Two previously developed GIS models – a static and dynamic mode – were populated with data for buildings in Wellington city and simulations run to determine the consequence of postearthquake fire. The models were corrected for shortcomings in the spread of fire between buildings with non-combustible claddings and the effect of vegetation between houses and suburbs.

Total property loss due to fire-spread between buildings with non-combustible claddings was assessed by developing and testing additional rules for the dynamic model. Using a survey of the buildings within the Wellington CBD, it was determined that the dominant fire-spread mechanism for these buildings was via non-fire rated roofs or openings in the walls.

Vegetation (between buildings and suburbs) facilitates fire-spread where it may not otherwise occur. Techniques were developed to incorporate vegetation into the static model. A pilot study in Karori suggests that loss estimates made without including vegetation may be 50% of those made with vegetation included, assuming all vegetation is flammable.

The dynamic fire-spread model was also modified to determine the spread of rural wildfire (firespread between built-up areas via intervening bush and scrub ) in a pilot study. The cell-based technique of modelling used could not accurately model the effect of wind or slope when the direction of maximum spread was not a sub-cardinal direction.