A novel cross-sensor calibration method to generate a consistent night-time lights time series dataset. Issue 14 (17th July 2020)

• Record Type: Journal Article
• Authors: Tu, Ying; Zhou, Hanlin; Lang, Wei; Chen, Tingting; Li, Xun; Xu, Bing
• Is Part Of: International journal of remote sensing. Volume 41:Issue 14(2020)
• Page Start: 5482
• Page End: 5502
• Publication Date: 2020-07-17
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2020.1731935
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: Night-time lights (NTLs) collected from the Defense Meteorological Satellite Program's Operational Linescan System (DMSP-OLS) and the Visible Infrared Imaging Radiometer Suite (VIIRS) of the Suomi National Polar Partnership satellite have been widely used in multiple disciplines. However, the defects of DMSP and VIIRS data itself, and the inconsistency between them, hinder their applications in long-term finer studies. Despite some effective efforts, existing relevant researches are still limited by the shortcomings of data inaccessibility, data deficiency neglection, and spatial resolution degradation. To resolve these issues, a novel cross-sensor calibration method was developed in this article by considering three Chinese metropolises (Beijing, Shanghai, and Guangzhou) as the study area. First, the original DMSP NTL images for 2000–2013 were calibrated through stepwise calibration, background noise removal and vegetation adjustment. Second, stable VIIRS annual composites for 2012–2019 were produced after seasonal noise removal, yearly aggregation, background noise removal, vegetation adjustment, and outliers correction. Third, a power regression model was applied to align pixel values of the processed DMSP and the processed VIIRS data for the overlapped years, and consistent NTLs for 2000–2019 were further generated using the regression results. The evaluations based on statistical coefficients, spatial patterns, profile curves, dynamic changes, and correlations with socioeconomic statistics, indicated the robustness and effectiveness of the proposed approach in filling the gaps between DMSP and VIIRS data. The consistent, continuous, and stable NTL time series could serve as input data for further applications, such as urban dynamics capture, economic growth estimation, and population distribution mapping.