The timely and reliable estimation of imperviousness is essential for the scientific understanding of human-Earth interactions. Due to the unique capacity of capturing artificial light luminosity and long-term data records, the Defense Meteorological Satellite Program (DMSP)’s Operational Line-scan System (OLS) nighttime light (NTL) imagery offers an appealing opportunity for continuously characterizing impervious surface area (ISA) at regional and continental scales. Although different levels of success have been achieved, critical challenges still remain in the literature. ISA results generated by DMSP-OLS NTL alone suffer from limitations due to systemic defects of the sensor. Moreover, the majority of developed methodologies seldom consider spatial heterogeneity, which is a key issue in coarse imagery applications. In this study, we proposed a novel method for multi-temporal ISA estimation. This method is based on a linear regression model developed between the sub-pixel ISA fraction and a multi-source index with the integrated use of DMSP-OLS NTL and MODIS NDVI. In contrast with traditional regression analysis, we incorporated spatial information to the regression model for obtaining spatially adaptive coefficients at the per-pixel level. To produce multi-temporal ISA maps using a mono-temporal reference dataset, temporally stable samples were extracted for model training and validation. We tested the proposed method in the Yangtze River Delta and generated annual ISA fraction maps for the decade 2000–2009. According to our assessments, the proposed method exhibited substantial improvements compared with the standard linear regression model and provided a feasible way to monitor large-scale impervious surface dynamics.
This is an open access article distributed under the Creative Commons Attribution License
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.