Content uploaded by S.W. Taylor
Author content
All content in this area was uploaded by S.W. Taylor on Apr 06, 2016
Content may be subject to copyright.
Flame temperature and residence time of fires
in dry eucalypt forest
B. Mike Wotton
A
,
F
, James S. Gould
B
,
E
, W. Lachlan McCaw
C
,
E
,
N. Phillip Cheney
B
and Stephen W. Taylor
D
A
Natural Resources Canada, Canadian Forest Service, Great Lakes Forestry Centre,
1219 Queen Street East, Sault Ste Marie, ON, P6A 2E5, Canada.
B
CSIRO Ecosystem Sciences and Climate Adaptation Flagship, GPO Box 284, Canberra,
ACT 2601, Australia.
C
Department of Environment and Conservation, Locked Bag 2, Manjimup, WA 6258, Australia.
D
Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre, 506 West Burnside
Road, Victoria, BC, V8Z 1M5, Canada.
E
Bushfire Cooperative Research Centre, Level 5, 340 Albert Street, East Melbourne,
VIC 3002, Australia.
F
Corresponding author. Email: mike.wotton@nrcan.gc.ca
Abstract. Temperature profiles of flames were measured using arrays of thermocouples on towers located in
experimental bushfires of varying intensity, carried out in dry eucalypt forest of different fuel age and structure. In-fire
video of flame-front passage and time series data from very fine exposed thermocouples were used to estimate the duration
of passage of the main flaming front in these experimental fires. Flame temperature measured at points within the flame
was found to vary with height; maximum flame temperature was greater in the tall shrub fuel than in the low shrub fuel
sites. A model to estimate flame temperature at any height within a flame of a specific height was developed. The
maximum flame temperature observed was ,11008C near the flame base and, when observation height was normalised by
flame height, flame temperature exponentially decreased to the visible flame tip where temperatures were ,3008C.
Maximum flame temperature was significantly correlated with rate of spread, fire intensity, flame height and surface fuel
bulk density. Average flame-front residence time for eucalypt forest fuels was 37 s and did not vary significantly with fine
fuel moisture, fuel quantity or bulk density.
Additional keywords: bushfire, Eucalyptus marginata, fire behaviour, flame measurement, fuel age, jarrah forest, Vesta.
Received 12 November 2010, accepted 26 May 2011, published online 15 December 2011
Introduction
Wildland fire is a complex phenomenon characterised by a wide
range of factors such as: wind, slope and atmospheric interac-
tions; fuel structure and properties; thermal degradation; fuel
combustion; and the physical and chemical properties of the
flames and emissions. Flames are constantly changing, com-
plex, random, pulsating, transient phenomena (Gaydon and
Wolfhard 1960; Johnson 1982; Chandler et al. 1983; Drysdale
1985). It is difficult to measure such phenomena in the labora-
tory, and more so in nature because wildland fires emit large
quantities of heat, are partly obscured by smoke and vegetation,
and are potentially dangerous to the observer (Gill and Knight
1991). However, flames are the most striking and readily
observable aspect of a wildland fire. Flame size and shape are
useful for describing the character of the fire and in predicting or
describing fire behaviour and effects (Pyne et al. 1996). Flame
height is related to rate of heat output or intensity (Byram 1959)
and therefore to the likelihood of scorching, crowning and
resistance to control. Flame height (along with convection
currents and velocity) influences the lofting of firebrands (Ellis
2000). The amount of radiant and convective heat from the
flame to its surroundings depends on the size, shape and tem-
perature of the flames.
Thermocouples have often been used to measure the
‘temperature’ of fire in wildland fuels (Walker and Stocks
1968; Packham 1970; Packham and Pompe 1971; Gill and
Knight 1991; Moore et al. 1995; Gould et al. 1997; Mangan
1997; Santoni et al. 2002; Butler et al. 2004a; Taylor et al.
2004; Santoni et al. 2006). However, temperatures from
wildfire flames measured by thermocouples are not measure-
ments of a single, well-defined quantity but an integration of
heat transfer (through radiative and convective processes)
CSIRO PUBLISHING
International Journal of Wildland Fire 2012, 21, 270–281
http://dx.doi.org/10.1071/WF10127
Journal compilation Ó IAWF 2012 www.publish.csiro.au/journals/ijwf
from a mixture of gases, liquids and solids with a range of
emissivities. Observed flame temperature can also depend on
the physical characteristics (e.g. thermal mass and cross-
sectional area) of the thermocouple used. Therefore, extra-
polation of a flame’s true radiative temperature is not
straightforward (Van Wagner 1967, 1970). Flame tempera-
tures vary widely and fluctuate rapidly (Philpot 1965;
Walker and Stocks 1968; Packham 1970; Packham and
Pompe 1971; Drysdale 1985; Gill and Knight 1991; Moore
et al. 1995; Gould et al. 1997; Mangan 1997; Wotton and
Martin 1998; Wotton et al. 1998; Knight et al. 2002; Santoni
et al. 2002; Butler et al. 2004a; Taylor et al. 2004; Santoni
et al. 2006). Errors in flame temperature measurement can
result in large error in estimated radiation output owing to
radiation’s relation to the fourth power of temperature
(Packham and Pompe 1971; Vines 1981; Chandler et al. 1983).
Wotton and Martin (1998) found in laboratory experimen-
tation in pine needle beds that free-burning flames, ranging
from 0.5 to 2.0 m in height, had a strong vertical variation in
flame temperature, with maximum temperature occurring at
the base of the flame. In that study, temperature in the flame
decreased with increasing height to a minimum of ,2008Cat
the (visually observed) flame tip. Marcelli et al. (2004)
also found similar decrease in measured temperature with
height in small flames (30 cm) burning across a fuel bed in a
laboratory. They found that temperatures near the observed
tip of the flames were ,300 to 4008C (from fig. 2 in Marcelli
et al. 2004). However, the properties of turbulent flames
spreading in natural forest stands may differ from flame
properties determined under still air conditions in the
laboratory.
The radiative qualities of wildfire flames (emission type,
temperature and emissivity) and the characteristics of the flame
front (flame shape, height and width) are key components in
modelling heat transfer in wildland fires (Sullivan et al. 2003).
Sullivan et al. (2003) noted a wide range of applications of heat
transfer models including: determining firefighter safety zones
(Butler and Cohen 1998; Sullivan et al. 2002; Putnam and Butler
2004; Zarate et al. 2008), devising guidelines for wildland–
urban interface buildings (Maughan et al. 1999; Ellis 2000;
Douglas and Ellis 2001; Cohen 2004) and developing physically
based fire propagation models (Van Wagner 1967; Frandsen
1971; Albini 1985; Santoni and Balbi 1998; Linn et al. 2002;
Butler et al. 2004b; Morandini et al. 2005). Thermal dynamics
of wildland fire flames are an important component of heat
transfer modelling. Flame temperatures are difficult to evaluate
from experimental fires in the field because the flames of
wildland fires are highly heterogeneous and turbulent (Butler
et al. 2004a; Taylor et al. 2004; Santoni et al. 2006), but field
observations are important for understanding the resultant
thermal dynamics during flame propagation over a range of
burning conditions.
In this paper, we characterise the time–temperature struc-
ture of flames in dry eucalypt forests having different fuel
ages, and examine the vertical distribution of temperature within
the flaming zone. The influence of fuel-bed characteristics
and resultant fire behaviour on maximum flame tempera-
ture, flame residence time and time–temperature exposure are
also examined.
Experimental methods and instrumentations (apparatus)
Site description
During the summers (January–March) of 1998, 1999 and 2001,
experimental fires were ignited at two sites in the jarrah
(Eucalyptus marginata) forest of south-west Western Australia
(Fig. 1) as part of Project Vesta (Cheney et al. 1998; Gould et al.
2007). The McCorkhill site (referred to hereafter as the TS site)
understorey was dominated by 2 m-tall shrubs (Taxandria par-
viceps; after 8 or more years unburnt, this understorey provides a
ladder fuel, increasing flame length). The Dee Vee Road site
(the LS site) had a sparse understorey of low shrubs (most shorter
than 0.5 m high) dominated by Bossiaea ornata. At each site,
several replicate plots were established in areas of forest unburnt
for 2–22 years. Plots were 200 200 m (4 ha) and were sepa-
rated by 3 m-wide bare-earth tracks.
Fuel description
Fuel and vegetation characteristics were assessed at up to 32
systematic sample points in each plot using a ranked sampling
approach as detailed in Gould et al. (2007, 2011). Five fuel strata
were recognised:
open overstorey canopy trees ,30 m top height with bark
fuel,
intermediate canopy trees up to 15 m high and associated
bark,
elevated fuel layer of upright shrubs and other understorey
plants without significant suspended material up to a height
of 3 m,
a near-surface fuel layer of twigs, bark, suspended leaves and
low shrubs up to 0.5 m high,
surface litter and duff layer with depth up to 50 mm.
Fuel characteristics assessed at each point included depth,
cover and loading of surface litter and near-surface fuel, and
height of elevated shrub fuel. Samples of surface litter and near-
surface fuel were harvested and oven-dried to determine load-
ings in tonnes per hectare. Sampling intensity was designed to
estimate the mean surface fuel loading with a standard error of
less than 15%. Elevated shrub fuels were also harvested from a
small number of quadrats of each fuel age to determine loadings
of live and dead material ,6 mm in diameter.
Fire behaviour experiments
Experimental fires were conducted under dry summer condi-
tions of moderate to high forest fire danger (McArthur 1967). On
each burn day, four or five fires were ignited simultaneously in
plots of each fuel age (five ages at TS, four ages at LS). Ignition
lines 120 m long near the upwind edge of each plot were lit with
drip torches, working outwards from the centre point of the
ignition line to complete the lighting operation in 2 min. Fire
spread was measured by electronic timers that recorded the
burn-through of thin nylon lines strung across the plot from a
centre-post. At 2-min intervals, experienced fire behaviour
observers recorded ocular estimates of fuel continuity, fuel type,
flame height, flame depth and flame angle, and placed metal tags
at the head of the fire, which were located and mapped post fire
and used to reconstruct fire spread across the plot. Additional
notes were made on up-draughts, down-draughts and smoke
Flame temperature and residence time in eucalypt forest fires Int. J. Wildland Fire 271
colour, and photographs were taken of characteristic fire
behaviour (Gould et al. 2007).
Flame temperature measurement
Flame temperatures were measured using chromel–alumel
thermocouple junctions (36-gauge or 0.127-mm diameter wire)
attached to 3 m-tall towers at 0.5, 1.0, 2.0 and 3.0 m above
ground surface. The fast response time (,0.4 s as reported in
Butler et al. 2004a), small thermal mass and cross-sectional area
of these thermocouple beads (,0.25-mm diameter) minimise
the effects of radiation and heat retention, providing reasonably
accurate measurements of gas temperature within the flaming
zone and flame zone depth. Butler et al. (2004a) describe
potential error in measurement of wildfire flame temperature by
exposed thermocouple junctions of this size and type and con-
clude that the upper limit of error is likely no more than 508C.
We feel variability in measurements of maximum temperature
from the turbulent flames in this experiment was much larger
than the radiation correction needed. Temperatures were
recorded at 1 Hz.
Temperature towers were located in the $5- and $7-year-
old fuel in the TS and LS sites. In 1998 at the LS site, only three
temperature towers were available, and all were arranged in a
single plot on each burn day. In 1999, six additional 3 m-tall
towers were deployed, permitting simultaneous flame tempera-
ture measurement in three plots of different fuel age on each
burn day. One tower was placed near the centre line of the fire
plot and the other two were placed ,50 m left and right of the
centre. All towers were placed 100–150 m from the ignition line
depending on the expected fire behaviour in a particular fuel
complex; the objective was to achieve an active spreading
headfire through the tower location.
At the TS site, where the dense shrub layer was expected to
produce higher flames, two higher thermocouple arrays supple-
mented observations from the nine 3 m-tall towers. These
consisted of a series of thermocouples attached to a thin steel
cable at 2-m intervals (to a maximum of 12 m) strung over a
horizontal tree limb near the centre of the 10- and 16-year-old
plots. Because limb heights varied and the array was difficult to
position exactly, the actual position of each thermocouple was
calculated using the distance of the lowest thermocouple to the
ground once the array was fixed in place.
In addition to the 2-min fire behaviour observations, the
approach and passage of the fire through tower locations
was recorded on hand-held video, and visual observations of
flame height and fire behaviour were recorded when possible.
To Pe r th
Pinjarra
Dwellingup
Boddington
Williams
Hamil
Harvey
40 km
Collie
Bunbury
Ludlow
Busselton
Kirup
Grimwade
Lewana
20 km
Nannup
20 km
Margaret River
Manjimup
McCorkhill site
Indian Ocean
Dee Vee site
Fig. 1. Location of the two study sites in south-west forest regions of Western Australia. McCorkhill was the tall
shrub (TS) site whereas Dee Vee was the low shrub (LS) site.
272 Int. J. Wildland Fire B. M. Wotton et al.
Experimental fires conducted after 14 January 1999 were also
monitored by an in-fire video camera on loan from the USDA
Forest Service Technology and Development Program at Mis-
soula, MT (Kautz 1997), which provided a visual record of the
passage of the flame past a single tower location on each fire.
The camera was placed in the centre of the plot with the oldest
fuel age to maximise the possibility of active fire spread through
the camera’s field of view. A 3-m temperature tower was
included in the field of view as a height reference for flame
structure.
Analysis
Temperature data from well-established, rapidly spreading head
or flank fires passing the tower were grouped according to fuel
age and fire day; temperature tower data from backing fires
(e.g. from spot fire growth) were excluded from analysis. We
examined the influence of fuel load, fire behaviour and fuel
moisture on three observed flame temperature characteristics for
each fuel age and burn day combination: maximum temperature
reached, duration above 3008C at each thermocouple closest to
the surface (i.e. flame-front residence time), and a time–
temperature integration of each temperature time series for the
period above 3008C. A list of variables used in this study is
given in the ‘Symbols and definitions of variables used
in analysis’ section.
For each thermocouple tower where headfire passage was
directly observed, maximum flame temperature was paired with
observed flame height to examine the relationship between
flame temperature, flame height and vertical position in the
flame.
Residence time of flaming combustion was defined as the
period where temperatures exceeded 3008C. This temperature
threshold has been found in laboratory and field-based studies
(Burrows 2001; Taylor et al. 2004) including the present one
(see later section on vertical temperature variation) to corre-
spond well with the temperature of the observed flame tip.
Positioning the thermocouples above surface fuels allowed us to
study only the passage of the main fire front; the duration of
longer-term surface fuel burnout, characterised by intermittent
patches of lower flame not attached to the main front, is not
quantifiable from these measurements but was quantifiable
through video analysis.
An independent estimate of residence time was also obtained
(as the quotient) of ocular estimates of flame front depth and
spread rate. Residence time was also estimated for each plot
where in-fire video was recorded. Video analysis estimated the
duration of the main tall, elevated flames from the active fire
front, the duration of the period of short (,30 cm high) but
continuous flaming directly following the main standing flames,
and the duration of intermittent, discontinuous flaming on the
surface at the thermocouple tower.
Results
Eleven sets of simultaneous experiments were conducted at TS
sites, and 12 sets at LS sites for a total of 98 experimental fires, of
which 44 were instrumented with thermocouple temperature
towers. The range of forward spread, flame height and fire
intensity for each age class of the instrumented fires is given in
Table 1. Fire behaviour ranged from slow-spreading surface
fires to high-intensity fires with sporadic crowning activity.
Observed flame lengths at 38 tower locations on 19 individ-
ual fires ranged from 1 to 14 m. In-fire videos captured flame
passage through thermocouple towers on seven fires, and
allowed more detailed, albeit qualitative, study of flame char-
acteristics associated with these temperature profiles. Fig. 2
shows an example of images captured from this in-fire
video. Examples of in-fire video footage from TS and LS
experimental fires are available in the Supplementary material
on the International Journal of Wildland Fire website (see
http://www.publish.csiro.au/?act=view_file&file_id=WF10127_
AC.zip).
Fig. 3 shows a typical temperature trace of a flame passing
through a 3-m temperature tower. This trace, from TS experi-
mental burn 16BN (a 16-year-old plot burned on 27 January
1999), shows temperatures within some of the larger flames
observed during the experiment. This flame had a well-
developed, relatively stable structure and an ocularly observed
height of 10 m.
Maximum flame temperature (T
fmax
)
For fires where in-fire temperatures were measured, maximum
temperatures (T
fmax
) observed were similar at the LS and TS sites
(1184 and 10988C) and were typically recorded on flames several
metres in height. To explore the factors potentially influencing
T
fmax
, we examined the maximum temperature data observed at
the lowest (0.5 m) thermocouple on all towers with a temperature
trace indicating clear passage of a flaming front through the tower
(e.g. Fig. 3). This analysis was carried out for a series of plot
average values, to account for high within-plot variability, and
for observations at individual tower locations. We estimated
Pearson correlation coefficients (Table 2) (and carried out formal
linear regressions) for mean maximum temperature and plot
average values of: forward spread rate (R), fire intensity (I
B
),
flame length (L
f
), under-forest canopy wind speed at 5 m (U
5
),
height of the elevated fuel (shrub) layer (H
e
), surface fuel bulk
density (BD
s
), near-surface fuel bulk density (BD
ns
), total fine
fuel load in surface and near-surface fuels (L
sns
), surface fuel
moisture content (M
fs
, the upper 5–10 mm of undecomposed
leaves from the litter layer), near-surface fuel moisture content
(M
fns
), and surface fuel profile moisture content (M
fsp
, full depth
of the litter layer above the mineral soil including surface litter).
Maximum flame temperature was significantly correlated
(P , 0.01) with R, I
B
, L
f
, and BD
s
and showed marginally
significant correlations (P values ,0.05) with U
5
and I
B
.
Maximum temperature data at individual tower locations
were only analysed for fire behaviour characteristics that could
be estimated at individual tower locations (by personnel who
documented fire behaviour ahead of the moving fireline): local
R, local L
f
and flame depth (D
f
). Results of individual
correlation tests (Table 3) were similar to the plot average
analysis for R, L
f
, H
f
and D
f
.
Flame residence time (RT
f
)
Overall mean residence time (RT
f
) for the passage of the main
flaming front for all fires with thermocouple measurements was
Flame temperature and residence time in eucalypt forest fires Int. J. Wildland Fire 273
37 s (n ¼ 77, s.d. 14). Mean overall RT
f
for the LS and TS sites,
36 s (s.d. 12) and 37 s (s.d. 16), were not significantly different
(n ¼ 77, F ¼ 0.27, P ¼ 0.60). We did not find a statistically
significant effect of fuel age on residence time at either site.
Mean RT
f
for each fuel age at the two sites is presented in
Table 4.
The same analysis conducted on the T
fmax
data was done for
RT
f
(Table 2), and only R had a (marginally) significant
relationship, with a P value of ,0.06. Correlation results at
individual tower locations (Table 3) showed the significance of
the relationship between R (at the tower location) and RT
f
disappeared; thus, the significance of the marginal relationship
in the plot average analysis is doubtful. The results are relatively
insensitive to 3008C as the lower threshold for residence time;
similar statistical comparisons were done at 200 and 4008C.
Estimated RT
f
based on observed D
f
/R from all experimental
fires given in Gould et al. (2007) was 37 s (n ¼ 98, s.d. 24).
Residence time in the LS and TS sites in the younger fuel age
classes (,3 years old) that were not instrumented with thermo-
couple towers were 28 s (n ¼ 8, s.d. 18) and 38 s (n ¼ 10, s.d. 18).
Correlation tests showed no significant effects of L
s
, L
ns
, BD
s
and BD
ns
on the observed RT
f
based on D
f
/R.
In-fire video of seven fires in 16-year-old fuel at the TS site
showed that the prevailing wind was entrained into the headfire
convection from behind the tall flames. Spot fires ahead of the
main fire developed in a circular pattern, indicating that the wind
in this area was often light and variable in direction. Generally,
the spots were drawn against the prevailing wind towards the
head of the fire in only the last few seconds before the front
arrived. The tall flames around the camera lasted from 9 to 13 s if
Table 1. Range of observed fire behaviour (minimum and maximum) during the instrumented experimental fires by site and fuel age based on mean
values per plot
Observed variable Tall shrub (TS) site fuel ages (years since fire)
5 7 10 16
Number of plots instrumented 7 8 5 10
U
5
( ms
1
) 1.1–1.6 1.4–1.6 1.1–2.0 1.0–2.4
M
fs
(%) 5.6–9.6 5.6–9.6 5.6–9.6 5.6–9.6
M
fns
(%) 6.6–10.5 6.6–10.5 6.6–10.5 6.6–10.5
M
fnp
(%) 6.1–9.5 5.5–9.5 6.1–9.5 5.5–9.5
L
s
(t ha
1
) 5.6–7.9 5.9–11.6 7.7–11.5 7.7–16.5
L
sns
(t ha
1
) 8.6–10.6 8.9–14.3 11.5–17.4 12.0–22.1
BD
s
(kg m
3
) 33.6–46.6 31.1–47.6 39.3–52.6 27.0–48.1
BD
ns
(kg m
3
) 4.7–6.2 4.6–8.0 5.2–8.4 4.7–8.7
H
e
(cm) 92–118 109–134 108–165 96–192
R (m h
1
) 189–673 211–947 420–1240 295–1160
H
f
(m) 1.2–7.5 1.6–6.7 2.5–9.1 2.3–14.2
D
f
(m) 1.7–5.0 1.9–14.5 4.5–5.0 2.2–4.0
A
f
(8) 55–100 57–95 75–87 69–91
L
f
(m) 1.2–9.1 1.7–6.8 2.6–9.2 2.3–14.4
I
B
(kW m
1
) 710–3040 1120–4120 2570–10 570 2320–8320
T
fmax
(8C) 676–1066 746–1039 722–944 709–1098
RT
f
(s) 23–32 25–60 21–44 20–64
Low shrub (LS) site fuel ages (years since fire)
78 91920
Number of plots instrumented 3 2 3 1 3
U
5
(m s
1
) 1.1–1.4 1.6–1.7 1.2–1.6 1.7 1.1–1.2
M
fs
(%) 6.1–7.1 6.6–6.9 6.1–6.8 6.5 6.1–7.1
M
fns
(%) 7.1–8.8 7.4–8.2 7.1–8.8 7.9 7.1–8.8
M
fnp
(%) 7.8–8.8 7.1–7.5 7.1–8.8 7.9 7.1–8.8
L
s
(t ha
1
) 7.7–13.6 6.1–7.3 7.3–8.8 11.6 10.2–17.9
L
sns
(t ha
1
) 11.5–13.6 9.1–9.7 9.2–17.5 12.5 13.2–19.9
BD
s
(kg m
3
) 44.1–54.2 24.7–35.5 38.1–50.4 41.1 38.6–80.7
BD
ns
(kg m
3
) 4.3–8.5 7. 1–9.1 5.1–9.9 6.3 8.0–11.6
H
e
(cm) 72–84 51–76 38–55 50 35–38
R (m h
1
) 410–490 546–566 160–480 490 180–304
H
f
(m) 1.4–3.2 3.2–4.0 1.1–1.9 3.7 1.1–3.0
D
f
(m) 2.1–8.0 7.0–9.7 1.0–2.3 13.4 2.1–3.0
A
f
(8) 73–88 69–100 84–104 75 84–95
L
f
(m) 1.4–3.3 3.3–4.3 1.1–1.8 3.8 1.1–3.0
I
B
(kW m
1
) 2220–2945 2510–2580 715–4200 3145 1345–2220
T
fmax
(8C) 920–1000 1018–1184 871–925 1046 679–941
RT
f
(s) 30–49 32–39 25–47 39 24–42
274 Int. J. Wildland Fire B. M. Wotton et al.
the fire passed uniformly through the camera site. If there were
violent swirls and flames from fuels not directly around the
camera, then the tall flames appeared to last up to 27 s (Table 5).
Behind the tall flames of the headfire, flames associated with
residual burning were low and depressed by downdraft wind
feeding into the headfire (Fig. 2). Continuous short flames
persisted for 37 to 74 s. Over the next 75 to 127 s, areas of
smouldering combustion increased and flames were increas-
ingly restricted to smaller areas of heavy fuels, although strong
gusts of wind over smouldering fuels could ignite them, indicat-
ing there was still substantial surface fuel available.
Time–temperature integration
Considering the duration of exposure above a specified thresh-
old temperature is important for the design of radiation safety
zones (Butler and Cohen 1998) and in understanding the effect
of a fire on vegetation survival. Each temperature trace at 0.5 m
was integrated over time when that trace exceeded 3008C (the
period of the passage of the main flame front). This results in a
time–temperature integration value (s 8C) for each tower in
each plot. A correlation table was used to examine the influence
of the fire behaviour factors (Table 2) on this time–temperature
integration, and only R was found to be statistically significant;
there was a marginally significant association between time–
temperature and both L
f
and H
f
(P ¼ ,0.07 and 0.08). These
relationships were also statistically significant when we exam-
ined paired data from individual tower locations and localised
observations of fire behaviour characteristics (Table 3).
Vertical flame temperature variation
Fig. 4 shows raw data for the relationship between flame tem-
perature and distance of the thermocouple measurement from
the flame tip. Despite the difference in fuel structure between the
Fig. 2. Flames burning in the surface fuel 22 s after the leading edge of the
fire in plot 16BN at the tall shrub (TS) site passed the camera (the fire is
burning directly away from the camera). The tall flames lasted 9 s, burning
out the fine fuel ,2.5 mm in diameter of the elevated and near-surface
layers. The downdraft behind the flames has flattened the flames of the fire
burning out the surface fuel layer.
0
200
400
600
800
1000
1200
15:23:00 15:23:30 15:24:00 15:24:30 15:25:00 15:25:30 15:26:00 15:26:30
Temperature (⬚C)
Time
0.5 m
1.0 m
2.0 m
Fig. 3. An example of a typical temperature trace of a flame passing through a 3-m thermocouple tower from an
experimental fire in the 16-year-old fuel at the tall shrub (TS) site (16BN). Ignition time was 1515 hours and tower
was located 142 m from ignition line. Flame height observed at this tower location was 10 m.
Flame temperature and residence time in eucalypt forest fires Int. J. Wildland Fire 275
two sites, this relationship differs little between observations
from the TS and LS sites.
Researchers who have examined the empirical distribution of
temperature above a buoyant plume in both laboratory
(e.g. McCaffrey 1979; Dupuy et al. 2003) and field (Mercer
and Weber 2001) settings describe three regimes for change in
temperature in buoyant plumes with height: a constant tempera-
ture range (nearer the base of the flame), a transitional zone
(beginning within the visible flame) and a plume region where
conventional plume theory holds. McCaffrey (1979) suggested
that in the intermittent zone, temperature decrease with height
scales with the first power of height, whereas in the plume
region, it scales as the 5/3rd power of height. Dupuy et al. (2003)
further showed that, in vertical flames, if temperature measure-
ment height is normalised using H
f
, a consistent scaled relation-
ship can be found over a range of H
f
.
Table 2. Correlations between observed plot average fuel bed and fire behaviour characteristics and measured flame temperature characteristics
(n 5 39 for all tests)
Plot average fuel or fire
behaviour characteristic
Resultant flame characteristics
Maximum temperature Flame residence time Time–temperature
R (P) R (P) R (P)
U
5
0.31 (0.052) 0.11 (0.519) 0.09 (0.936)
R 0.45 (0.004) 0.30 (0.059) 0.32 (0.048)
I
B
0.32 (0.050) 0.27 (0.100) 0.24 (0.133)
L
f
0.40 (0.012) 0.24 (0.138) 0.31 (0.052)
M
fs
0.05 (0.773) 0.07 (0.652) 0.06 (0.722)
M
fns
0.02 (0.918) 0.07 (0.675) 0.04 (0.813)
M
fnp
0.10 (0.540) 0.11 (0.500) 0.17 (0.294)
BD
s
0.47 (0.003) 0.23 (0.160) 0.30 (0.062)
BD
ns
0.03 (0.854) 0.04 (0.797) 0.02 (0.889)
L
sns
0.10 (0.546) 0.07 (0.687) 0.04 (0.786)
H
e
0.16 (0.328) 0.20 (0.209) 0.20 (0.602)
Table 3. Correlations between flame characteristics from individual tower observations and local fire behaviour observations from the
location of the tower
Local fire behaviour characteristic
(at tower location)
Resultant flame characteristics
Maximum temperature Flame residence time (.3008C) Time–temperature
nR(P) nR(P) nR(P)
R 82 0.37 (0.001) 72 0.13 (0.285) 82 0.23 (0.045)
L
f
82 0.27 (0.014) 72 0.11 (0.0352) 82 0.20 (0.067)
H
f
82 0.26 (0.019) 72 0.12 (0.330) 82 0.20 (0.076)
D
f
41 0.33 (0.036) 35 0.03 (0.846) 41 0.22 (0.158)
Table 4. Flame-front residence time for each experimental site and
fuel age since last fire
Tall shrub (TS) Low shrub (LS)
Residence time (s) Residence time (s)
Age n mean s.d. Age n mean s.d.
51029 107 8 35 13
7 14 44 20 8–9 12 36 14
10 4 30 11 19–20 11 35 8
16 18 39 16
Table 5. Duration of headfire combustion at a particular point,
timed from in-fire video observations, tall shrub (TS) block (source
Gould et al. 2007)
Block Date Duration of
tall flames (s)
Duration of
continuous
short flames (s)
Duration of
intermittent
flaming (s)
16BN 22-Jan-99 11 58 111
16CS 29-Jan-99 9 42 100
16EN 31-Jan-99 12 37 97
16DN 5-Feb-99 12 (27
A
) 58 126
16IS 7-Feb-99 13 37 75
16FS 8-Feb-99 22
A
61 127
16GS 9-Feb-99 14 74 122
A
Denotes the total time that the camera was within flames originating from
two different flame locations.
276 Int. J. Wildland Fire B. M. Wotton et al.
We plotted maximum flame temperature against normalised
flame height (height of the thermocouple, H
t
, divided by flame
height, H
f
) in log-log space (not shown) and observed a very
similar relationship to that observed in earlier studies. The
relationship seemed to separate into the three regions described
by McCaffrey (1979) with a linear relationship (in log-log
space) between flame temperature and normalised height for
heights above ,0.5 H
f
. However, when we plotted T
fmax
as a
function of log of normalised height alone, the relationship
appeared quite linear (Fig. 5), suggesting that a single exponen-
tial model adequately described flame temperature variation.
A linear regression run on the log-transformed data gave a
significant relationship with R
2
¼ 79% (n ¼ 173, F ¼ 634,
P , 0.0001). The equation for this flame temperature,
T
f
(H
t
, H
f
), as a function of normalised flame height (H
t
/H
f
)
curve is then
T
f
ðH
t
; H
f
Þ¼a þ b lnðH
t
=H
f
Þð1Þ
where a ¼ 334 (s.e. 10) and b ¼258 (s.e. 10). It seems doubtful
this relationship should hold above the tip of the flame itself;
however, the model fit showed no strong deviation bias for
distances up to a full flame length above the tip (Fig. 5).
Discussion
In the vertical plane, flames in freely burning wildland fires are
typical turbulent diffusion flames with two visually identifiable
regions. First, a region of continuous flames directly connected
to the burning fuel can be visually characterised by height or
length, depth and angle. Above this region are intermittent flame
flashes; detached envelopes of burning gas separated from the
continuous flame. Understanding temperature in these regions
and above the visible flame is an important part of understanding
heat transfer mechanisms in and above wildfires. Along the
horizontal plane, one might also separate the flaming zone into
two main regions: the first is characterised by the main flame
volume (the distinct and connected upright flame) and the sec-
ond is an area of smaller flaming that persists, becoming more
discontinuous as lighter fuels are consumed but heavier or more
compact fuels continue to burn.
Rothermel and Deeming (1980) suggested that flame resi-
dence is equal to the time from initial temperature rise to the time
of ‘definite drop’ after reaching peak temperature (Fig. 6);
however, residence times, defined by these two points in the
time–temperature history of a point, can be quite short and,
based on our observations, may underestimate the duration that a
point is within the main flaming front. Cheney (1981) defines
residence time simply as the period of flaming combustion.
Visual estimates of flame-front residence times from in-fire
video of Project Vesta experimental fires (Gould et al. 2007)
were much longer than suggested by the definition of Rothermel
and Deeming (Table 5). In the present study, we characterised
flame residence time using thermocouples elevated above the
top of the fuel bed; thus for the purposes of this study, and to be
consistent with other observations of flaming residence time in
wildland fires in various fuel complexes with similar equipment,
we define flame-front residence time at any point as the length of
time that a point is within the main flame envelope. In this study,
because the thermocouples are elevated as much as 0.5 m above
the surface, the period of extended burnout of heavier or more
compacted fuels characterised by small flames is not repre-
sented in the estimates of residence time.
Flaming residence times were on average ,37 s for both the
TS and LS sites. Despite the differences in fuel load and structure,
no significant differences in residence time were found between
sites or among plot ages (time since fire). This residence time
average for dry eucalypt sites is close to the value of ,33 s found
in the laboratory by Burrows (2001), who burned piles of jarrah
leaves equivalent to ,10 t ha
1
fuel load (similar to the surface
loads found in the older age classes of the present study).
0
200
400
600
800
1000
1200
⫺5 ⫺2.5 0 2.5 5 7.5 10
Flame temperature (⬚C)
Distance of temperature measurement from the flame tip (m)
Low shrub
Tall shrub
Fig. 4. Maximum flame temperature as a function of distance down from
the flame tip (,0 m above flame tip).
0
200
400
600
800
1000
1200
0.01 0.1 1 10
Flame temperature (⬚C)
Normalised flame hei
g
ht (m)
Low shrub
Tall shrub
Fig. 5. Maximum flame temperature observed as a function of log normal-
ised height within the flame (thermocouple location has been normalised by
dividing by the flame height). The straight line represents the simple least-
squares fit to the log-transformed data (R
2
¼ 0.79, P , 0.0001).
Flame temperature and residence time in eucalypt forest fires Int. J. Wildland Fire 277
Particle flaming time is a function of particle diameter
(Clements and Alkidas 1973; Cheney et al. 1990; Burrows
2001). Residence time has been described and modelled (Ander-
son 1969; Burrows 2001) solely as a function of the dimensional
characteristics of the fuel being burned (i.e. diameter); thus a
particular fuel bed could be thought of as having a characteristic
residence time (Burrows 2001) depending on the average
diameter of the fuel particles making up the bed. However,
within a fuel bed, the residence time also includes the time to
burn horizontally along the particles and down into the fuel bed.
In these studies, the surface litter fuel bed made up ,80% of the
total fine fuel load. This fuel bed was stratified with a loosely
compacted layer of fresh litter material above a denser layer of
decomposing material. Continuous flames of the fire front burn
this loosely compacted material, and further combustion down
into the fuel bed is limited by ash deposition and compaction of
the lower layers, which burn by intermittent flaming or smoul-
dering combustion. Gould et al. (2007) found that the bulk
density at both sites remained more or less constant after 5 years.
The constant residence time suggests that the characteristic of
the loosely compacted fresh litter fuel does not change with fuel
age, thus explaining the poor correlation of RT
f
with fuel load
and bulk density.
Taylor et al. (2004) found residence times at the surface
of ,45 s for fully active crown fires (and closer to 20 s for
flaming in the canopy). Similarly, Cahoon et al. (2000) listed a
range of 30 to 60 s for boreal crown fires. Overall, our observa-
tions compare reasonably with other high-intensity fires in
forested canopies (Cahoon et al. 2000; Burrows 2001; Taylor
et al. 2004).
Flame-front residence time interpreted from thermocouple
traces showed neither strong influence of fuel bed (such as fuel
load or bulk density) nor fire behaviour characteristics (such as
rate of spread or flame length). Although we found a weak
relationship between residence time and rate of spread at the plot
level, no significant relationship was found for individual tower
observations; this relationship requires further study, perhaps in
more controlled settings.
Maximum flame temperatures observed showed sustained
values similar to those reported in other studies. In concurrent
studies on the same experimental fires in northern Canadian
boreal forest, Taylor et al. (2004) found maximum temperatures
(1-s average) of ,12008C, whereas Butler et al. (2004a)
observed maximum flame temperatures of ,13008C. Taylor’s
temperatures were similar to those reported in the present study.
Typically, maximum flame temperatures were observed at the
base of the flame that agreed with other observations from both
the laboratory (Wotton and Martin 1998) and the field (Taylor
et al. 2004). Butler et al. (2004a) observed maximum flame
temperatures in thick crown-fire flames to occur higher above
the surface (at mid-stand height) than observed in the present
study, which they speculate is due to the in-draught of cool air
through the lower part of the canopy.
Flames we observed were typically smaller, thinner and
better mixed than the deep flames in a high-intensity crown fire
in the boreal forest. As expected, correlation analysis showed a
significant dependence of maximum temperature on fire behav-
iour characteristics (rate of spread, intensity and flame length)
and surface fuel bulk density. These parameters relate directly or
indirectly to flame size and hence the relative position in the
0:10:00 0:10:15 0:10:30 0:10:45 0:11:00 0:11:15 0:11:30 0:11:45 0:12:00
Time since ignition (minutes:seconds)
0
200
400
600
800
1000
1200
Temperature (⬚C)
Flame residence time (t
r
)
Reference: Rothermel and Deeming (1980)
Observed flaming
period at 0.5 m
(a)
(b)
Fig. 6. Comparison of (a) the flame residence time defined by Rothermel and Deeming (1980); and (b) the
observed flaming period at 0.5 m in 8-year-old fuel TS experimental fire.
278 Int. J. Wildland Fire B. M. Wotton et al.
flame of the lowermost thermocouple (which was at a constant
height throughout the experiment). Similarly, analysis of the
time–temperature data showed significant relationships with
spread rate and flame length, both indicators (indirectly and
directly respectively) of flame size (spread rate is related to
flame size through fire intensity). Relationships were quite
weak; however, this is to be expected given the strong variability
in the raw flame temperature signal (Fig. 3).
Earlier research with fine-gauge thermocouples during labo-
ratory fires (Wotton and Martin 1998) suggested that the
normalised flame height approach led to significant scatter in
the vertical flame temperature relationship, but when the flame
tips were taken as the origin of measurement, a coherent model
of flame temperature variation with distance from the flame tip
could be developed for flames ranging from 0.5 to 2.0 m in
height. However, these observations were made in a controlled
no-wind and no-slope laboratory environment. Although using
flame tip as the common origin of the measuring system within
the flame does organise the data into a coherent structure that
changes systematically with distance from the tip (Fig. 5), the
use of a normalised flame height provides a simpler and easier to
use model form. Further field research should be carried out to
examine whether this normalised flame temperature model
holds consistently over a large range of flame heights.
The temperature at the flame tip was between 200 and 4008C
(Fig. 5). This is slightly higher than values found in previous
laboratory work from well-established flames spreading in still
air (Wotton and Martin 1998); however, a flame tip temperature
of 300 to 4008C is consistent with other observations (Thomas
1967; McCaffrey 1979; Marcelli et al. 2004) and is a reasonable
estimate for turbulent, wind-driven flames in a natural forest.
If a consistent temperature model can be applied to flames for
a wide range of fires in different fuel types, then the next step in
understanding flame dimensions and their effect on fire sup-
pression is to compare the structure of flames for fires spreading
in different fuel beds. Visually, at least, flame volume appears to
change much more than rate of spread for fires in fuels of
different structure and load, particularly for lower wind speeds.
Because emissivity of flames decreases rapidly with flame depth
below 3 m (Hagglund and Persson 1974; A
`
gueda et al. 2010),
then flames in fuel-reduced areas may have quite different
radiation characteristics compared with fires in long-unburnt
fuels, and influence the difficulty of suppression of fires in
different fuel ages.
Conclusions
Flame temperatures were measured using thermocouples during
Project Vesta’s experimental fires in 1998 and 1999. The
maximum temperature observed was ,11008C, which is con-
sistent with observations of temperature from large flames in
other experimental fires. The temperature of the visible flame tip
was ,3008C, which agreed with observations from other
research. Flame-front residence time (the duration of flaming at
any given point) was estimated from thermocouple traces at a
height of 0.5 m and was not found to vary significantly with plot
age (which corresponded generally to fuel load) or with exper-
imental site. Average flame-front residence time for dry euca-
lypt forest fires was 37 s.
Despite the large range in flame heights and fire intensities
observed during the experiment, a simple model of flame
temperature variation with height above the ground was found.
Using height above the fuel bed normalised by flame height,
a basic exponential structure seemed to provide a sound yet
simple model to predict temperature change with height over the
range of observed flame heights. We do not suggest such
a simple exponential structure should fit all flame types univer-
sally, and indeed the three-phase structure found in other studies
of buoyant plumes in more controlled settings (e.g. McCaffrey
1979; Dupuy et al. 2003) suggests a more complex, yet still
readily generalisable, structure.
Symbols and definitions of variables used in analysis
A
f
, flame angle (8), based on Cheney (1981) definitions
BD
ns
, near-surface fuel bulk density (kg m
3
)
BD
s
, surface fuel bulk density (kg m
3
)
D
f
, flame depth (m), based on Cheney (1981) definitions
H
e
, elevated fuel height (cm)
H
f
, flame height (m), based on Cheney (1981) definitions
H
t
, height of thermocouple (m)
I
B
, fire line intensity (kW m
1
) (Byram 1959)
L
f
, flame length (m), calculated from flame height and flame
angle
LS, low shrub
L
s
, surface fuel load ,6 mm in diameter (t ha
1
)
L
sns
, surface plus near-surface fuel load ,6 mm in diameter
(t ha
1
)
M
fns
, dead fuel moisture content of near-surface fuel ,6mmin
diameter (%)
M
fs
, dead fuel moisture content of surface fuel ,6mm in
diameter (%)
M
fsp
, profile dead fuel moisture content of surface fuel ,6mm
in diameter (%)
R, forward rate of spread (m h
1
)
RT
f
, flame residence time (s)
T
f
, flame temperature (8C)
T
fmax
, maximum flame temperature (8C)
TS, tall shrub
U
5
, 5-m wind speed under forest canopy (m s
1
)
Supplementary material
Supplementary material that presents two in-fire videos of the
fire-front passage through 3-m thermocouple towers in LS
(20-year-old plot) and TS (16-year-old plot) sites is avail-
able from the International Journal of Wildland Fire
website (see http://www.publish.csiro.au/?act=view_file&file_id=
WF10127_AC.zip).
Acknowledgement
The authors thank staff from the Department of Conservation and Land
Management (now Department of Environment and Conservation) and
CSIRO Forestry and Forest Products (now CSIRO Ecosystem Sciences) who
contributed to the design, planning and implementation of Project Vesta.
The Canadian Forest Service’s support of their scientists’ participation in
Project Vesta experiments is gratefully appreciated. Project Vesta was
sponsored by the Department of Conservation and Land Management,
Flame temperature and residence time in eucalypt forest fires Int. J. Wildland Fire 279
CSIRO Forestry and Forest Products, Australasian Fire Authorities Council,
Hermon Slade Foundation, Forest and Wood Products Research and
Development Corporation, Insurance Council of Australia and Isuzu Trucks.
We also thank reviewers who provided valuable comments on earlier ver-
sions of the manuscript. We thank Andrew Sullivan of CSIRO for the editing
production of the in-fire videos.
References
A
`
gueda A, Pastor E, Perez Y, Planas E (2010) Experimental study of the
emissivity of flames resulting from the combustion of forest fuels.
International Journal of Thermal Sciences 49(3), 543–554. doi:10.1016/
J.IJTHERMALSCI.2009.09.006
Albini FA (1985) A model for fire spread in wildland fuels by radiation.
Combustion Science and Technology 42, 229–258. doi:10.1080/
00102208508960381
Anderson HE (1969) Heat transfer and fire spread. USDA Forest Service,
Intermountain Forest and Range Experimental Station, Research Paper
INT-69. (Ogden, UT)
Burrows ND (2001) Flame residence time and rates of weight loss of
eucalypt forest fuel particles. International Journal of Wildland Fire
10, 137–143. doi:10.1071/WF01005
Butler BW, Cohen JD (1998) Firefighter safety zones: a theoretical model
based on radiative heating. International Journal of Wildland Fire 8,
73–77. doi:10.1071/WF9980073
Butler BW, Cohen J, Latham DJ, Schuette RD, Sopko P, Shannon KS,
Jimenez D, Bradshaw LS (2004a). Measurements of radiant emissive
power and temperatures in crown fires. Canadian Journal of Forest
Research 34, 1577–1587. doi:10.1139/X04-060
Butler BW, Finney MA, Andrews PL, Albini FA (2004b). A radiation-driven
model for crown fire spread. Canadian Journal of Forest Research 34,
1588–1599. doi:10.1139/X04-074
Byram GM (1959) Combustion of forest fuels. In ‘Forest Fire: Control and
Use’. (Ed. KP Davis) pp. 61–89. (McGraw-Hill: New York)
Cahoon DR, Stocks BJ, Alexander ME, Baum BA, Goldammer JG (2000)
Wildland fire detection from space: theory and application. In ‘Biomass
Burning and its Inter-relationship with the Climate System’. (Eds JL
Innes, M Beniston, MM Verstraete) pp. 151–169. (Kluwer Academic
Publishers: Dordrecht)
Chandler C, Cheney P, Thomas P, Trabaud L, Williams D (1983) ‘Fire in
Forestry, Volume I: Forest Fire Behaviour and Effects.’ (Wiley:
New York)
Cheney NP (1981) Fire behaviour. In ‘Fire and the Australia Biota’.
(Eds AM Gill, RH Groves, IR Noble) pp. 151–175. (Australian Academy
of Science: Canberra, ACT)
Cheney NP, Gould JS, Hutchings PT (1990) Sampling available fuel and
damage to trees retained after thinning and burning. Management of
eucalypt regrowth in East Gippsland. CSIRO Division of Forestry and
Forest Products and Department of Conservation and Environment.
Technical Report 12. (Melbourne)
Cheney NP, Gould JS, McCaw LW (1998) Project Vesta: research initiative
into the effects of fuel structure and fuel load on behaviour of wildfires in
dry eucalypt forest. In ‘Proceedings 13th International Conference of
Forest Fire and Meteorology’, 27–31 October 1996, Lorne, VIC.
pp. 375–378. (International Association of Wildland Fire: Missoula, MT)
Clements HB, Alkidas A (1973) Combustion of wood in methanal flames.
Combustion Science and Technology 7, 13–18. doi:10.1080/
00102207308952337
Cohen JD (2004) Relating flame radiation to home ignition using modelling
and experimental crown fires. Canadian Journal of Forest Research 34,
1616–1626. doi:10.1139/X04-049
Douglas GB, Ellis PF (2001) Integrating land use planning and construction
standards for protection for bushfire in NSW – a model. In ‘Proceedings,
2001 Australasian Bushfire Conference’, 3–6 July 2001, Christchurch,
New Zealand. (Eds G Pearce, L Lester) pp. 173–180. (New Zealand
Forest Research Institute: Rotorua)
Drysdale D (1985) ‘An Introduction to Fire Dynamics.’ (Wiley:
Chichester, UK)
Dupuy JL, Marechal J, Morvan D (2003) Fires from a cylindrical forest fuel
burner: combustion dynamics and flame properties. Combustion and
Flame 135, 65–76. doi:10.1016/S0010-2180(03)00147-0
Ellis PF (2000) Review of current methodology of assessment of bushfire
hazard and the prescription of appropriate separation distances and
building standards. CSIRO Forestry and Forest Products Client Report
901. (Canberra, ACT)
Frandsen WH (1971) Fire spread through porous fuels from the conservation
of energy. Combustion and Flame 16, 9–16. doi:10.1016/S0010-2180
(71)80005-6
Gaydon AG, Wolfhard HG (1960) ‘Flames: their Structure, Radiation and
Temperature.’ (Chapman & Hall Ltd: London)
Gill AM, Knight IK (1991) Fire measurement. In ‘Conference on Bushfire
Modelling and Fire Danger Rating Systems. Proceedings’, 11–13 July
1998, Canberra, ACT. (Eds NP Cheney, AM Gill) pp. 137–147. (CSIRO:
Melbourne)
Gould JS, Knight IK, Sullivan AL (1997) Physical modeling of leaf scorch
height from prescribed fires in young Eucalyptus sieberi regrowth forest
in south-eastern Australia. International Journal of Wildland Fire 7
,
7–20. doi:10.1071/WF9970007
Gould JS, McCaw WL, Cheney NP, Ellis PF, Knight IK, Sullivan AL (2007)
‘Project Vesta – Fire in Dry Eucalypt Forest: Fuel Structure, Fuel
Dynamics and Fire Behaviour.’ (Ensis–CSIRO: Canberra ACT, and
Department of Environment and Conservation: Perth, WA)
Gould JS, McCaw WL, Cheney NP (2011) Quantifying fine fuel dynamics
and structure in dry eucalypt forest (Eucalyptus marginata) in Western
Australia for fire management. Forest Ecology and Management.
doi:10.1016/J.FORECO.2011.04.022
Hagglund B, Persson LE (1974) An experimental study of radiation
from wood flames. Forvarets Forhningsanstalt, FAO 4 Rapport
C4589
]
D6(A3)
Johnson VJ (1982) The dilemma of flame length and intensity. Fire
Management Notes 43(4), 3–7.
Kautz J (1997) Appendix C – insulated boxes for protecting video cameras.
Surviving fire entrapments: comparing conditions inside vehicles and
fire shelters. (Ed. R Mangan) USDA Forest Service, Technology and
Development Program, Technical Report 9751–2817-MTDC,
pp. 39–40. (Missoula, MT)
Knight IK, Sullivan AL, Hutchings PT, Cheney SP (2002) Feasibility study
of a gas-fired bushfire flame front simulator. Final Report, CSIRO
Forestry and Forest Products Client Report 1062. (Canberra)
Linn RR, Reisner J, Colman J, Winterkamp J (2002) Studying wildfire
behaviour using FIRETEC. International Journal of Wildland Fire 11,
233–246. doi:10.1071/WF02007
Mangan R (1997) Surviving fire entrapments: comparing conditions inside
vehicles and fire shelters. USDA Forest Service, Technology and
Development Program, Technical Report 9751–2817-MTDC.
(Missoula, MT)
Marcelli T, Santoni PA, Simeoni A, Leoni E, Porterie B (2004) Fire spread
across pine needle fuel beds: characterization of temperature and
velocity distributions within the fire plume. International Journal of
Wildland Fire 13, 37–48. doi:10.1071/WF02065
Maughan D, Krusel N, Boura J, Caling T (1999) House safety zones:
a theoretical model (draft). CFA Risk Management Report.
McArthur AG (1967) Fire behaviour in eucalypt forests. Commonwealth of
Australia Forestry and Timber Bureau Leaflet 107. (Canberra, ACT)
McCaffrey BJ (1979) Purely buoyant diffusion flames: some experi-
mental results. US National Bureau of Standards, Center for Fire
Research, National Engineering Laboratory Report NBSIR 79–1910.
(Washington, DC)
280 Int. J. Wildland Fire B. M. Wotton et al.
Mercer GN, Weber RO (2001) Fire plumes. In ‘Forest Fires: Behavior and
Ecological Effects’. (Eds EA Johnson, K Miyanishi) pp. 225–255.
(Academic Press: San Diego, CA)
Moore PHR, Gill AM, Kohnert R (1995) Quantifying bushfires for ecology
using two electronic devices and biological indicators. CalmScience
Supplementary 4, 83–87.
Morandini F, Simeoni A, Santoni PA, Balbi JH (2005) A model for the
spread of fire across a fuel bed incorporating the effects of wind and
slope. Combustion Science and Technology 177, 1381–1418.
doi:10.1080/00102200590950520
Packham DR (1970) Heat transfer above a small ground fire. Australian
Forest Research 5(1), 19–24.
Packham DR, Pompe A (1971) Radiation temperatures of forest fires.
Australian Forest Research 5(3), 1–8.
Philpot CW (1965) Temperature in a large natural-fuel fire. USDA Forest
Service, Pacific Southwest Forest and Range Experiment Station,
Research Note PSW-90. (Berkeley, CA)
Putnam T, Butler BW (2004) Evaluating fire shelter performance in
experimental crown fires. Canadian Journal of Forest Research 34,
1600–1615. doi:10.1139/X04-091
Pyne SJ, Andrews PL, Laven AR (1996) ‘Introduction to Wildland Fire.’
2nd edn. (Wiley: New York)
Rothermel RC, Deeming JE (1980) Measuring and interpreting fire
behaviour for correlation with fire effects. USDA Forest Service,
Intermountain Forest and Range Experiment Station, General Technical
Report INT-93. (Ogden, UT)
Santoni PA, Balbi JH (1998) Modelling of two-dimension flame spread
across sloping fuel bed. Fire Safety Journal 31, 201–225. doi:10.1016/
S0379-7112(98)00011-3
Santoni PA, Marcelli T, Leoni E (2002) Measurement of fluctuating
temperatures in a continuous flame spreading across a fuel bed using a
double thermocouple probe. Combustion and Flame 131, 47–58.
doi:10.1016/S0010-2180(02)00391-7
Santoni PA, Simeoni JL, Rossi F, Bosseur F, Morandini F, Silvani X, Balbi
JH, Canellieri D, Rossi L (2006) Instrumentation of wildland fire:
characterisation of a fire spreading through a Mediterranean shrub. Fire
Safety Journal 41, 171–184. doi:10.1016/J.FIRESAF.2005.11.010
Sullivan AL, Knight IK, Cheney NP (2002) Predicting the radiant heat
flux from burning logs in a forest following a fire. Australian Forestry
65, 59–67.
Sullivan AL, Ellis PF, Knight IK (2003) A review of radiant heat flux models
used in bushfire applications. International Journal of Wildland Fire 12,
101–120. doi:10.1071/WF02052
Taylor SW, Wotton BM, Alexander ME, Dalrymple GN (2004) Variation in
wind and crown fire behaviour in a northern jack pine–black spruce
forest. Canadian Journal of Forest Research 34, 1561–1576.
doi:10.1139/X04-116
Thomas PH (1967) Some aspects of the growth and spread of fire in the open.
Forestry 40, 139–164. doi:10.1093/FORESTRY/40.2.139
Van Wagner CE (1967) Calculation on forest fire spread by flame radiation.
Canadian Department of Forestry and Rural Development, Forestry
Branch Publication 1185. (Ottawa. ON)
Van Wagner CE (1970) On the value of temperature data in forest fire
research. Canadian Forestry Service, Petawawa Forest Experimental
Station Internal Report PS-20. (Ottawa. ON)
Vines RG (1981) Physics and chemistry of rural fires. In ‘Fire and the
Australia Biota’. (Eds AM Gill, RH Groves, IR Noble) pp. 129–149.
(Australian Academy of Science: Canberra, ACT)
Walker JD, Stocks BJ (1968) Thermocouple errors in forest fire research.
Fire Technology 4, 59–62. doi:10.1007/BF02588607
Wotton BM, Martin TL (1998) Temperature variation in vertical flames
from a surface fire. In ‘Proceedings 3rd International Conference on
Forest Fire Research and 14th Conference on Fire and Forest Meteorol-
ogy’, 16–20 November 1998, Luso, Portugal. pp. 533–545. (ADAI:
Coimbra, Portugal)
Wotton BM, Martin TL, Engel K (1998) A vertical flame intensity
profile from a surface fire. In ‘Proceedings 13th International
Conference of Forest Fire and Meteorology’, 27–31 October 1996,
Lorne, VIC. pp. 175–182. (International Association of Wildland Fire:
Missoula, MT)
Zarate L, Arnaldos J, Casal J (2008) Establishing safety distance for wildland
fires. Fire Safety Journal 43, 565–575. doi:10.1016/J.FIRESAF.
2008.01.001
www.publish.csiro.au/journals/ijwf
Flame temperature and residence time in eucalypt forest fires Int. J. Wildland Fire 281
A preview of this full-text is provided by CSIRO Publishing.
Content available from International Journal of Wildland Fire
This content is subject to copyright.
