Satellite imaging and artificial intelligence are among the technologies being leveraged to help insurers model and manage wildfire risk, a natural peril that is far less understood than hurricanes or earthquakes.
The current Los Angeles fires – expected to be among the worst-ever US wildfires in terms of insured losses – are heaping pressure on underwriters to continue providing property cover in areas they had already deemed high risk.
Even before this latest disaster, insurers had been withdrawing from the region. The total exposure of California’s fire insurer-of-last-resort, the FAIR Plan, had rocketed to $458bn by September 2024, compared with $153bn in 2020. The number of FAIR policies in force more than doubled over the same period, to 452,000 from 203,000.
Part of the solution for insurers returning to California is better modelling, and permission to use those models in setting rates.
To-date, wildfire modelling has lagged behind modelling of other natural catastrophe perils, such a hurricanes and earthquakes. In InsuranceERM’s special report on the subject, published this month, wildfire modelling experts discussed the challenges of wildfire modelling.
“It’s almost unimaginable how complicated it is to simulate what happens in a real fire,” said Nancy Watkins, principal and consulting actuary at consultancy Milliman.
Daniel Bannister, weather and climate risks research lead at broking and advisory firm WTW, said: “One of the main problems in wildfire modelling is predicting when and where fires will start, especially since nearly 80% of wildfires are caused by humans.”
“It’s harder to predict human behaviour than it is to model natural events like windstorms,” he added.
But advances in satellite and drone imagery, combined with the use of AI, are helping wildfire insurers understand property risk and portfolio aggregations better than ever before.
According to Sarah Russell, managing director at mapping and spatial analysis firm Bellwether that sits within X, the innovation arm of tech giant Alphabet: “AI has changed everything for the field of wildfire risk. It has made it possible to do things that we didn’t think possible, like infer and forecast ember spotting. AI makes it possible to learn the dynamics of fire risk in a given geography and then apply that learning in a brand new, ‘foreign’ land.”
- Read more about wildfire risk modelling in InsuranceERM’s special report Natural Catastrophes: Flood, Fire & Storm, free to download here
- InsuranceERM and Bellwether are also hosting a free webinar on wildfire risk management on 19 February at 3pm GMT. Register to join here