Using a combination of nighttime light and MODIS data to estimate spatiotemporal patterns of CO emissions at multiple scales.

Accurate mapping spatiotemporal patterns of CO emissions and understanding its driving factors are very important, it is useful for the scientific and rational formulation of carbon emission reduction policies. Nevertheless, due to data availability issues, most studies have been limited to the global and national scales, and the models used were relatively simple. In this paper, we used the 500 m Visible Infrared Imaging Radiometer Suite Day/Night Band (VIIRS-DNB) data and the 250 m Moderate Resolution Imaging Spectroradiometer normalized difference vegetation index (MODIS NDVI) and proposed an improved CO emissions index (ICEI) to calculate CO emissions. Compared with the total nighttime light (NTL), the average regression coefficient (R) can be improve from 0.73 to 0.78. We also used the coefficient of variation, spatial autocorrelation, and geographically weighted regression models to analyze the temporal and spatial variation mode of CO emissions, as well as the associated correlation and heterogeneity, at three different administrative unit scales during 2012-2019. Our experimental results demonstrate that: (1) the improved index (ICEI) is better than the traditional variable (NTL) in estimating CO emissions; (2) the highest CO emissions are primarily gathered in the developed coastal areas in eastern China; and (3) at the provincial level, the added value of the secondary industry is the most significant factor, whereas the added value of the tertiary industry is negatively correlated with CO emissions.