September 28, 2018 by Greg Meckbach
Climate change has turned the risk of wildfire into a year-round peril in Western Canada, a geography expert says.
In the past, wildfire season started in April or May and ended some time in October, said Dominic Ford, senior geospatial analyst for Verisk Analytics Inc.
“That’s not really the case anymore,” Ford said during a presentation Tuesday at the Insurance Analytics Canada Summit. “Whether you are looking at California or British Columbia, wildfire season is pretty much all year round now and never really stops.
“With climate change, we are seeing increased wind levels that change the way fires spread. They spread more rapidly, they are harder to control, they are harder to contain, and that puts additional risk on to your property and homeowners.”
Wildfires in the summer of 2017 caused about $127 million in insured losses in British Columbia, Catastrophe Indices and Quantification Inc. (CatIQ) reported at the time. Before this year, 2017 was the worst wildfire season ever in B.C. However, the 2018 wildfire season was the worst in B.C. when measured land burned – more than 13,000 square kilometres – The Canadian Press reported this week.
The 2016 wildfire in Fort McMurray, Alta. set a record because it cost the P&C industry nearly $4 billion.
Worldwide, insured losses from wildfires in 2017 were about US $14 billion, the highest ever recorded, Swiss Re said in a report released in early 2018. Climate change will increase frequency and severity of wildfire because of warmer temperatures and longer drought periods, Swiss Re added in Natural catastrophes and man-made disasters in 2017.
Global warming “has reduced the accumulation of snowpack during the cool-and-wet winters, and brought forward the onset of spring,” Swiss Re noted. “In turn, this has led to extension of the warm-and-dry summers in which wildfires can start and spread.”
Artificial intelligence and machine learning is currently being used to make wildfire risk models from data sets; the data sets can be as large as the equivalent of 1,000 full-length movies. Data includes topography, road networks and images created by remote sensing.
AI is already being used in software that looks at electronic photos and detects things like property damage. “But when you are looking at remote sensing data, satellite and aerial imagery, the challenges are a little different,” Ford said. “It’s not the same as picking out a cat and saying ‘This is a cat, this is a house and this is damage to your car.’”