LAUSR

Abstract Artificial nighttime lights (NTL) generated by human activities offer a unique opportunity to understand urban environments. Although previous studies have widely used NTL images to map urban extent at multiple scales, it remains a challenging task to address how NTL respond exactly to urbanization and thus to map urbanization from NTL. In this study, using monthly Suomi National Polar-orbiting Partnership/Visible Infrared Imaging Radiometer Suite (NPP/VIIRS) NTL images between 2013 and 2017, we developed a method to decompose time-series NTL signal into annual and seasonal components. Further, we proposed an NTL-based indicator for the detection of impervious surfaces change (ISC) by integrating annual increment and seasonal variation of NTL brightness. The indicator was then used to identify ISC by using a thresholding method. The application of the methodology in the conterminous United States (CONUS) revealed a more rapid urbanization in the southern CONUS than the northern states and a northeastern-southwestern gradient of NTL seasonality. It was also found that NTL of November and December provided the most accurate characterization of urban extent for most areas in the CONUS. The detection of ISC in four representative regions (i.e. Dallas-Fort Worth-Arlington, greater Washington D.C., Denver-Aurora, and Atlanta) resulted in a moderate to high accuracy with the overall accuracy of ~80% and the Kappa value ranging from 0.56 to 0.73. Despite of this, the results showed a low accuracy of NTL-derived changing year of ISC (Kappa: 0.28) because of the existence of temporal inconsistency between NTL increase and ISC. The proposed method has the potential to timely map urban expansion at large geographical scales (e.g., continental and global) in a cost-efficient manner.