Bushfire conditions under a warming climate – the value of regional climate modelling

Long-term changes to the climate affect the incidence of weather conditions suitable for bushfires as well as the growth and dry matter accumulation that contribute to fuel loads. Estimating the changes to fire danger with a changing climate is an essential task for the long-term planning of bushfire response. Our best tools for projecting changes to the climate system are climate models, however the coarse scale of global climate models (GCMs) does not show the regional detail of changes. In this study, we use output from the dynamical regional climate model (RCM) named conformal cubic atmospheric model (CCAM) to examine the simulation of Tasmanian fire weather, and illustrate some advantages in using RCM outputs over GCM outputs. This study focuses on bushfire weather; the atmospheric conditions that influence the incidence of bushfire. The dynamics behind one synoptic pattern associated with elevated bushfire weather conditions are examined, and the pattern is more reliably simulated in the RCM compared to GCMs. This suggests that the greater resolution offers some advantages, and further investigation of other atmospheric dynamics associated with elevated bushfire weather conditions is justified using these model outputs. A standard index of bushfire weather, the McArthur Mark 5 Forest Fire Danger Index is calculated directly from the model output but assuming a drought factor of the maximum 10. The model output has a fine spatial scale (0.1 degree lat/lon) and temporal scale (3-hourly is used) and generates output variables that are directly comparable to the observations used to calculate fire weather indices. The temporal variability of FFDI at Hobart compares well to observations at all scales, from sub-daily to seasonal and inter-annual. The spatial pattern of cumulative FFDI, the frequency of events of elevated FFDI and the maximum FFDI also compare well to observations. These results suggest that the RCM simulations offer some more detail than GCMs, and are likely to be a useful tool for examining the change in fire danger indices with a warming climate. The results indicate that the fine resolution model output has some advantages over GCM outputs. The results also suggest that after more careful validation and ground-truthing from observations, RCM output can used directly to examine simulated events of high fire danger, rather than using the projected trend in temperature, wind and relative humidity as a basis to manipulate observed datasets. These results suggest that the direct model outputs are a useful tool to examine changes to fire weather dynamics and high fire danger events with a warming climate. Further analysis of fire weather dynamics and fire danger will be pursued using these model outputs.