The accuracy of a LC map can thus affect the accuracy of hydrological modelling results. However, because LC maps are not typically produced specifically for hydrological studies, the conventional LC map accuracy metrics may not be the most relevant. In this study, we proposed a new metric for LC map accuracy assessment, calculated as the root-mean-square-deviation (RMSD) of the mapped (i.e. estimated) and “ground truth” runoff curve numbers (CN) at randomly-sampled locations. The new metric, CN-RMSD, assesses the accuracy of the direct runoff estimates derived from the LC map, and its benefit over the traditional LC accuracy assessment metrics is that it more heavily weights LC classification errors that cause greater errors in estimated runoff. Ground truth CN data can be collected much in the same way as ground truth data is collected for the traditional accuracy metrics, although a soil map can improve the accuracy of the ground truth CN values. Some potential applications of CN-RMSD, e.g. for LC map selection and LC map fusion, are also discussed.