A novel method to extract urban human settlements by integrating remote sensing and mobile phone locations

• Main Authors: Bin Chen, Yimeng Song, Bo Huang, Bing Xu
• Format: Article
• Language: English
• Published: Elsevier 2020-06-01
• Series: Science of Remote Sensing
• Subjects: Urban human settlements; Location-based data; Landsat; MODIS; POIs
• Online Access: http://www.sciencedirect.com/science/article/pii/S266601722030002X

Satellite-based human settlement extraction methods have limited practical applications, due to merely studying the difference between human settlements and other land cover/use types in physical attributes (e.g., spectral signature and land surface temperature) instead of considering basic anthropogenic attributes (e.g., human distribution and human activities). To deal with this challenge, we proposed a novel method to accurately extract human settlements by integrating mobile phone locating-request (MPL) data and remotely sensed data. In this study, human settlements for selected cities were mapped at a medium resolution (30 ​m) by redistributing the MPL data using Landsat Normalized Difference Vegetation Index (NDVI) adjusted weights, with an overall accuracy of above 90.0%. Additionally, by extending the proposed method to the MPL and Moderate Resolution Imaging Spectroradiometer (MODIS) data, a coarse-resolution (250 ​m) map of human settlements in China was created with an overall accuracy of 95.2%. Compared with the widely used nighttime light based methods, the proposed method could solve the long-existing problems such as data saturation and blooming effects, as well as characterizing human settlements with fine spatial details. Our study provides an alternative approach to human settlement extraction by combining its physical and anthropogenic attributes, and it can be easily adjusted with multi-scale remotely sensed data and applied to human settlement extraction at different scales.