A 10-m resolution impervious surface map for the Lancang-Mekong Basin from Sentinel-1 and Sentinel-2 remote sensing images

Here, the 2016-2021 impervious surface maps exclusively for the Lancang-Mekong basin is prepared. We demonstrate a machine learning framework that fuses Sentinel-1 Synthetic Aperture Radar images and Sentinel-2 Multispectral images to extract impervious surface pixels on Google Earth Engine. A train sample migration strategy is also proposed to automate impervious surface mapping at different times, eliminating the collection of additional train samples from this vast region. Finally, a quantitative assessment is performed with validation samples interpreted from Google Earth. Result shows that the overall accuracy and Kappa of the final impervious surface maps range from 91.45% to 92.44% and 0.829 to 0.848, respectively. This atlas caters for the measurement of impervious surface coverage and configuration, and thus sound urban planning. <br> The impervious surface map includes three sets of files. (1) Single-band impervious surface map in GEOTIFF format with 10-m spatial resolution, divided into 69 tiles. Mapped value of 0 signifies non- impervious surface and other values (integers from 2016 to 2021) signify the time when the pixel converted to impervious surface. And the maps are projected with WGS 1984 geographic coordinate system (EPSG: 4326). (2) Distribution map of the 69 tiles in ArcGIS Layer format for users to access and download the corresponding map tiles. (3) Thumbnails of the annual impervious surface map in JPEG format. This dataset can be processed with GIS software and programming packages such as ArcGIS, QGIS, ENVI and GDAL.