Assessing mitigation of the risk from extreme wildfires using MODIS hotspot data

Abstract

A taxing question in wildfire risk management concerns the effectiveness of fuel reduction burning on mitigating the risk to Australia's communities from extreme wildfires. Extreme wildfires account for the majority of wildfire damage over recent years. In part the uncertainty arises from how questions are framed, in part from the lack of effective datasets to conduct definitive analyses. The development of a climatology of satellite-derived fire hotspots over the last thirteen years has permitted new approaches to the problem that yields answers that may provide valid insights into improving community safety. The complexity of the problem necessitated focusing on a study domain. The effectiveness of fuel reduction arises from the inter-annual accumulation of leaf litter in eucalyptus dominated forests. The forests of mainland southeast Australia were selected to meet that criterion in a well understood setting. Within the domain, included hotspots for each fire year were separated, aggregated and used to generate a 2km radius buffer area. This was designed to minimise issues associated with hotspot positional errors and to, in some sense, account for associated milder fire areas that would not trigger the hotspot algorithm. Small scale or low intensity fuel reduction burns are missed by this approach, but they are unlikely to be effective against an extreme wildfire. If intensive fuel reduction were effective, then there would be an expectation of burnt areas in preceding years affecting, in a measurable way, the extent of an extreme wildfire. To do this, a subset of the buffered area was generated for hotspots associated with extreme wildfires. These were identified based on the expected spatiotemporal clustering of their hotspots and on-going research into these events. Two hypotheses are explored. Firstly, as is often claimed, fuel reduction burning may be effective. Overlaps of specific sets of hotspots were analysed with respect to this. If this is not the case, this may be due to the small proportion of the domain burning each year, making overlaps unlikely. Secondly, if fuel reduction burning is not effective, this may be due to remoteness and ruggedness. The distributions of extreme wildfires and other fire were compared with respect to these environmental parameters. By examining the four preceding years it was found that some extreme wildfires were not affected despite appreciable overlap with preceding fuel reduction activity. The main proportion of an extreme wildfire that was recently burnt is estimated at 19%, including the downwind edge. In terms of distance in from the edge of a forested area, it was found that fuel reduction occurred mainly towards the edge, while extreme wildfires were relatively much more frequent in the interior. Further, extreme wildfires present greater operational challenges in rugged landscapes. It was found that fuel reduction is more frequent in flat landscapes, while extreme wildfire is most common on rugged landscapes. The implications of these patterns for mitigation of the impacts of extreme wildfires are discussed. Statistically, most extreme wildfires are unlikely to encounter previously burnt areas, but that this does not imply that the latter prevents the former. Where they do overlap there is little indication of an interaction. Mitigation of the risks arising from extreme wildfires through fuel management, ignition prevention or response arrangements, must take account of the nature of these fires.