Logic combination and diagnostic rule-based method for consistency assessment and its application to cross-sensor calibrated nighttime light image products

Abstract

With observations from the Defense Meteorological Satellite Programme's Operational Line Scanning System (DMSP/OLS, 1992–2013) and the Suomi National Polar-Orbiting Partnership's Visible Infrared Imaging Radiometer Suite (NPP/VIIRS, 2012-), night-time light (NTL) imagery has become one of the most unique and widely-used data for understanding human activities due to its unique low-light detection capability and close correlation with socioeconomic development. Its capability for long-term observation has been further enhanced by the recent advancement in cross-sensor calibrated NTL products, which address the inconsistency between DMSP-OLS and VIIRS data and combine them together as extended NTL time series (ENTL). Despite the prosperity of cross-sensor calibration models, comprehensive and in-depth assessments of temporal consistency of their resulting ENTL products remain scarce or constrained at an aggregated scale. This study developed a new assessment scheme based on logical combinations and diagnosis rules for NTL intensity trends. Compared to previous schemes, the proposed scheme offers significant advantages in fine-grained, non-subjective intervention and semi-automation for NTL intensity consistency assessment, and its derived consistency profile layer of ENTL products can more effectively inform end-users in ENTL products selection of products and account for uncertainty in their analysis. Based on the assessment, we generated a standard light intensity dynamic trend layer (SNID) to illustrate the characteristics of global NTL intensity variations over the period from 1992 to 2020 and the applied this layer to validate the effectiveness and applicability of six most representative ENTL products. Our results showed that the scheme can automatically generate NTL intensity consistency features at a finer spatial scale than the previous TSOL-based method, and revealed for the first time a fact that has been neglected before, that is, there was a distinct gap in NTL intensity consistency among different ENTL products, with the percentage of well-matched units fluctuating from 52.81 % to 84.46 %. These variations were particularly evident in regions with high light intensity, rural areas, and high-latitude regions, reflecting the influence of spatial heterogeneity and calibration strategies. In summary, this study refines the detection process for the consistency profile of ENTL products, significantly enhancing their reliability in socioeconomic analysis and urban expansion research. By revealing the intensity consistency differences among various products, it provides critical guidance for users in data selection and application, helping to better address uncertainty.