House loss patterns in the Australian 2019–20 fire disaster confirm the reliability of official risk mapping

This research paper examines the spatial patterns of house and building loss during the 2019–20 wildfire disaster in New South Wales (NSW), Australia, to test the reliability of official wildfire risk mapping. The study used the NSW Government’s Bush Fire Prone Land (BFPL) mapping and analysed data from Building Impact Assessment (BIA) surveys, which tracked 8656 destroyed buildings (2580 houses) after the fires, and Microsoft Building Footprints (MBF) to assess loss patterns by distance from flammable vegetation. GIS mapping was employed to calculate proximity, and statistical models evaluated the influence of fire weather and vegetation type. The research focused on buildings within, adjacent to, and beyond 100 m from mapped flammable vegetation and considered land use, seasonality, and fire danger ratings as key variables.

The findings demonstrate that 81% of house losses and 80% of building losses occurred within flammable vegetation, while 94% of house losses happened within 100 m from it. Buildings beyond 200 m were at minimal risk, but under extreme fire weather, losses extended further from vegetation. Most losses occurred adjacent to eucalypt forests, especially in summer. The authors conclude that the current 100 m buffer in NSW risk mapping accurately predicts house loss risk overall but may be less reliable in increasingly extreme fire conditions due to climate change. They recommend reconsidering the buffer for grasslands, improving risk mapping, and using intensive fuel management near wildlands, alongside more detailed house-level assessments to refine predictions and policy.