干旱区生态环境知识资源中心

Mapping impervious surface area (ISA) in arid/semiarid regions is of great importance because of the vulnerable ecological situation there. However, using remote sensing data for ISA mapping in arid/semiarid landscapes is still very difficult because of the complexities of ISA, such as bare soils, impervious surface areas, and sparse vegetation cover. The Visible Infrared Imaging Radiometer Suite with Day/Night Band (VIIRS-DNB) carried by the Suomi National Polar-orbiting Partnership, focusing on earth nighttime light, provides novel ways to map urban information, but is still little used in ISA mapping. In this paper, we explore a new index based on VIIRS-DNB data called the improved impervious surface index (IISI). We use a linear regression model for fractional ISA estimation in an arid/semiarid urban region, and the IISI was used as an independent variable. Reference ISAs from selected sites were mapped from Landsat images and were used as a dependent variable in this model. The results indicate that the IISI-based model provides good estimation with coefficient of determination (R-2)of 0.70, much greater accuracy than that using normalized VIIRS-DNB data. This method is very valuable for fractional ISA mapping in an arid/semiarid region.