Content uploaded by Miltiadis Athanasiou
Author content
All content in this area was uploaded by Miltiadis Athanasiou on Nov 19, 2018
Content may be subject to copyright.
The browsing of UC Digitalis, UC Pombalina and UC Impactum and the consultation and download of
titles contained in them presumes full and unreserved acceptance of the Terms and Conditions of
Use, available at https://digitalis.uc.pt/en/terms_and_conditions.
As laid out in the Terms and Conditions of Use, the download of restricted-access titles requires a
valid licence, and the document(s) should be accessed from the IP address of the licence-holding
institution.
Downloads are for personal use only. The use of downloaded titles for any another purpose, such as
commercial, requires authorization from the author or publisher of the work.
As all the works of UC Digitalis are protected by Copyright and Related Rights, and other applicable
legislation, any copying, total or partial, of this document, where this is legally permitted, must contain
or be accompanied by a notice to this effect.
Observations on wildfire spotting occurrence and characteristics in Greece
Author(s: Athanasiou, Miltiadis; Xanthopoulos, Gavriil
Published by: Imprensa da Universidade de Coimbra
Persistent URL: URI:http://hdl.handle.net/10316.2/44582
DOI: DOI:https://doi.org/10.14195/978-989-26-16-506_65
Accessed : 19-Nov-2018 12:01:58
digitalis.uc.pt
pombalina.uc.pt
AdvAnces in
Forest Fire reseArch
2018
EDITED BY
DOMINGOS XAVIER VIEGAS
ADAI/CEIF, UNIVERSITY OF COIMBRA, PORTUGAL
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
https://doi.org/10.14195/978-989-26-16-506_65
Advances in Forest Fire Research 2018 – Page 588
Observations on wildfire spotting occurrence and characteristics in
Greece
Miltiadis Athanasiou1*; Gavriil Xanthopoulos2
1Wildfire Management Consulting and Training. 8 Thoma Paleologou st., Acharnes, 13673,
Athens, Greece, {info@m-athanasiou.gr*}
2Hellenic Agricultural Organization "Demeter". Institute of Mediterranean Forest Ecosystems.
Terma Alkmanos, Ilisia, 11528, Athens, Greece, {gxnrtc@fria.gr}
Abstract
This paper presents a study on the phenomenon of spotting in some of the most common forest vegetation
types in Greece, during wildfires in the 2007-2017 period. Monitoring and documenting selected wildfires
during this period, noting the appearance or absence of spot fires and the prevailing conditions at the time, a
database of 166 field observations was developed. The database includes information on the number of
observed spot fires (Nκ), the in situ measured relative humidity (RH, %) values, the wind speed, the forest
fuel type where the firebrands had landed, namely maquis, phrygana and grasses, the maximum spotting
distance (Dκ, m) from the fire perimeter, the fire perimeter segment (head or flank) where the firebrands
came from, and the fire type, namely surface, passive crown, active crown and plume dominated fire.
The database was analyzed first by examining the correlation of RH values with Nκ for the three fuel
types. An ordinal variable named Κκ was created in order to represent the following four empirical classes:
a) no spotting (Nκ=0, Κκ=0), b) rare spotting (Nκ<3, Κκ=1), c) limited spotting (3 ≤ Νκ ≤ 9, Κκ=2) and d)
profuse/massive spotting (Νκ ≥ 10, Κκ=3). At RH values higher than 46%, no spotting ignition was recorded.
Massive spotting that triggered extreme fire behavior, was documented for RH values lower than 17%. The
RH thresholds for spotting occurrence that were identified for the three forest fuel types on which the
firebrands landed, are presented and discussed. The Dκ and the Nκ were correlated with both the fire type
and the fire segment on which they were observed. Their descriptive statistics are also presented and
discussed. The study confirmed the great spotting potential of the plume dominated wildfires, both in regard
to spotting distance and the number of spot fires.
Keywords: Spotting, Spot fire, Forest fire, Wildfire behaviour, Firefighting, Greece
Spotting ignition is one of the three significant mechanisms of wildfire spread. It can be considered
as a discontinuous fire spread mechanism (Koo et al. 2010) that is synonymous with solid mass
transport (Albini1979, Alexander 2009). The transport of burning fire embers outside the fire
perimeter, is a cause of serious concern to firefighters because it affects fire behaviour and difficulty
of control and poses a serious threat for them and for civilians.
Spotting involves the source of firebrands, how far they travel, and the probability of ignition on
landing (Rothermel 1983). It is mainly caused by lofted firebrands, including burning tips of branches,
cones, and pieces of bark, that fly and land beyond the main fire perimeter, but may also be caused by
burning cones or pieces of wood rolling down steep slopes (Van Wagner 1988). The type of forest
vegetation that is burning is important for the creation of firebrands.
The probability of ignition at the point where a firebrand lands, is a function of both firebrand size
and temperature. It has been found that as firebrand size is reduced, increased temperature is required
for ignition (Hadden and Scott 2011). Additionally, the probability of ignition depends on the
characteristics of the dead fuels where the firebrand lands, such as fuel quantity, dimension (fineness),
arrangement (compactness and continuity) and fuel moisture content (FMC, %). Atmospheric relative
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 589
humidity (RH, %) affects directly the FMC of dead fuels, the effect being more dramatic and the
response faster for the finer ones. Thus, significant differences in spotting may exist, depending on the
forest vegetation properties. Firebrands in the flaming phase are more capable to ignite fuel beds with
no air flow than the ones in the glowing phase with air flow (Ganteaume et al. 2009) while when fire
danger is high, the ignition probability of flaming firebrands that land on fine fuels, approaches 100%
(Ellis 2012).
A spotting distance of up to 200 meters (m) corresponds to short-range spotting and is common in
high intensity wildfires, while distance values between 200 m and 1 kilometer (km) (Bushfire CRC
2009) or between 200 m and 2 km (Alexander 2009) can be considered as medium-range spotting. A
spotting distance greater than 1 or 2 km, which is very common in some forest types, such as the
eucalypt forests of Australia, can be considered as long-range spotting.
The number of firebrands generated and the rapidity of development of the spot fires, determine the
magnitude of the phenomenon and its effect on wildfire behaviour (Ellis 2012). Spotting usually
exacerbates fire suppression activities and plans, is the leading cause of loss of structures in fires in
Wildland Urban Interface (WUI) areas and is a major concern regarding the safety of firefighters and
the public (Alexander 2009).
In Greece, spotting occurs often in all of its Mediterranean vegetation types, such as the
Mediterranean pine forests, the evergreen shrublands (maquis), the low scrubland vegetation called
phrygana and the grasslands (Athanasiou and Xanthopoulos 2013). As the fuel characteristics are
important for the creation of firebrands and the characteristics of the fuel bed where firebrands land
affect the probability of ignition, in addition to the FMC, the objective set for this study was to examine
the occurrence and characteristics of spotting in three main fuel types in Greece, namely maquis,
phrygana and grasses, especially in relation to the prevailing RH, since this affects the FMC of the
dead fuels. Most of the work has been carried out as part of the Ph.D. dissertation of the first author
(Athanasiou 2015).
Systematic observation, recording and measurements of spotting on maquis, phryganic areas and
grass during the spread of wildfires, started in 2007 in Greece and continues until today. The procedure
followed has been described in Athanasiou and Xanthopoulos (2010). An initial data set of 75 cases
was analysed and preliminary findings were presented in 2013 (Athanasiou and Xanthopoulos 2013).
Ninety-one (91) additional spotting observations that were collected during the following fire seasons,
resulted in a total of 166 cases in which the phenomenon was either present or absent, presenting an
opportunity for testing and extending the initial conclusions. The length of observation for each case
varied from at least five minutes to almost half an hour, depending on the conditions and the potential
risk.
The database that was developed, consists of 166 spotting observations (n=166) that include
information about a) the number of the spot fires (Nκ), b) the in situ measured RH values, c) the wind
speed at the height of 10 m (Wind10m, km/h) and at eye level (calculating one from the other, depending
on which of the two was actually measured), d) the forest fuel type where the firebrands had landed,
e) the maximum spotting distance from the fire perimeter (Dκ, m), f) the fire perimeter segment (head
or flank) where the firebrands came from, and g) the fire type, namely surface, passive crown, active
crown and plume dominated fire during the spread of which the measurements had been conducted
(Figure 1).
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 590
a. The first spot fire is recorded at 15:23:56 at a distance of 200 m
b. Two more spot fires at 15:24:04 at 80 and
110 m (while the first one at the left is growing)
c. Five more spot fires have occured at 180, 200, 230
and 240 m and some of them have already merged, two
minutes and six seconds later (at 15:26:10).
d. Three more spot fires at 15:26:54 at 50
and 80 m (in total, eleven spot fires recorded,
within a period of 2 minutes and 58 seconds)
e. Wildfire evolution at 15:28:10
Figure 1 - Spot fires (Nκ=11 & Dκ=240 m) recorded on a 84% slope, mainly covered by Sarcopoterium spinosum and
Cistus spp. (phryganic vegetation), while a wind driven passive crown wildfire spreads through an Aleppo pine stand.
Weather conditions: T=29οC, RH=43.5%, Wind speed at eye level= 15 km/h
Meteorological measurements and the relative necessary information about month and time of day,
fuels and topography that had been collected on site, were utilised for calculating fine (1-h) dead Fuel
Moisture Content (FDFMC, %) values by using Rothermel’s methodology (1983). The database was
analysed first by examining the correlation of RH and FDFMC values with the Nκ for the three fuel
types. An ordinal variable named Κκ was created in order to represent the following four empirical
classes: a) no spotting (Nκ=0, Κκ=0), b) rare spotting (Nκ<3, Κκ=1), c) limited spotting (3 ≤ Νκ ≤ 9,
Κκ=2) and d) profuse/massive spotting (Νκ ≥ 10, Κκ=3). The Nκ and the Dκ were also examined for
correlation with both the fire type and the fire perimeter segment on which they had been observed.
For the observations/records where Nκ > 1, the Dκ value was the distance of the farthest spot fire from
the fire perimeter (e.g. Figure 1).
At RH values higher than 46%, no spotting ignition was recorded. Massive spotting that triggered
extreme fire behaviour, was documented for RH values lower than 17%. The RH and FDFMC
thresholds for spotting occurrence that were found, for the three forest fuel types on which the
firebrands landed, are reported in Table 1. The descriptive statistics of Nκ and Dκ in relation to both
the fire type and the fire perimeter segment, are also presented in Table 2 and 3, respectively. Dκ
values were only available for 58 of the 67 cases in which spotting was observed.
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 591
Table 1 - Ranges of values and thresholds of RH and FDFMC for spotting occurrence, on maquis, phrygana and
grass
n- RH – FDFMC
RH – FDFMC
Κκ
n
Maquis
Phrygana
(Sarcopoterium spinosum)
Grass
(15-62.5) -
(4-14)
0
99
43-(15 – 62.5) – (4-14)
29-(15 - 50) – (4-12)
27-(23.8-55) – (4-
11)
(15–46) -
(3-11)
1
31
9-(20-38.5) – (4-11)
15-(15-46) – (3-10)
7-(16-31) – (3-6)
(14–35) -
(2-9)
2
18
7-(18-33) – (4-8)
4-(34.5-35) – (6-9)
7-(14-30) – (2-7)
(13–46) -
(3-8)
3
18
6-(16-21.4) – (3-7)
1-(43.5) - (8)
11-(13-46) – (3-8)
Total:
166
65
49
52
Table 2 - Descriptive statistics of Nκ, per fire type and perimeter segment for the 67 cases with spotting
Surface
(n= 22)
Passive crown
(n= 21)
Active crown
(wind driven) (n= 7)
Crown (plume)
(n= 17)
Nκ
Flank
(n=7)
Head
(n=15)
Flank
(n=7)
Head
(n=14)
Head
(n=7)
(n=17)
Mean value
1
2
3
5
3
21
Median
1
1
3
4
2
22
Mode
1
1
3
1
1
30
Std. Dev.
1
2
2
6
2
9
Minimum
1
1
1
1
1
5
Maximum
3
8
7
22
6
30
Wind10m
8-30
6-111
10–26
0–27
RH
16-42
15-46
20-35
14-46
20-35
13-21
FDFMC
3-7
3-10
4-9
2-11
4-9
3-7
Table 3 - Descriptive statistics of Dκ, per fire type and perimeter segment (n=58)
Surface (n= 22)
Passive crown (n= 17)
Active crown
(wind-driven) (n= 7)
Crown (plume)*
(n= 12)
Dκ (m)
Flank
(n= 7)
Head
(n= 15)
Flank
(n= 6)
Head
(n= 11)
Head
(n= 7)
(n= 12)
Mean value
51
132
71
118
229
392
Median
20
70
60
100
250
250
Mode
15
10
N/A
100
250
200
Std. Dev.
58
143
48
85
175
308
Minimum
10
5
15
20
50
150
Maximum
150
500
150
300
500
1,200
Wind10m
8-30
6-36
10–26
0–27
RH
16-42
15-46
20-35
15-46
20-35
13-18
FDFMC
3-7
3-10
4-9
3-11
4-9
3-6
For the 67 cases in which spotting was observed, RH and FDFMC were also plotted versus Nκ
(Figure 2 & 3).
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 592
Figure 2 - Plot of RH and Nκ for the 67 records with spotting (fi: absolute frequency)
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 593
Figure 3 - Plot of FDFMC and Nκ for the 67 records with spotting (fi: absolute frequency)
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 594
The maximum RH value at which spotting occurred (Nκ=1) on maquis, was 38.5% (Table 1, Figure
2) and the corresponding calculated FDFMC value was 11% (Figure 3). The RH and FDFMC
thresholds were 46% and 10% (Nκ=1) for phrygana and 46% and 8% for grass (Nκ=12). Regarding
phrygana, it is worth noting that there was also one observation (Figure 1) of Kκ=3 spotting class
(Nκ=11), at relatively high RH and FDFMC values, 43.5% and 8% respectively. In this case, the
radiation emitted against the fuel bed, from the leaning smoke column, played a crucial role in
preheating the fuels including the fine dead ones (Figure 1). Moreover, regarding the previously
mentioned maximum RH and FDFMC datapoint for grass, 12 spot fires were recorded on a grassland,
at RH=46% (Figure 2) and FDFMC=8% (Figure 3), at a Dκ of 30 m. The fire brands originated from
a torching Pinus halepensis tree and one of them broke into smaller ones upon landing. According to
Gould et al. (2007) this is a potentially important notification, and it should be included in any future
analysis if available.
The RH versus Nκ (Figure 2) and FDFMC versus Nκ (Figure 3) plots, show that spotting tends to
be rare when RH > 40% or when FDFMC > 10%. The former trend is consistent with Weir’s (2004)
conclusion who examined 99 prescribed fires and found that spotting occurrence was very rare when
RH exceeded 40% The latter one is in general agreement with the finding of Manzello et al. (2006)
that embers with mass of 0.5 gr could ignite pine needles with fuel moisture of 11% or less and with
the finding of Ellis (2000) that for fine fuel moistures below 9%, flaming embers with mass between
0.7 and 1.8 gr had a 100% probability of igniting the Monterey pine litter while glowing ones had
lower probabilities. With a light wind (1 m/sec) the probability of ignition was found to be about 20%
at fine fuel moisture content of 9% and approximately 65% at a fine fuel moisture content of 3.5%
(Ellis 2000).
Τhe great spotting potential of the plume dominated wildfires, both in regard to the Dκ, and the Nκ,
was confirmed (Table 2 & 3). Future modeling of the shed-vortex transport (Berlad and Lee 1968) and
of the plumes’ characteristics, may shed light into the long-distance transport aspect for this fire type.
Regarding wind driven fires, Wind10m was not found to be a Dκ predictor, not even a poor one, for the
“head spread” subset. A possible reason is because a fire brand does not always originate at the fire
perimeter, so the total horizontal distance it has traveled, is not known and is not necessarily equal to
Dκ (the distance between the fire perimeter and the farthest recorded spot fire). Furthermore, in
addition to the ambient atmospheric conditions, the trajectories of fire brands are also affected by the
tilted or vertical turbulence and currents of the convection column.
Additionally, the spot fires that were documented at the flanks of surface and passive crown fires,
were not in a windless environment and there was a component of wind of unknown velocity and
direction that temporarily drove them. Moreover, fire behavior is sometimes a poor predictor of
spotting distance or number of spot fires: as found by Racher (2003), wildfires producing the most
distant or numerous landing embers are not always those with the greater rate of spread, flame height
or flame depth zone.
The analysis of the subset of 67 records of spotting occurrence did not show a strong correlation
between Nκ and RH or FDFMC. Tables 1, 2 and 3 as well as Figures 2 and 3, may offer some
guidelines and relative practical advice to firefighters and fire behaviour practitioners but they are not
predictive tools. Number and size of embers produced by various fuel types, is essential information,
in order to estimate how many embers could be carried downwind and how far downwind they will
go, to determine whether spotting (Nκ and Dκ) will affect the fire rate of spread and to assess whether
the spot fires will be numerous enough to merge readily. According to the field observations, the
patterns of spatial distribution of spot fires vary, depending on the fire segment the firebrands originate
from, the wind field and the landscape (Figure 4a & 4b).
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 595
Figure 4a - An along-wind pattern of landing embers, on grass, downwind from a surface head fire
Figure 4b - An across-wind pattern of landing embers, on phrygana beyond the flank of a surface wildfire that
spreads through rough landscape
The along-wind pattern of groups of spot fires has been observed to form a roughly elliptical shape
(Figure 4a) and seems to be more predictable, while the across-wind one (Figure 4b) is irregular.
Although these patterns have not yet been thoroughly described, they may allow firefighters to get a
feeling on what to expect, a practically useful information: those patterns may play a crucial role in
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 596
the behavior as the proximity among the spot fires affects their merging rate and the overall effective
rate of spread and intensity. As proposed by Gould et al. (2007), they can be studied in the field, leading
to the development of empirical functions.
The findings of the work presented here is that no spotting ignition was recorded on maquis,
phrygana and Mediterranean grasslands, at RH values higher than 46% and that massive spotting that
triggered extreme fire behaviour, was documented for RH values lower than 17%.
The RH threshold below which a spot fire is most likely to occur seems to be close to the value of
40%. However, it was also observed that a significant number of spot fires may take place even if RH
values range between 40% and 46%. The finding that this is most likely to occur on fine fuels
(phrygana and grass) should be taken into consideration, in operational firefighting.
The patterns of the spatial distribution of spot fires, and their basic characteristics in head and flank
fires, can be included in practical guidelines about spotting
Future work is expected to shed additional light on the issues discussed in this paper, as field data
continue being collected, ultimately improving fire behaviour prediction and firefighter safety in
Greece.
The research reported here is part of the Ph.D. thesis of the first author. It was sponsored, in part,
by the International Association of Wildland Fire (IAWF) through the Doctoral Student Scholarship
Award for 2014. Participation of the second author was in the frame of the project MedWildFireLab
(“Global Change Impacts on Wildland Fire Behaviour and Uses in Mediterranean Forest Ecosystems,
towards a «wall less» Mediterranean Wildland Fire Laboratory”) a European ERANet FORESTERRA
project with funding from the General Direction for the Development and Protection of Forests and
Agro-environment of the Greek Ministry of Environment and Energy.
Albini F.A., 1979. Spot fire distance from burning trees: a predictive model. USDA Forest Service,
Intermountain Forest and Range Experiment Station, Technical Report INT-56. (Ogden, UT).
Alexander, M.E. 2009. Some pragmatic thoughts on the prediction of spotting in wildland fires.
MITACS/GEOIDE Conference on Forest Fire Modelling, June 22-23, 2009 – Hinton, AB.
Athanasiou, M. and G. Xanthopoulos 2010. Fire behaviour of the large fires of 2007 in Greece. In
proceedings of the 6th International Conference on Forest Fire Research, 15-18 November 2010,
Coimbra, Portugal. D.G. Viegas, Editor. ADAI/CEIF, University of Coimbra, Portugal. Abstract p.
336, full text on CD.
Athanasiou M. and G. Xanthopoulos 2013. Observations of the spotting phenomenon, in wildfires in
Greece. 30-40 p. In proceedings of the 16th National Forestry Conference, October 6-9, 2013,
Thessaloniki, Greece. Hellenic Forestry Society. 1144 p. (in Greek).
Athanasiou, M. 2015. Development of an optimal methodology for forecasting forest fire behaviour
in Greece. PhD dissertation, Department of Geology and Geoenvironment, National and
Kapodistrian University of Athens, 408 p. (In Greek with English abstract).
Berlad AL, Lee SL 1968. Long-range spotting. Combustion and Flame 12, 172–174.
Bushfire CRC. 2009. Victorian Bushfires 2009 Research Taskforce. October 2009. Final Report.
Advances in Forest Fire Research 2018 - D. X. Viegas (Ed.)
Chapter 3 – Fire Management
Advances in Forest Fire Research 2018 – Page 597
Byram, G.M. 1959. Combustion of forest fuels; Forest fire behavior. in Davis, K.P., ed. Forest fire:
Control and use. New York: McGraw-Hill: 61-89, 90-123.
Ellis, P.F.M. 2000. The aerodynamic combustion characteristics of eucalypt bark – a firebrand study.
A Thesis submitted to the Australian National University for the Degree of Doctor of Philosophy.
Ellis, P.F.M. 2012. A review of empirical studies of firebrand behavior. CSIRO Ecosystem Science
and CSIRO Climate Adaptation Flagship.
Ganteaume A., Lampin-Maillet C., Guijarro M., Hernando C., Jappiot M., Fonturbel T., Perez-
Gorostiaga P. and J.A. Vega. 2009. Spot fires: fuel bed flammability and capability of firebrands to
ignite fuel beds. International Journal of Wildland Fire, 18, 951-969.
Gould, J.S. McCaw, W.L. Cheney N.P. Ellis, P.F. Knight, I.K. Sullivan, A.L. 2007. Project Vesta-Fire
in dry eucalypt forests: fuel structure, fuel dynamics, and fire behaviour. Perth, WA: Ensis-CSIRO,
Canberra ACT and Department of Environment and Conservation. 218 p.
Hadden, R., Scott, S., 2011. Ignition of Combustible Fuel Beds by Hot Particles: An Experimental and
Theoretical Study. Fire Technology 47:341–355.
Koo, E., Pagni, P.J., Weise, D.R., Woycheese, J.P., 2010. Firebrands and spotting ignition in large-
scale fires. International Journal of Wildland Fire 19:818–843.
Manzello, S.L., Cleary, T.G., Shields, J.R., and Yang, J. C. 2006. On the ignition of fuel beds by
firebrands. Fire and Materials 30: 77–87.
Racher, B. 2003. Prescription development for burning two volatile fuel types. Lubbock. TX: Texas
Tech University. 72 p. Ph.D. dissertation.
Rothermel, R.C., 1983. How to predict the spread and intensity of forest and range fires. Gen. Tech.
Rep. INT-143. Ogden, UT: U.S. Department of Agriculture, Forest Service, Intermountain Forest
and Range Experiment Station. 161 p.
Van Wagner, C.E. 1988. Effect of slope on fires spreading downhill. Can. J. For. Res. 18:818-820.
Weir, John. 2004. Probability of Spot Fires During Prescribed Burns. Fire Management Today. 64.
24-26.
