Insurable losses due to wildland fire: can spatial models provide useful estimates of risk?

点击翻译按钮获取中文摘要

Wildland fires pose significant threats to the settlements, infrastructure and industry embedded in Canada's forested lands. The recent mass evacuations and extensive structural damage in Fort McMurray, Alberta represented four billion dollars of insured losses. The frequency of such events and the exposure to losses are likely to increase due to climate warming, and continued industrial development and population growth in forested regions of Canada. The resultant losses could threaten insurance company profitability or even viability. A common method to assess these risks is by means of computer-based models of fire occurrence and spread. In Canada, the wildfire simulation engine Burn-P3 is often used for this purpose. Our partner, Desjardins Groupe d'Assurance Générales (DGAG), is interested to determine if Burn-P3's estimated fire hazard maps are reliable enough for use in their business, and is also concerned about their scalability to annual predictions at national extents. This proposal was developed to seeks to assess and compare the reliability of Burn-P3 and an alternate hazard mapping methodology developed by the applicant. The two methods will be compared against each other and against independent fire-occurrence data over a large study area in the Abitibi-Témiscamingue region of Quebec, Canada. Their relative computational efficiency and scalability will also be assessed.