Determining house destruction functions using fire simulator outputs for landscape risk planning (Wildfire Conference. Day 2 - Wednesday Nov. 13th, 2019)

By Pau Costa Foundation on

2019 Wildfire Conference Adressing the Challenges of Bushfire Management

Presentations Notes 2019: DAY 1 (Day 2 - Wednesday Nov. 13th, 2019).

Determining house destruction functions using fire simulator outputs for landscape risk planning

Thomas J. Duff, Bianca Tumino, Trent Penman

Abstract:

The protection of human lives and property is a primary focus of managers in fire prone landscapes. As managers have limited resources to prepare for wildfires, determining where in the landscape to focus risk reduction efforts is an enduring challenge. The simulation of hypothetical fires across a landscape using fire simulator systems can allow potential fire behaviour to be visualised and the exposure of assets to be estimated. To quantify risks, relationships between fire exposure and asset impacts must be determined. We demonstrate an approach whereby impact functions are fitted using case study fires. We collated data for wildfires where there had been house destruction in Victoria Australia; including information on burnt area, fuel, weather, house locations and topography. Fire behaviour was reconstructed using the simulator PHOENIX RapidFire. Landscape data and simulator outputs were superimposed over maps of impact and these were used to train a Boosted Regression Tree to predict destruction likelihood. The best model explained over 50% of the variation in house impacts, with fire and landscape predictors each contributing to 50% of variation. Incorporating empirically fitted asset impact functions in fire simulation systems can provide for improved projections of likely risk and consequent efficient resource allocation.

Login or register to download this file