Accurately Predicting Fire Season Severity From Analysis of Historic Fire Climatology
BY ANDY ACKLAND AND MUSA KILINC
Climate change is increasing the severity of bushfire seasons across the globe. There is a growing need for fire managers to be more proactive and less reactive in planning for severe bushfires.
Recent trends in bushfires are concerning. Victoria, Australia, has had 5.8 million hectares burned by bushfires since 2002-03; of that, four million hectares were burned in three events of 1.2 million hectares or greater.
While it is important to look forward and better understand the effects of climate change on bushfire risk and fire management, much can be learned and applied from looking back and comprehensively analysing historic fire climatology.
In Victoria, Andy Ackland developed a fire climatology analysis system that has been operationally applied to seasonal preparedness planning since 2007. This system compiles details of bushfire house loss, fatalities, and area burned, along with monthly rainfall and climate patterns for every bushfire season back to 1900. The system provides a powerful platform for investigating rainfall and climate patterns associated with Victoria’s worst seasons. Linear plots (locally termed spaghetti plots) provide a simple means of comparing to analogue years, such as those that have similar climate drivers such as El Nino.
A single state-wide indicator, called Victorian cumulative monthly rainfall anomaly, tracks the severity of drought leading into the fire season. While a simplistic measure, this area-averaged rainfall anomaly has proven to be particularly effective because it is most driven by Victoria’s highest rainfall areas, which contain the state’s most bushfire-prone rugged eucalypt forests.
The system reveals that all of Victoria’s worst bushfire seasons (including 2019-20, 2008-09, 200607, 1982-83 and 1938-39) followed a period of severe cumulative rainfall deficit commencing the autumn prior to the fire season. Virtually all seasons with a January-to-October anomaly beyond -150 millimetres go on to have severe bushfires.
This system allows accurate identification of a concerning rainfall pattern as early as the autumn prior; this was the case ahead of the catastrophic 2008-09 Black Saturday and 2019-20 Black Summer seasons, with such analysis prompting early seasonal preparedness planning and communications.
The system also enables accurate early prediction of lower-severity bushfire seasons (such as 2021-22 and 2022-23) from tracking climate drivers that give rise to wetter (positive) cumulative rainfall anomaly, such as the combination of La Niña and negative Indian Ocean Dipole.
Furthermore, the system reveals some recent concerning changes in patterns of rainfall anomaly and bushfire season severity.
The chance of having a serious season (or worse) has changed from 22 per cent annual exceedance probability or AEP (approximately five times a decade) to 33 per cent AEP (approximately 2-3 years). In other words, there’s now a 33 per cent chance any season will be at least severe (three fatalities and/or 100 houses).
The chance of getting at least an important season (up to 30 houses lost or a single fatality) has increased to 57 per cent from 36 per cent. In other words, there’s now a 57 per cent chance in any season there will be up to 30 homes lost and/or one fatality.
Development of this fire climatology analysis system has not just been for scientific and analytical purposes, but also to ensure it is clear, simple, and useful for operational staff involved in fire season preparedness planning. Each year, fire managers face difficult, multi-million-dollar investment decisions regarding preparedness planning for the next fire season. Decisions typically need to be made in July and August for fire seasons that typically peak in in January and February in Victoria – six to seven months ahead. There can be significant implications, particularly if fire agencies are under-resourced leading into a major fire season, but also if over-resourced for a low-severity season. This system has led to accurate prediction of fire season severity virtually every year since 2007. Senior operational staff now regularly ask for the system’s data to help inform their preparedness decisions.
Referring to analysis of historical fire climatology alongside analysis of regional rainfall, fuel conditions, and fire occurrence has enabled accurate prediction of above-average, average- or below-average Victorian bushfire potential every year for the past seven years in Australia’s quarterly National Seasonal Bushfire Outlook, co-ordinated by the Australasian Fire Authorities Council. There has been found to be strong statistical predictive power in referring to patterns from similar historic seasons, particularly those climate drivers that give rise to very wet or very dry seasons.
Climate change is clearly increasing the threat of bushfires across the globe. A comprehensive understanding of our historic fire climate can form a powerful reference or foundation for understanding future change. There is predictive power in historical fire climate analysis, but it is important to contextualise this in a shifting climate – recent trends seem to be showing concerning persistence, with fire climate patterns from the earlier 1900s showing less relevance in a warming world.
ABOUT THE AUTHORS
Andy Ackland and Musa Kilinc are predictive services specialists with the Country Fire Authority, Victoria, Australia. Ackland has worked as a predictive services specialist within the Country Fire Authority in Victoria since October 2020. Before that, since 2008, Ackland worked in the Victorian Department of Environment, Land, Water and Planning where he led development of bushfire risk modelling capability using PHOENIX Rapidfire bushfire simulator. Ackland is an experienced fire behaviour analyst, with deployments to South Australia, Tasmania as well as British Columbia, Canada, in 2009 and 2015.
Kilinc joined the Country Fire Authority in 2015 as a predictive services specialist. Since joining CFA, Kilinc has supported the development of bushfire simulator capability and development. He has conducted various research investigations with colleagues looking into ignitability and the development of fire spread models. Kilinc has also contributed to the evaluation of fire behaviour models and is a strong advocate in science translation. Kilinc works as a fire behaviour analyst at the state and incident level control centres during the summer period and supports capability development of FBANs in Victoria.
