Extracting Urban Water Bodies from Landsat Imagery Based on mNDWI and HSV Transformation

Urban water bodies are critical for sustainable urban ecological and social development. However, the complex compositions of urban land cover and small water bodies pose considerable challenges to urban water surface delineation. Here, we propose a novel urban water extraction algorithm (UWEA) that is efficient in distinguishing water and other low-reflective objects by combining the modified normalized difference water index (mNDWI) and HSV transformation. The spectral properties of urban land covers were analyzed and the separability of objects in different color spaces was compared before applying the HSV transformation. The accuracy and robustness of the UWEA were validated in six highly urbanized subregions of Beijing, Tokyo, and New York, and compared with the mNDWI and HIS methods. The results show that the UWEA had the fewest total errors (sum of omission and commission errors) for all the validation sites, which was approximately 3% fewer errors than those of the mNDWI and 17% fewer errors than those of the HIS method. The UWEA performed best because it was good at identifying small water bodies and suppressing reflective surfaces. The UWEA is effective in urban water monitoring and its thresholds are also robust in various situations. The resulting highly accurate water map could support water-related analyses. This method is also useful for scientists, managers, and planners in water resource management, urban hydrological applications, and sustainable urban development.