Extraction of Urban Built-Up Areas Using Nighttime Light (NTL) and Multi-Source Data: A Case Study in Dalian City, China

Abstract

The rapid urban development associated with China’s reform and opening up has been the source of many urban problems. To understand these issues, it is necessary to have a deep understanding of the distribution of urban spatial structure. Taking the six districts of Dalian as an example, in this study, we integrated the enhanced vegetation index, points of interest, and surface temperature data into night light data. Furthermore, herein, we analyze the kernel density of the points of interest and construct three indices using image geometric mean: a human settlement index (HSI), a HSI-POI (HP) index, and a HSI-POI-LST (HPL) index. Using a support vector machine to identify the land type in Dalian’s built-up area, 1000 sampling points were created for verification. Then, the threshold boundary corresponding to the highest overall accuracy of each index and kappa coefficient was selected. The relevant conclusions are as follows: As compared with the other three types of data, the HPL index constructed in this study exhibited natural and social attributes, and the built-up area extracted using this method had the highest accuracy, a high image spatial resolution, and was able to overcome the omission issues observed when using one or two data sources. In addition, this method produces richer spatial details of the actual built-up area and provides more choices for assessing small-scale urban built-up areas in future research.