While this may well be the case for the test fires lit for Project Vesta it is not necessarily the case elsewhere. In fact, much experience in the ACT region suggests that FFDI over-predicts potential fire behaviour (specifically rate of spread, ROS). A paper by Steve Kessell (using data from NSW National Parks & Wildlife Service) in 1990 looked at 32 fires in which observed fire behaviour was compared to predictions from FFDI and the 3-strata Rothermel model. In general, Rothermel predicted well, while FFDI grossly over-predicted.
I would thus recommend that we treat predicted fire behaviour from FFDI as being potentially unreliable, without suggesting in which way it would be in error.
Having said that, an awful lot of the error that comes out of fire behaviour predictions arises not from the meter used, but from the inputs used. The potential errors in temperatures, humidity, wind, fuel and slope are vast, and a level of expertise is needed before reliable predictions are possible.
FFDI was based solely on available fine fuel – ie ground fuels. It is patently obvious to anyone with fire experience that this is incorrect. The spread of a fire into shrub layers, if not the canopy, can drastically alter the behaviour of the fire.
Observations of local fires suggest that a simplistic assumption of "shrubs mean a higher ROS" is wrong.
While the greater fuel loads due to shrubs will cause larger flames and possibly higher ROS, the wind speeds experienced by the flames may be lower, due to the increased resistance to wind from the shrub layer. The balance between these two factors will reflect all conditions at the time.
Certainly a fire in shrubs will have higher flames and higher intensity than one in surface fuels only. All we can say about its ROS is that it may increase.
Importantly, we need a fire model that considers all potential fuel sources.
A fire will generate a convection column of rising heated gasses and smoke. Air is drawn in along the ground from all directions to replace those gasses. The net effect is an indraft, around the fire perimeter, which carries the flames inwards, leaning over burnt ground. Any bulk wind will work to override this effect at the head of the fire (and reinforce it elsewhere around the perimeter). When the bulk wind exceeds the indraft, the flames will start to lean into unburnt fuel, causing a rapid acceleration of the headfire. Observations of mild fires suggest a typical indraft of around 3 km/hr. A 3 km/hr wind is needed to override this, but this is the wind under the forest canopy. In eucalypt woodlands, this is about 20% of the bulk winds (as used in forecasts). Thus 3 km/hr under the canopy is equivalent to a forecast wind speed of about 15 km/hr.
Care is needed in other forest types. For example an unthinned pine plantation may have wind at flames at around 10% of bulk winds, while an open, thinned plantation may get 40% of bulk winds at the flames. These correspond to bulk winds of 30 km/hr and 8 km/hr to override the indraft. Obviously careful observation of the effects of the vegetation structure on winds at flame level are needed. Further when a fire burns into vegetation with less canopy wind resistance, it becomes more likely that strong winds will occur at flame level (given the same bulk winds). The fire is then likely to accelerate.
There are however other factors that affect wind speeds that must be considered as well.
When working at any of these tasks it is important to recognise that there are concerns that you may not recognise that are significant for others. Thus, for example, a stable weather outlook may be the basis for setting the objectives and strategies at a fire, while the tactics for some sectors may need to be based on gusty, unpredictable winds.
It is a major issue with predicting fire behaviour that it is scale-dependent. The fire itself sets the scale, and we have to base our prediction on information that is appropriate for that scale. (As an extreme instance, you wouldn’t refer to a world atlas to get slope data. Incorrectly sourced slope data can have a major effect on predicted ROS.) So, a line fire has a different scale than a point fire, and thus its behaviour would be different, all other things being equal. A backburn lit by a series of spot fires will burn slowly until they start to merge, then they will acquire a larger scale and accelerate (i.e. become like a line fire).
However, all firefighters assigned to large fires need to know enough about the subject to predict potentially dangerous changes in fire behaviour in their sector.
Australian Capital Territory Emergency Services Bureau
Concepts and Risk Management