Fire weather, atmospheric stability environments, and fire activity in southern Australia

There is a general acceptance in the Australian fire community, as there is internationally, that enhanced or erratic fire behaviours are associated with unstable atmospheric environments, and that these effects are not represented in the widely used fire danger indices that are more closely related to wind driven fire behaviour. The Haines Index (HI) is widely used to identify such unstable environments. However, in southern Australia a number of studies, plus undocumented but extensive experience, show that the HI "saturates", with a very large proportion of summertime days having HI greter or equal to 5 over vast areas, and so does not discriminate the extreme days. An exception is in Tasmania, where Bally (1995) showed that there was a relationship between the HI and days of greater than normal fire activity that was not accounted for by McArthur's Forest Fire Danger Index (FFDI).

Using NWP data to circumvent the spatial and temporal sampling deficiencies (in a fire weather sense) of Australia's radiosonde network, an 8-year spatial data set of the HI, and other measures of atmospheric stability, has been generated and the climatological distribution of HI and its temperature lapse and dewpoint depression ingredients have been analysed. Based on the climate of the ingredients an open-ended extension of the HI is proposed -the "Continuous Haines" (CH).

To assess the relation between this new CH and the other stability indicators calculated, and fire events, a number of examples of extreme fire behaviour, notable pyrocumulus cloud development, or where greater than expected fire activity during prescribed burns was reported was accumulated - some 60 event-days in total. Plotting NWP fields for these events showed that the CH forecast focused maxima in the region of the major events, and provided, from a meteorological perspective, far more informative spatial fields than did the traditional HI.

This does not, of course, show that CH provides information additional to the FFDI about these cases. Accordingly, for each event the value of the CH on that day was compared with its climatological distribution at that location, and a similar comparison made for the FFDI. Comparing CH with FFDI for these events shows that

1. CH shows some sensitivity in identifying the most extreme events compared to FDI in terms of percentiles of the local climate of each index.

2. A number of prescribed fire events that showed unexpected activity show that CH was climatologically extreme when FDI was not unusual

3. A number of nocturnal breakout events were associated with very high values of CH, in situations of relatively low, or decreasing, FFDI.

4. Most of the spectacular pyrocumulus events had statistically extreme FFDI as well as statistically extreme CH, suggesting that the huge moisture release from very large consumption rates is an important factor in these developments.

5. There are examples where a weak frontal wind change had passed the fire ground, with decreasing FFDI, but observed deep pyrocumulus development and observed high CH suggests that the fire was interacting with the layer of atmosphere above the shallow frontal inversion.

Other measures of atmospheric stability do not overall appear to correspond as closely with observed fire behaviour as does the CH, in spite of their arguably greater physical basis. This needs to be further explored, but may well to be due to the fact that the CH includes both lapse rate and humidity information. Assessing the ingredient fields of temperature lapse and dewpoint depression shows that for some of the fire events the CH maxima were temperature lapse dominated, while for others the dewpoint depression dominated, and that the maxima in fields of ingredients did not necessarily coincide. These suggest that perhaps there are more than one physical processes associated with atmospheric stability that might enhance fire activity, and this is a fruitful avenue for further research.

However, the evidence presented so far for the application of CH in operations is only encouraging. It is desirable that a rigorous comparison with some fire activity measures be conducted, but to date fire activity/behaviour data sets in Australia are not well suited to this analysis. It is hoped that this will be alleviated in the future. In the meantime, extra-ordinarily positive forecaster feedback was received about the utility of CH on the day of the catastrophic fires in Victoria in February 2009, which, while not quantitative, is extremely gratifying.