Object orientated image analysis of high resolution imagery was undertaken as part of a review of the bushfire hazard and threat across a local government area in South Australia. The project included a GIS-based analysis of bushfire hazard and threat to assets across the 500km² study area. A key aspect of the project was the mapping of fuel loads across this large area which was undertaken through a process of satellite image analysis and interpretation, and on-ground fuel load mapping.
A hybrid unsupervised spectral classification and object oriented remote sensing methodology was developed to map vegetation density based on high resolution satellite imagery. Imagery from two dates in 2004 and 2007 were analysed enabling change detection to assess fire events that occurred between 2004 and 2007 and other changes to fuel loads. The hybrid approach enabled mapping of individual scattered tree canopies from the spectral classification as well as attribution of vegetation patches with a density code from the object orientated analysis.
The main changes in vegetation density and fuel load were attributed to bushfires, timber harvesting and increases in vegetation density where revegetation had been undertaken several years ago. This analysis was effective for mapping changes in fuel load at a regional scale but is not applicable for detecting changes at a local property scale. In addition, because the fuel hazard in the burnt areas is likely to increase quickly over the coming year, the assessment must be recognised as a snap-shot coverage.

The resulting data combined with other asset and topography data was used in a Bushfire Risk Assessment Project. The data was converted to raster format and summed using an existing formula to calculate the risk and threat areas. A range of bushfire threat and hazard maps were produced.