A 10-year-old satellite built for daylight imaging just started seeing Earth after dark. ESA switched Sentinel-2A on at night in December 2025. It was a test, not a normal mission mode. Sentinel-2’s multispectral imager usually relies on reflected sunlight, so the instrument is normally switched off when the satellite passes over the dark side of Earth. This time, ESA left it running. The targets were carefully chosen: gas flares, bright lights, wildfires, fishing activity and marine activity. The campaign covered selected locations during one full Sentinel-2A cycle, with Level-1B data collected in specialist modes. The results were better than expected. Sentinel-2A captured gas flares from oil production in the Middle East. That was the easier part, because flares are bright enough to stand out clearly at night. Then it picked up subtler signals: a wildfire in India, fishing boats off South Korea, gas flares in Iraq and Qatar, and city lights. That’s the interesting part. Sentinel-2A launched in 2015 to observe land, coasts, forests, farms and water in daylight. It carries a 13-band multispectral imager, has a 290 km-wide swath, and became one of the core workhorses of Copernicus, Europe’s Earth observation programme. Now it’s the older sibling. Sentinel-2B launched in 2017. Sentinel-2C launched in 2024 to take over from Sentinel-2A. Sentinel-2D is still to come. With the newer satellites handling the main operational mission, ESA is using Sentinel-2A’s remaining time in orbit to test ideas for Sentinel-2 Next Generation. Night imaging is one of those ideas. The next generation mission is being designed to deliver higher resolution imagery and to image some parts of the planet when the Sun is down. Sentinel-2A is helping ESA work out which signals are useful enough, strong enough and technically realistic enough to build into that future mission. Gas flares could support emissions monitoring. Fishing boat lights could help track marine activity, including poorly monitored fishing. Wildfires could be easier to detect in some night scenes because the background is darker. City lights could reveal urban expansion, infrastructure growth, electrification, disaster recovery and conflict impacts. For researchers, the spatial detail is the real prize. Most nighttime lights work relies on sensors such as DMSP, VIIRS and Black Marble. Those datasets are incredibly useful, but they’re much coarser than Sentinel-2. Sentinel-2 can get down to 10 to 60 metre resolution depending on the band. That opens the door to seeing night activity at the scale of ports, road corridors, settlement edges, mine sites, flare stacks, fishing clusters and fire lines. There are constraints. This doesn’t turn Sentinel-2 into a full-time night lights mission. Energy use matters. Calibration matters. Signal strength matters. Target selection matters. Processing Level-1B nighttime data also takes specialist care. But the direction is clear: the old boundary between “daylight optical imagery” and “nighttime activity monitoring” is starting to move. The data is also opening up. The December 2025 campaign was first limited to expert users through the Copernicus Data Space Ecosystem. In May 2026, ESA said a consolidated nighttime dataset from the campaign would be made public to all users, expected by the end of Q2 2026... I'm looking forward to this...