Comparing the performance of daily forest fire danger summary metrics for estimating fire activity in southern Australian forests

Abstract

Fire danger indices integrate weather and fuel variables to indicate the potential for wildland fires to ignite, spread, resist suppression and cause damage. McArthur’s Forest Fire Danger Index (FFDI) is applied across much of Australia, with the forecast daily maximum value used to inform fire management planning decisions and issuance of public warnings. Variations in daily maximum FFDI and the hourly changing of FFDI values during the day (including use of different soil moisture deficit indices) were compared against five binary fire activity statistics in six forested areas in southern Australia, with performance assessed using Theil–Sen regression lines fitted to rank percentile curves. Fire activity rates were similar on days with wide and narrow hourly FFDI distributions except in one study area where days with wide distributions experienced more fires. The maximum hourly FFDI metric performed the best of all the metrics tested, though there were no statistically significant differences among any of them. There was also little difference in the performance of metrics determined using alternative calculations and different drought indices. These results suggest that the current use of the forecast hourly maximum FFDI is appropriate and that using alternative methods to determine Drought Factor offers little benefit.