AI-driven maps track disappearing Arctic lakes and permafrost thaw in near real time
Scientists have launched an online platform that uses satellite photos and artificial intelligence to monitor permafrost melt across the Arctic as lakes appear and vanish.
Researchers combined satellite data, historic aerial photos and machine-learning tools to build a live-update map of permafrost degradation throughout the Arctic. The platform flags thermokarst ponds—water-filled depressions that form when ice-rich soil melts—and tracks lakes that drain abruptly, both clear indicators of subsurface thaw. An international collaboration via the Permafrost Discovery Gateway has catalogued over 4 million lakes and 70 million trough ponds, enabling rapid detection of changes.
In the past two decades, the north-western Alaskan peninsulas have lost many long-standing lakes, with nearly 200 shrinking dramatically in the summer of 2018 and about 30 more draining in June-July 2026 despite a colder winter. While seasonal water-level shifts and wildlife activity also affect lake size, the new near-real-time data give planners concrete guidance on safe construction sites and help scientists study thaw dynamics more closely.
Why it matters
Real-time permafrost maps let Arctic residents and planners avoid damage from sudden ground collapse.
In this story
