Urban Impervious Surface Extraction Using Seasonal Time Series SAR Images

Abstract

Urban impervious surface (UIS) is an essential indicator to measure urban ecology, and understanding its temporal and spatial distribution is of great significance for alleviating urban diseases, such as heat island. Although remote sensing techniques have been widely used to extract UIS, the capability of synthetic aperture radar (SAR) technology in UIS extraction has not been thoroughly investigated. Accordingly, this article systematically analyzed the role of the backscattering and interferometric coherence features of dual polarized (VH/VV) SAR in UIS extraction, taking Tianjin city, China, as a study case. Specifically, we used seasonal time series Sentinel-1B backscattering and coherence images to extract UIS via two object-based classifiers. We further investigated the seasonal, polarization, and precipitation effects on UIS extraction using SAR images. Our findings suggested that, for coherence images, the extraction accuracy of UIS in summer is the highest, while for backscattering that in spring and winter are the highest. Our results also suggested that VH polarization is more suitable for UIS extraction compared with VV polarization. These findings of this study confirmed the utility and effectiveness of time series SAR backscattering and coherence images in UIS extraction, providing an essential reference for UIS extraction in other places with a similar climate. We encourage more efforts to be made toward UIS extraction using SAR images and toward designing a filling scheme for the UIS extraction in a continuous manner, especially in cloudy and rainy areas.

Keywords

• Coherence images
• synthetic aperture radar (SAR) backscattering
• seasonal effect
• time series SAR images
• urban impervious surface (UIS)