Monitoring of Urban Impervious Surfaces Using Time Series of High-Resolution Remote Sensing Images in Rapidly Urbanized Areas: A Case Study of Shenzhen

Abstract

Knowledge of impervious surface changes is important for understanding the urban environment and human activity. Most of previous studies have investigated impervious surface change at a macro level (e.g., urban expansion) using medium-resolution images but ignored the subtle changes within urban areas. High-resolution images have great potential to precisely monitor the detailed characteristics of impervious surfaces. However, very few studies focused on this issue using multitemporal high-resolution data. In this study, we aimed to resolve these problems and investigate the impervious surface characteristics using high-resolution time-series data. The experiments were performed on Shenzhen, a megacity in China that has experienced rapid urbanization over the past three decades. The images were acquired by QuickBird (2.4 m), WorldView-2 (2 m), and WorldView-3 (1.2 m) at ∼2-year intervals from 2003 to 2017. The presented method integrating multiple features was found to be effective in extracting impervious surfaces from the high-resolution images (kappa coefficient greater than 0.90), and the average accuracy of the change detection was 75%. Courtesy of the high-resolution imagery, it was revealed that the impervious surfaces can be converted back to pervious surfaces, and some regions have shown repeated changes due to the urban renewal planning. It was also found that impervious surfaces in Shenzhen gradually increased before 2012, but subsequently showed a decreasing tendency, reflecting the adjusted strategies for urban development. Our results demonstrate that high-resolution images are essential for precise impervious surface monitoring, and can provide deep insights into urban development patterns during the process of urbanization.