Kimberley Opie

Incorporating vegetation attenuation in radiant heat flux modelling

Models of radiant heat flux (RHF) are critical for understanding wildfire behaviour and the effect a fire may have on homes and people. Various models have been presented in the literature for wildfire RHF, many being based on the Stephan–Boltzmann equation for radiative heat transfer. Most models simplify the fire and receiver interaction by considering a single fuel type at a given separation distance from a receiving point (e.g. on a building requiring protection). However, wildfire is an inherently spatial phenomenon, in that a fire may progress across the landscape towards a building across complex terrain and through spatially varying fuel types. This spatial variation influences the fire behaviour as well as the level of RHF incident on the building. In this study, we present methods for incorporating spatially varying topography and fuels into existing RHF modelling equations. In this way, we achieve a time-dependent profile of the RHF incident on homes, while accounting for attenuation due to fuels and topography that lie between the building and the fire front. The model is applied to the prediction of damage in a fire that occurred in South Australia in 2005. Although only coarse spatial information was available for determining the spatial distribution of fuels, modelled RHF was a significant indicator of house damage. Attenuation due to vegetation between homes and the fire was shown to reduce the modelled RHF exposure of homes. However, this was not shown to increase the significance of predicted house damage in the case of this fire event.

An automated method for normalising large Landsat time series datasets to like values for change detection

The increased availability of Landsat imagery provides the opportunity to monitor changes in the land surface over long time frames. This paper presents an automated method for normalizing satellite images to a calibrated reference image using pseudo invariant features (PIFs). The method is repeatable, objective and fast, and has been used to prepare a large collection of Landsat images for flood inundation detection.