A New Spectral Index for Mapping Water Surfaces in Urban Contexts

This article proposes a modification of the modified normalized difference water index (MNDWI) called rescaled water index (RWI), aiming to enhance the delineation and performance of water body mapping in urban contexts. To evaluate the proposed index, we compare its performance in mapping water surfaces (WSs) in different cities in South America with the results obtained by other well-known water indices, namely, normalized difference water index (NDWI), MNDWI, automated water extraction index (AWEIsh, and AWEInsh). We utilized images from the Sentinel satellites (2A and 2B), all acquired in June 2021, focusing on cities located south of the Tropic of Capricorn to obtain images with the highest incidence of shading, whose spectral response complicates the mapping of WSs. In this scenario, the present study revealed that selecting the best single index for mapping all WSs is an arduous task, as their performances varied across the analyzed locations. Overall, the performance of the indices, evaluated by the partial receiver operating characteristic curve and non-water misclassification point restrictions, revealed quite close results, especially considering the RWI, NDWI, and MNDWI. In the six cities analyzed, the RWI outperformed in three. Additionally, when considering the results for all the cities, the RWI outperformed all the other indices. Our study also rainforced the importance of the complementarity of the indices to improve the performance of the mappings. Such results, therefore, were interpreted as an essential contribution to water body mapping, considering its practical applications in environmental monitoring and water resource management in urbanized areas.