Inland Water Body Mapping Using Multitemporal Sentinel-1 SAR Data

• Published in: IEEE journal of selected topics in applied earth observations and remote sensing 2021, Vol.14, p.11789-11799
• Main Authors: Marzi, David, Gamba, Paolo
• Format: Article
• Language: English
• Subjects: <italic xmlns:ali="http://www.niso.org/schemas/ali/1.0/" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">k -means; Climate change; Climate prediction; Climate studies; Clustering; Data mining; Earth; Feature extraction; Geomorphology; Inland waters; Land cover; Land surface; Land use; Mapping; Optical sensors; Resolution; Sentinel-1; Spatial discrimination; Spatial resolution; Surface morphology; Synthetic aperture radar; synthetic aperture radar (SAR); time series analysis; Water bodies; water mapping
• ISSN: 1939-1404; 2151-1535
• DOI: 10.1109/JSTARS.2021.3127748

Climate change studies require increasingly detailed information on land cover and land use, to precisely model and predict climate based on their status and changes. A fundamental land cover type that needs to be constantly monitored by the climate change community is water, but currently there is a lack of high-resolution water body maps at the global scale. In this article, we present a fully automated procedure for the extraction of fine spatial resolution (10 m) inland water land cover maps for any region of the Earth by means of a relatively simple k -means clustering model applied to multitemporal features extracted from Sentinel-1 SAR sequences. Indeed, due to heavy cloud coverage conditions in many locations, multispectral sensors are not suitable for global water body mapping. For this reason, in this work, we deal only with SAR data, and specifically with multitemporal Sentinel-1 data sequences. The experimental results, obtained for three geographical areas selected because of their wide diversity in terms of geomorphology and climate, show an almost complete consistency with existing datasets, and improve them thanks to their finer spatial details.