1. Introduction
Human activities are continuously changing the Earth’s natural surface and the urban systems, comprising a continuum of socioeconomic and demographic phenomena. Monitoring global human footprint patterns is crucial for understanding global environmental change, sustainability, and socioeconomic status (Leu et al., 2008; Venter et al., 2016; Zhu et al., 2020). Shifts in societies, cultures, economic system structures, policies, technologies, and behaviors are rapidly affecting global ecosystems (Malecki, 1997; Ojima et al., 1994; Steffen et al., 2006). Human footprint expansion and reconstruction actions are driven by social-ecological changes, such as population growth and the consequential needs for natural resources. These drivers have modified the long-term land cover and land use features over large areas of land surfaces (Deng et al., 2009; Turner, 2010). Meanwhile, disturbance stresses caused by social shocks and behavioral changes (e.g., armed conflict and gathering events) engendered short-term changes on the local scale (Baumann and Kuemmerle, 2016). Data-intensive frameworks for monitoring human-induced land changes at large-scale have become essential to enable a more timely, comprehensive, and deeper understanding of human activity dynamics. This demand calls for the need for reliable, timely, and large-area consistent information. In contrast to available socioeconomic and field survey data, the remote sensing nighttime light (NTL) imagery provides a reliable measure of human activity changes at the global scale with fine temporal and spatial resolutions (Jensen and Cowen, 1999; Xie and Weng, 2016).
The remotely sensed NTL data provides the direct imprint of both the spatial extent and emission intensity of the artificial light, which is a good indicator of the human footprint changes (Elvidge et al., 1997; Levin et al., 2020; Zhang and Seto, 2011). Characteristics of artificial nocturnal illumination are often associated with the economic and demographic structures of modern society (Green et al., 2015). The artificial NTL intensity is strongly responded to the growth or decrease of the health and development of the society (Hölker et al., 2010). Strong correlations have been found among the NTL trends and socioeconomic status (Ma et al., 2012), which enables us to estimate the spatial-temporal dynamics of society based on the NTL changes. Accurate results have been produced by using the NTL datasets as the major inputs for mapping the urbanization processes (Shi et al., 2014), estimating Gross Domestic Product (GDP) and mapping poverty (Yu et al., 2015), monitoring natural hazards and recurrent disaster impacts on underserved communities (Machlis et al., 2022; Román et al., 2019), armed conflicts (Li et al., 2018), cultural behaviors (Liu et al., 2019; Román and Stokes, 2015), and detecting long-term landscape changes in the urbanized regions (Chen et al., 2019).
Compared with the previous Defense Meteorological Satellite Program’s Operational Line Scanner (DMSP/OLS) sensor, the Visible Infrared Imaging Radiometer Suite (VIIRS) Day/Night Band (DNB) provides higher spatial resolution NTL data with significant improvement in its quality, traceability, and consistency (Elvidge et al., 2017). However, high uncertainties caused by both the dynamics of the emission sources and the environmental impacts still exist in the VIIRS DNB observations (Coesfeld et al., 2018; Elvidge et al., 2022; Li et al., 2020; Wang et al., 2021) which makes time series analysis (e.g., daily) of DNB observations extremely challenging. Moreover, the VIIRS DNB observations are inevitably subject to the extraneous impacts of the angular effects, surface BRDF and albedo, lunar phases, atmospheric effects, cloud and snow contamination, and vegetation canopy (Wang et al., 2021). To alleviate the significant variation caused by the external effects, previous studies used the monthly or annual composited NTL data to smooth the variation (Elvidge et al., 2021; Levin, 2017; Liu et al., 2019; Yang et al., 2020), which is a viable solution but also significantly reduces the data temporal frequencies, making it difficult to provide timely information and monitor short-term changes (Xie et al., 2019; Zhao et al., 2020; Zheng et al., 2021).
NASA’s Lunar-BRDF-corrected Black Marble NTL product (VNP46A2) provides daily 15-arc-second spatial resolution Visible Infrared Imaging Radiometer Suite (VIIRS) Day/Night Band (DNB) data with operational correction for the lunar phase effects (Román et al., 2018). Significant reduction of the temporal variation has been achieved with the correction of the major sources of noise from the lunar cycle (Elvidge et al., 2022; Wang et al., 2021) which provides new opportunities for analyzing NTL dynamics based on the daily DNB data for the first time. However, there are still some remaining factors that could cause large variations in NASA’s Black Marble products, such as cloud and snow missed in the Quality Assurance (QA) flag (Wang et al., 2021), vegetation phenology, surface albedo (Levin, 2017; Tang et al., 2021), and angular effect from the illuminating artificial lights (Li et al., 2022; Tan et al., 2022), which makes their direct usage for time series analysis difficult.
In this study, we aimed to develop a new algorithm for continuous monitoring of NTL changes based on daily VIIRS DNB observations from NASA’s Black Marble standard product suite, which adds robustness to the large variation caused by cloud and cloud shadow missed in the Level 3 QA flagging process, vegetation phenology, snow, and angular effects introduced by illuminating artificial lights.