A Genetic Algorithm-Based Urban Cluster Automatic Threshold Method by Combining VIIRS DNB, NDVI, and NDBI to Monitor Urbanization

Accurate and timely information related to quantitative descriptions and spatial distributions of urban areas is crucial to understand urbanization dynamics and is also helpful to address environmental issues associated with rapid urban land-cover changes. Thresholding is acknowledged as the most popular and practical way to extract urban information from nighttime lights. However, the difficulty of determining optimal threshold remains challenging to applications of this method. In order to address the problem of selecting thresholds, a Genetic Algorithm-based urban cluster automatic threshold (GA-UCAT) method by combining Visible-Infrared Imager-Radiometer Suite Day/Night band (VIIRS DNB), Normalized Difference Vegetation Index (NDVI), and Normalized Difference Built-up Index (NDBI) is proposed to distinguish urban areas from dark rural background in NTL images. The key point of this proposed method is to design an appropriate fitness function of GA by means of integrating between-class variance and inter-class variance with all these three data sources to determine optimal thresholds. In accuracy assessments by comparing with ground truth—Landsat 8 OLI images, this new method has been validated and results with OA (Overall Accuracy) ranging from 0.854 to 0.913 and Kappa ranging from 0.699 to 0.722 show that the GA-UCAT approach is capable of describing spatial distributions and giving detailed information of urban extents. Additionally, there is discussion on different classifications of rural residential spots in Landsat remote sensing images and nighttime light (NTL) and evaluations of spatial-temporal development patterns of five selected Chinese urban clusters from 2012 to 2017 on utilizing this proposed method. The new method shows great potential to map global urban information in a simple and accurate way and to help address urban environmental issues