How AI could become an unlikely ally in the fight against forest fires

Firefighters traditionally operate on satellite data, but it is incredibly slow, often with a 12-hour lag, rendering the data ineffective. With a quicker flow of information at their disposal, responders’ chances of tackling a forest fire effectively are vastly improved.

“Speed is one of the key points. We want to be accurate, but also we want to be very fast in the predictions,” said Dr Rossella Arcucci, an Assistant Professor in Data Science and Machine Learning at Imperial College London.

Dr Arcucci is another industry expert dedicated to utilising the power of AI to improve emergency responses to forest fires.

But rather than using robotics, her focus is on using social media to identify an inferno’s point of ignition and improve wildfire forecasting. She is working in partnership with the European Space Agency on this research.

Their AI system scans social media – from TikTok and Instagram to Snapchat and X (formerly Twitter) – for videos and images of the wildfire. The content is then filtered and checked to ensure its reliability. 

“If you are posting something about a wildfire, you are not in the middle, hopefully you are running … so there is an error of a few kilometres,” said Dr Arcucci. 

“We combine the geolocation from the social media post with the risk map and then [the AI software] gives the point with higher probability to be the ignition point.”

The error margins are very low, said Dr Arcucci, as the AI tool ensures cross-validation. “Usually when a wildfire is happening there is never only one person talking about it,” she added.