Planned burning is an important fuel management tool applied across thousands of hectares of Australian forest and woodland each year. Remote sensing is being successfully applied to determine the severity of wildfires; but its application to planned burning is more difficult because the effects are generally less severe and may be masked by unburnt vegetation in the shrub or canopy layers. The aim of this study was to investigate the use of the Landsat 8 Operational Land Imager (OLI) sensor to deliver fire severity assessments of planned burns in Australian forests and woodlands. Landsat 8 OLI was selected for investigation because its spectral and spatial (30 metres) resolution is suitable for detecting effects of fire, and the data are openly available on a routine basis over all of Australia.

The procedure was based on the FIREMON method and tested using a 938ha planned burn in the Australian Alps which was conducted from 30 March – 4 April 2015. This burn was selected for analysis because it included a wide range of fire severities due to an intensely burning escape on the eastern edge. Cloud-free Landsat 8 OLI images were acquired on 3 January and 15 April 2015 and a Normalised Burn Ratio (NBR) algorithm was applied. NBR is the normalised difference index of the near infra-red and short-wave infra-red bands of the Landsat 8 OLI sensor. Ground-truthing was conducted 23rd-24th April at 105 plots using a modified version of the FIREMON method. Pixels were assigned to one of three classes using the change in NBR values: Unburnt (-0.66-0.15), Low (0.15-0.42) or High (0.42-1.10).

The training accuracy of the classification was 81% and the Kappa statistic was 0.66. This level of accuracy is comparable with other studies and indicates substantial agreement between the mapping and the ground-truthing.