NDBSI: A normalized difference bare soil index for remote sensing to improve bare soil mapping accuracy in urban and rural areas

• We developed NDBSI according to the spectral characteristics of multiple soil types. • The improved performance of NDBSI in soil mapping has been found in six study areas. • NDBSI exhibited a stronger ability to distinguish bare soil and impervious surfaces. As a component of the Earth’s land surface, bare soil is an important indicator of urbanization and therefore plays an irreplaceable role in regional ecosystems. Given that the spectral signatures of bare soil are complicated and easily confused with impervious surfaces, few soil spectral indices have been purposed and widely used. In this study, we developed a soil index, the normalized difference bare soil index (NDBSI), by analyzing the spectral characteristics of multiple soil types. Using Landsat 8 OLI images from six urban and rural areas in China, we found that NDBSI was more effective in mapping bare soil than currently used indices (i.e. bare soil index (BI), product index for dark soil (PIDS), and biophysical composition index (BCI)), especially in distinguishing between bare soil and impervious surface. Moreover, NDBSI showed good performance in identifying bare soil from red brick, which is helpful for soil mapping in industrial urban areas, where red brick is widely used as roofing materials. Our study may be the first attempt to develop a soil index considering the spectral characteristics of various soil types and verify its effectiveness in multiple regions. More validation of NDBSI with remote sensing images with different spatial resolution will facilitate its applications, and ultimately promote more accurate soil mapping in the future.