Cambridge researchers are using AI to predict climate disasters by identifying landslides in satellite data within hours, providing emergency teams with faster and clearer maps to target rescue efforts after earthquakes and extreme storms.
When natural disasters strike, time is everything. However, traditional damage assessments can take days or even weeks—precious time that can cost lives in isolated or hard-hit areas. Now, scientists at the University of Cambridge have created artificial intelligence that can detect landslides in satellite images within hours, offering a breakthrough tool for emergency response.
The system using AI to predict climate disasters was put to the test after Taiwan’s 7.4-magnitude earthquake in April 2024, the strongest to hit the country in a quarter century. Thousands of landslides ripped through mountain communities, burying homes and blocking critical roads. Within just three hours, the AI identified more than 7,000 landslides from satellite data—a process that normally takes human analysts several days.
The innovation lies in how the system processes information. Standard optical satellite images provide clear detail but fail when cloud cover or nightfall hides the ground. By combining optical data with radar imaging, which works in all weather and lighting conditions, the Cambridge team created a system that remains reliable exactly when it is needed most.
The scale of the problem underscores the urgency of this technology. Landslides already cause more than $20 billion in damages globally each year, and climate change is exacerbating their frequency and severity.
Heavy rainfall, deforestation, and rising temperatures destabilize slopes, turning extreme weather events into cascading disasters. Recent tragedies in Europe highlight the risks: a glacier collapse in Switzerland triggered a deadly landslide, while record floods in Valencia overwhelmed the city’s defenses.
The AI system is part of CoMHaz, a Cambridge program focused on understanding how disasters compound one another. Earthquakes can trigger landslides, which then block rivers and create flooding, multiplying the devastation. By detecting landslides rapidly, emergency teams can anticipate these chain reactions before they escalate further.
Real-world trials using AI to predict climate disasters are already underway. In Nepal, the system is being trained on local terrain around Butwal, a city threatened by unstable slopes. Here, the goal is to create an early warning system that predicts risks before they occur, allowing residents to evacuate in time.
One of the major hurdles for AI disaster tools has been trust. Emergency managers are cautious about relying on automated predictions for life-or-death decisions. To address this, the Cambridge researchers are designing their system to be transparent, showing the features in satellite images that triggered each alert. Partnerships with the European Space Agency and the World Meteorological Organisation, along with open data challenges, are helping refine both accuracy and reliability.
The potential impact of using AI to predict climate disasters extends well beyond landslides. By providing rapid and reliable damage assessments, the AI allows rescue teams to prioritize their resources, reopen vital roads sooner, and reach communities that might otherwise be overlooked. Similar systems could eventually be adapted to track floods, wildfires, or even urban building collapses, offering faster responses to a wide range of disasters. Beyond emergency response, the technology could also support long-term planning by identifying vulnerable infrastructure, guiding safer construction, and helping governments allocate funds more effectively.
For communities living in landslide-prone regions, this technology could mean the difference between survival and tragedy. It represents a new frontier where artificial intelligence moves from abstract theory into a practical, life-saving tool, helping societies confront the accelerating threats of climate change. By merging advanced computing with on-the-ground needs, the system highlights how innovation can enhance resilience, safeguard lives, and offer hope in an increasingly unstable world. Together, these advances remind us that human ingenuity remains our strongest defense against nature’s most unpredictable forces.
