A comparative study of NPP-VIIRS and DMSP-OLS nighttime light imagery for derivation of urban demographic metrics

Nighttime light imageries from two generations of sensors - Defense Meteorological Satellite Program's Operational Linescan System (DMSP-OLS) and Visible Infrared Imaging Radiometer Suite carried by the Suomi National Polar-Orbiting Partnership (NPP-VIIRS), were utilized to characterize urban areas from the perspective of three aspects (i.e., urban extent, population, and socio-economic activities). The results reveal that remotely sensed nighttime light signal was highly correlated to urban extent, population, and socio-economic activities. The number of urban areas and the total area through thresholding indicates that a 0.06 threshold significantly reduced the blooming effect around large urban areas without attenuating the number of individual small urban clusters for the DMSP-OLS imagery, while this value was smaller than 0.01 for the NPP-VIIRS image. This smaller threshold suggests the superiority of NPP-VIIRS over DMSP-OLS for the delineation of small urban clusters. Furthermore, a comparison between equivalent diameter (ED) and the ratio of area to perimeter suggests that the extracted urban areas (ED < 50KM) from the NPP-VIIRS imagery was more effective to reveal the spatial complexity than those from DMSP-OLS data. The intra and extra structures of urban areas tended to be hidden in the DMSP-OLS imagery, while this information was much clearer in the NPP-VIIRS imagery. However, NPP-VIIRS imagery did not show apparent advantages over DMSP-OLS on the modeling of overall urban population at the county level in the U.S.A. although NPP-VIIRS had stronger intensity to reflect urban population density. The correlations between socio-economic indicators and nighttime variables demonstrate that NPP-VIIRS can depict urban socio-economic activities better than DMSP-OLS. It is also apparent that education and income index were negatively correlated with light intensity per person, but Gini Coefficient was positively correlated with average light intensity (per KM), implying that larger inequality of income or wealth existed in the urbanized areas with stronger night-lights intensity.