Mengqing Geng, Xuecao Li, Shirao Liu, Guojiang Yu, Yuyu Zhou, Peng Gong
Abstract
The availability of long-term, annually harmonized nighttime light (NTL) data is pivotal for monitoring human activities across past decades. The Defense Meteorological Satellite Program (DMSP) has provided over 20 years of NTL observations, facilitating extensive global and regional studies. With the termination of DMSP NTL in 2013 and the subsequent introduction of the Visible Infrared Imaging Radiometer Suite (VIIRS) NTL data, we previously developed the global harmonized NTL dataset (H-NTL-v1), which offers a temporally extended and consistent time series data from 1992 to 2022. Despite its widespread use, the H-NTL-v1 dataset has been affected by noise from auroras and transient lights, particularly in high-latitude areas. In this study, we present an innovative method that employs temporal frequency analysis and Pareto surface optimization to address these challenges, resulting in an improved global harmonized NTL dataset (H-NTL-v2). This enhanced dataset markedly improves the consistency between historical DMSP (1992–2013) and the DMSP-like estimates post-2014. For aurora affected city lights, here we did not implement a specific correction algorithm. Our results indicate that the improved H-NTL-v2, spanning 1992 to 2022, significantly reduces auroral noise and inter-annual variability. Compared to the original H-NTL-v1, the improved H-NTL-v2 demonstrates strong agreement with DMSP observation in 2012 and 2013. It exhibits significantly diminished fluctuation in total lit pixels and digital numbers, particularly for low-luminance pixels. This refined dataset minimizes noise impacts from auroras and other sources, enhancing the application potential of NTL data in studies of light pollution, urban slums, and poverty and inequality in developing regions.
An efficient method for aurora and noise reduction with a harmonized nighttime light dataset.pdf
