Fire in Dry Eucalypt Forest:

Transcription

1 Project Vesta Fire in Dry Eucalypt Forest: fuel structure, fuel dynamics and fire behaviour. Project Vesta is Australia s most recent and significant study of forest fire behaviour. This comprehensive research project investigated the behaviour and spread of high-intensity bushfires in dry eucalypt forests with different fuel ages and understorey vegetation structures. The project was designed to quantify age-related changes in fuel attributes and fire behaviour in dry eucalypt forests typical of southern Australia. What are the problems? After many wildfires there were reports of unusual and unexpected fire behaviour. Some fires spread more rapidly than predicted; others appeared to blow-up unexpectedly and exhibit violent fire behaviour when previously they had been burning relatively quietly. Existing systems for predicting forest fire behaviour were based on low-intensity fires and while they are excellent for planning and predicting prescribed fire they could underpredict the spread of high-intensity wildfires by a factor of 3 or more. Project Vesta provided the data to develop a better fire behaviour prediction system to predict the spread and intensity of wildfires. The effectiveness of prescribed burning for reducing fire intensity and making fire suppression easier and safer has been long recognised by operational firefighters. However, there are conflicting reports about the duration and scale of effects as fuel accumulates with age and there has been limited data quantifying the effects of fuel load and structure on fire behaviour. Project Vesta demonstrated that hazard reduction by prescribed burning will reduce the rate of spread, flame height and intensity of a fire and reduce the potential for spotting. These effects may persist for a considerable time (up to 20 years) in forests containing rough-barked trees and shrubby understoreys. There was uncertainty about predicting the behaviour of wildfires in different fuel types. Currently the single factor used to describe forest fuels is fine fuel load. The contribution of live and dead shrubs and tree bark to wildfire behaviour was unknown and separate fire spread models were proposed for fuels with different structure. Project Vesta provided new models to estimate the fuel characteristics of different fuel types and identified better fuel parameters to predict the behaviour of fires in dry eucalypt forest.

2 Aim The four main scientific aims of Project Vesta were: 1. To quantify the changes in the behaviour of fires in dry eucalypt forests as fuels develop with age (i.e. time since fire). 2. To characterise wind speed profiles in forests with different overstorey and understorey vegetation structures in relation to fire behaviour. 3. To develop new algorithms describing the relationship between fire spread and wind speed, and fire spread and fuel characteristics including load, structure and height. 4. To develop a National Fire Behaviour Prediction System for dry eucalypt forests. Experiments Experimental fires were conducted in four hectare plots (200 x 200 m) at two sites with different understorey fuels ranging in age from 2 to 22 years since fire in south-western Australian eucalypt forests. During the summers of 1998, 1999 and 2001, 104 experimental fires were lit under dry summer conditions of moderate to high forest fire danger. On each burn day sets of fires were burnt simultaneously in plots of each fuel age. Experimental fire behaviour ranged from slow-spreading surface fires to high-intensity fires spreading more than 1000 m h -1 with sporadic crowning activity. The intensity of the experimental fires ranged from 0 (no fire spread) to 10,570 kw m -1 with flame heights up to 25 m. Results Key findings Fuels 1. A robust and practical hazard scoring system has been developed that numerically characterises the identifiable fuel strata in the forest which comprise: surface fuel of leaves, twigs and bark; near-surface fuel of grasses, low shrubs and suspended dead fine fuel; elevated fuel contributed by tall shrubs; and bark on intermediate and overstorey trees. 2. Hazard scores provide an effective way to quantify changes in fuel structure as fuels develop with time since burning. All fuel characteristics increased with time since burning. There was no evidence of any characteristic declining within 22 years of burning although the percent cover and the height of short shrubs appeared to reach an equilibrium level. 3. Numerical values of fuel structure correlate with fire spread and flame height and the hazard scoring system can be used to provide inputs for predicting fire behaviour and suppression difficulty. This represents a significant advance from previous fire behaviour models based solely on surface fuel load.

3 4. Summer fires of moderate intensity ( kw m -1 ) consume substantial quantities of bark (5 8 t ha -1 ) in long-unburned forest (20+ years) which adds substantially to the overall fire intensity, flame height, spotting potential and difficulty of suppression. Prescribed burning at very low intensity can allow bark to continue to accumulate on the upper stems of rough-barked trees. Key findings - Wind 1. The difference between the wind driving the fire front and the wind measured in the forest is a significant source of error for empirical fire behaviour modelling. 2. This error was minimised for the experiments by using 4 anemometers at 5 m height, each spaced 40 m apart and averaging both wind and fire spread over the longest interval possible while retaining uniform topography. 3. The minimum error in estimating wind at the fire front, and hence error in predicting the rate of spread due to wind measurement alone, was ±15%. 4. Wind gusts in the forest do not persist. Wind felt at one location in the forest is different to wind at another location as nearby as 40 m away, thus fire behaviour can vary widely at different locations. 5. Gustiness can raise the wind speed above a critical threshold and cause the rate of spread to increase dramatically. Key findings - Fire behaviour 1. A fire burning away from an established line of fire, such as when a wind change turns a flankfire into a headfire, will immediately burn at its potential rate of spread. However, flame height will increase progressively as elevated fuel is consumed. 2. Rate of spread is directly related to wind speed measured at 5 m in the forest above a threshold wind speed of about 5 km h Rate of spread is directly related to characteristics of the surface fuel bed and understorey layers but is only weakly related to fuel load alone. The near-surface fuel is the principal layer responsible for determining rate of spread. The near-surface layer provides a common fuel descriptor for a wide range of dry eucalypt forest types that are visually very different because of the characteristics of the understorey shrubs. 4. A new model has been developed to predict fire spread from fine fuel moisture, wind speed, surface fuel hazard score and a combined variable of near-surface fuel hazard and height. 5. The new fire spread model predicts the potential rate of spread of an established line of fire. Fires will burn at below their potential rate of spread during initial stages of development until the headfire is at least 100 m wide (typically 1-2 hours), and if the width of the headfire is constrained. Existing guides remain valid for predicting the behaviour of prescribed burns lit from point ignition sources under mild burning conditions.

4 Project Vesta Fire in Dry Eucalypt Forest: fuel structure, fuel dynamics and fire behaviour. 6. A model to predict flame height from rate of spread and elevated fuel height has been developed to better describe suppression difficulty and to facilitate the prediction of maximum spotting distance. Tall bushfire flames can reach a maximum temperature of 1000 C to 1100 C at the base of the flaming zone. 7. The experiments confirmed that firebrand generation and spotting behaviour are intimately linked to the behaviour of the convection column, and hence fire behaviour. The fire spotting model of Ellis (2000) provides a satisfactory basis to predict maximum spotting distance of fibrous firebrands using flame height and wind speed above the forest canopy. A separate model is required to predict the maximum possible spotting of candle bark firebrands. 8. There will be occasions when conditions exist that produce abnormally severe fire behaviour. Further research is needed to understand fire-atmospheric interactions leading to abnormal fire behaviour, including conditions immediately following a dry cool change. Key findings - Effectiveness of prescribed burning 1. Hazard reduction by prescribed burning will reduce the rate of spread, flame height and intensity of a fire, as well as the number and distance of spotfires by changing the structure of the fuel bed and reducing the total fuel load. 2. The persistence of this effect will be determined by the rate of change in fuel characteristics over time. Even when the surface fuel and understorey layers have stabilised, the hazard score rating of fibrous-barked trees will continue to increase and will increase the difficulty of suppression. 3. Stimulation of understorey shrub regeneration after burning will not increase the rate of spread of a fire until such time as a significant near-surface fuel layer accumulates. 4. Younger fuels produce fewer firebrands because fire intensities are low and less bark is consumed than in older fuel types. Using prescribed burning to lower bark hazard from a score of 3 to 2 reduces the density of firebrands by threefold. Outcomes 1. Development of a fuel hazard assessment guide that provides a quantitative description of fuel hazard and its effect on fire behaviour. This guide integrates Project Vesta research findings with the Victorian Overall Fuel Hazard Guide and is applicable to dry eucalypt forests throughout southern Australia. 2. Development of the fire spread component for a national fire spread prediction system for dry eucalypt forests. This model can be presented as equations, tables or nomogram (see over). The model requires the input of surface fuel moisture which can be obtained by direct measurement or predicted from weather variables using existing systems developed for specific fuel types. 3. The framework for a model to predict spotting distance in stringybark forest using rate of spread, bark hazard score and wind speed. 4. Important fire safety information for firefighters engaged in initial attack has been made available in training guidelines and videos The Dead Man Zone and Fire Behaviour in Dry Eucalypt Forest Fuels.

5 National Fire Spread Model for Open Dry Eucalypt Forest Near-surface fuel height (cm) (1) Near-surface fuel hazard score (2) m wind speed in the open (km h -1 ) (4) Surface fuel hazard scores (3) Dead fuel moisture (%) (5) Rate of spread on level ground (m h -1 ) (6) Slope ( o ) Forward rate of spread (m h -1 )

6 How to use the National Fire Spread Model The rate of forward spread is estimated by tracing a line through the graphs to connect the values of the variables of near-surface fuel height, near-surface fuel hazard score, surface fuel hazard score, wind speed and dead fuel moisture content. The example at left demonstrates the procedure for calculating the rate of spread under the following conditions: near-surface fuel height 20 cm near-surface fuel hazard score 3 surface fuel hazard score 3.5 average wind speed at 10 m in the open 20 km h -1 dead fine surface fuel moisture content 6% slope 15 o Apply these values to the fire spread model as follows: 1. Starting in graph (1), trace a vertical line from 20 cm near-surface fuel height to the diagonal line. 2. At the point of intercept, trace a horizontal line to the near-surface fuel score graph (2) to intercept the appropriate line for the near-surface fuel hazard score From this point of intercept in graph (2), trace a vertical line downward into graph (3) until it intercepts the appropriate surface fuel hazard score From this intercept trace a horizontal line into graph (4) until it intercepts the appropriate 10 m wind speed of 20 km h From this intercept trace a vertical line into graph (5) until it intercepts the appropriate fine dead fuel moisture content 6%. 6. From this intercept trace a horizontal line to the right hand axis of graph (5) to read the rate of spread on level ground value of approximately 1050 m h -1 (between 1000 and 1100 m h -1 ). From this intercept continue to trace the horizontal line into graph (6) until it intercept the appropriate slope (15 o ). From the slope intercept line trace a vertical line to the bottom axis to read the rate of spread value of approximately 3000 m h -1. Ensis/T6-FBP-BushfireResearch/Aug07 Project Vesta Fire in Dry Eucalypt Forest: fuel structure, fuel dynamics and fire behaviour Contacts Jim Gould Ensis Bushfire Research Ensis Forest Biosecurity and Protection, CSIRO P.O. Box E4008 Kingston ACT Lachie McCaw Department of Environment and Conservation Locked Bag 2 Manjimup WA This document summarises the full report arising from Project Vesta, October The report was compiled by Jim Gould, Lachie McCaw, Phil Cheney, Peter Ellis, Ian Knight, and Andrew Sullivan. The authors acknowledge the generous support given by a wide range of agencies, including the Australasian Fire Authorities Council (AFAC), Hermon Slade Foundation, Forest and Wood Products Research and Development Corporation, Bushfire CRC, other research providers, local government authorities and corporate sponsors (Isuzu Trucks and the Insurance Council of Australia).