An Improved Cross-Sensor Calibration Approach for DMSP-OLS and NPP-VIIRS Nighttime Light Data

Defense Meteorological Satellite Program Operational Linescan System (DMSP-OLS) and Suomi National Polar-orbiting Partnership Visible Infrared Imaging Radiometer Suite (NPP-VIIRS) nighttime light (NTL) data have been widely used to monitor human activities and urbanization. However, the DMSP-OLS sensor has no on-board calibration, the DMSP-OLS and NPP-VIIRS data are not spatially consistent and continuous due to the differences in spatial resolution and sensor design between satellites, which makes it difficult to use both datasets at the same time for spatio-temporal consistency analysis. Based on this, this study proposed a new approach for systematically calibrating the DMSP-OLS and NPP-VIIRS NTL data, and rapidly generated a continuously consistent NTL dataset from 1992 to 2022. First, the DMSP-OLS data were calibrated using the invariant target method. Secondly, the NPP-VIIRS data were subjected to outlier elimination and time-series comparability calibration. Third, a new fourth-degree polynomial fit calibration model is proposed to calibrate the NPP-VIIRS data into the DMSP-OLS data scale. Finally, this research obtained a long-time series (1992–2022) “DMSP-OLS-like” NTL dataset, and analyzed the dynamics NTLs at different scales. Compared with other fitting methods, the “DMSP-OLS-like” dataset in this research has a clearer city hierarchy. It significantly reduces the saturation phenomenon and spillover effect, has a good spatial pattern and spatio-temporal consistency, and is highly compatible with the relevant socioeconomic reference quantities. The correlation coefficients <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i><sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of total DN with population, GDP, and electricity consumption were 0.922, 0.922, and 0.988, respectively. The results showed that the proposed approach was effective, which is superior to those of existing researches. Our results effectually solved the problem that the two NTL datasets could not be used at the same time, and improved the calibration accuracy in the two NTL datasets, which provides a new source of data for relevant studies such as urban and environmental issues in long-time series.