Mapping 10 m global impervious surface area (GISA-10m) using multi-source geospatial data

Abstract. Artificial impervious surface area (ISA) documents the human footprint. Accurate, timely, and detailed ISA datasets are therefore essential for global climate change studies urban planning. However, due to lack of sufficient training samples operational mapping methods, at a 10 m resolution still lacking. To this end, we proposed method leveraging multi-source geospatial data. Based on existing satellite-derived maps crowdsourced OpenStreetMap (OSM) data, 58 million were extracted via series temporal, spatial, spectral, geometric rules. We then produced dataset (GISA-10m) from over 2.7 Sentinel optical radar images Google Earth Engine platform. test that independent set, GISA-10m achieves an overall accuracy greater than 86 %. In addition, was comprehensively compared with datasets, superiority confirmed. The road further investigated, courtesy dataset. It found China US have largest areas road. rural be 2.2 times while 1.5 larger regions. accounts 14.2 % ISA, 57.9 which is located in top countries. Generally speaking, sampling able achieve rapid efficient mapping, potential detecting other land covers. also shown can improved by incorporating OSM could used as fundamental parameter system science, will provide valuable support planning water cycle study. freely downloaded https://doi.org/10.5281/zenodo.5791855 (Huang et al., 2021a).