Post-disaster Power Supply Dynamics Observed by Night-time Light Images

Natural disasters annually claim thousands of lives worldwide and severely disrupt critical infrastructure. Rapid detection of damaged regions and the subsequent tracking of post-disaster recovery are essential for efficient humanitarian relief and the optimal allocation of aid funds, ultimately reducing post-disaster mortality and suffering. While traditional satellite imagery has long been a crucial tool for disaster mapping and emergency management, it historically focuses on observable "physical damage" (such as destroyed buildings or flooded lands) rather than critical "soft damage" like widespread electricity supply halts.

Addressing this critical gap, night-time light (NTL) remote sensing images provide a unique capability to map power outage regions and monitor the post-disaster recovery of electricity networks. These images serve as indispensable tools for decision-making in humanitarian aid, empowering governments and aid agencies to formulate targeted post-disaster policies. The night-time light remote sensing team at Wuhan University has leveraged these observations for numerous natural disaster mapping and recovery evaluation efforts. By producing specialized disaster-related products grounded in a deep physical understanding of NTL imaging mechanisms, the team has successfully supported the United Nations Office for the Coordination of Humanitarian Affairs (OCHA), the International Charter Space and Major Disasters, and various international users in critical relief work for both climate-driven and geophysical disasters.

The 2022-2023 Pakistan Floods, fueled by unprecedented monsoon rains and accelerated glacier melt, affected over 33 million people, displacing millions and devastating agricultural lands. Experts attribute the severe intensity of these extreme events to human-induced climate change, which acts as a profound threat multiplier. Similarly, in September 2023, Storm Daniel brought catastrophic rainfall to northeastern Libya. Exacerbated by a warming Mediterranean Sea, this extreme weather event—calculated by scientists to be up to 50 times more likely due to climate change—led to the devastating collapse of two poorly maintained dams above the city of Derna. Over 30 million cubic meters of water cascaded into the city, claiming thousands of lives, displacing tens of thousands, and sweeping away a quarter of the urban infrastructure. Both unprecedented disasters vividly highlight how climate change is accelerating the frequency and severity of extreme weather events, plunging vulnerable communities into compounding crises.

These two disasters destroyed a vast amount of power supply infrastructure, heavily impacting the basic living conditions of local residents and crippling critical facilities like hospital services. Accurately mapping the regions impacted by power outages and monitoring the timeline of power recovery from these two disasters helps to accurately allocate aid goods to where they are needed most. In addition, timely information regarding the stability of the post-disaster power supply is essential for aid groups and local governments to safely and effectively repair the power grid.

Under the GEO Night Light (Night-time Light Remote Sensing for Sustainable Development Goals) initiative, the team at Wuhan University, in collaboration with the United Nations Satellite Centre (UNOSAT), delivered two primary types of Earth Observation data products to UNOCHA and the International Charter Space and Major Disasters, which distributed these products to local governments for post-disaster assistance and reconstruction.

The first type of product is a time-series curve of the power supply index derived from VIIRS DNB data. This product reflects the continuous power supply situation on a daily scale. While the raw data allows for a time-series curve to be drawn for each individual pixel on the ground, the operational reports utilized an aggregated power supply index at the administrative region level. Consequently, decision-makers could easily track the power recovery trajectory for specific districts. Notably, the foundational data source is NASA's VNP46A2, which the team meticulously processed by correcting observation geometric effects and applying gap-filling algorithms to reconstruct data obscured by heavy cloud cover during the storms.

The second product is a high-resolution change detection map utilizing imagery from the SDGSAT-1 satellite, developed and operated by the International Research Center of Big Data for Sustainable Development Goals (CBAS). This change detection map provides granular power supply change information at the spatial scale of buildings. By comparing pre- and post-disaster imagery, emergency responders can pinpoint exactly which critical facilities—such as individual buildings or shelters—have lost or regained power whenever an emergency request is activated.

The data sources used in this application are highly specialized yet seamlessly integrated. The first primary source consists of night-time light data derived from the VNP46A1 and VNP46A2 suites of the NASA Black Marble product. Black Marble is a premier standard data product for night-time light observations. The primary sensor for this dataset is the Suomi NPP/VIIRS Day/Night Band (DNB), which offers a 742-meter spatial resolution at the nadir point. The VNP46A2 product provides atmospheric and moonlight-corrected night-time radiance from the Earth's surface, alongside essential quality flag layers that help mask out noise. Concurrently, the VNP46A1 product provides crucial viewing angle information (including viewing zenith and azimuth angles). Together, these datasets offer a comprehensive, daily-scale reflection of night-time light dynamics and their environmental impacting factors.

The second data source is high-resolution night-time light imagery acquired by the SDGSAT-1 satellite, operated by the International Research Center of Big Data for Sustainable Development Goals (CBAS). This satellite provides a multispectral night-time light RGB band at a 40-meter resolution, as well as a panchromatic band at a 10-meter resolution. Typically, the 40-meter multispectral data is preferred because its true-color composite is far more visually intuitive for human analysts. This high-fidelity dataset accurately reflects power supply conditions at a very high spatial resolution corresponding to building scales.

The third type is auxiliary data, including administrative boundary shapefiles and vital Point of Interest (POI) data, such as the exact locations of hospitals and building footprints. This auxiliary dataset is able to convert raw image pixels into actionable ground information, linking specific location names to their corresponding night-time light dynamics.

The application of night-time light remote sensing has successfully transitioned beyond the academic community, proving to be an indispensable operational tool in disaster mapping and humanitarian aid. During massive climate-driven crises like the 2022-2023 Pakistan Floods and the catastrophic 2023 Libya Floods, localized power outages were accurately mapped and promptly delivered to critical end-users, including UNOCHA and the International Charter Space and Major Disasters. This real-time intelligence made a tangible impact on local aid work, optimizing the distribution of life-saving supplies to areas cut off from the grid.

Beyond climate-driven events, this methodology has also been crucial in responding to geophysical disasters, notably the devastating 2023 Turkey Earthquake and the 2023 Morocco Earthquake. Detailed analyses of the recovery process following the Turkey earthquake revealed that while power restoration occurred, it exhibited a high degree of spatial heterogeneity. These nuanced insights have directly contributed to the groundwork of the UNHCR and numerous non-governmental organizations, ensuring that vulnerable, slower-to-recover communities receive targeted support.