Content uploaded by Ioannis Mitsopoulos
Author content
All content in this area was uploaded by Ioannis Mitsopoulos
Content may be subject to copyright.
Climatic Change (2011) 109:331–347
DOI 10.1007/s10584-011-0026-8
Impact of drought on wildland fires in Greece:
implications of climatic change?
Alexandros P. Dimitrakopoulos ·M. Vlahou ·
Ch. G. Anagnostopoulou ·I. D. Mitsopoulos
Received: 24 June 2008 / Accepted: 21 January 2011 / Published online: 17 February 2011
© Springer Science+Business Media B.V. 2011
Abstract An increasing trend and a statistically significant positive correlation
between wildfire occurrence, area burned and drought (as expressed by the Stan-
dardized Precipitation Index, SPI) have been observed all over Greece, during the
period 1961–1997. In the more humid and colder regions (Northern and Western
Greece) the number of fires and area burned were positively correlated to both
summer (SPI6_October) and annual drought (SPI12_September), whereas in the
relatively more dry and hot regions (Southern and Central Greece) the number of
fires and area burned were correlated only to summer drought. In 1978, Greece
entered a period of prolonged drought, possibly as a result of the global climatic
change. Data analysis of the period 1978–1997 revealed a statistically significant
increase in the mean annual number of fires, the area burned and the summer
and annual drought episodes in the relatively more humid and colder regions
(Northern and Western) of Greece (which in the past were characterized by less
fires and area burned) compared to the more dry and hot regions (Southern and
Eastern Greece), which always presented high fire activity. Additionally, analyzing
the two sub-periods (1961–1977, 1978–1997) separately, drought was significantly
correlated only to fire occurrence during the years 1961–1977, whereas during 1978–
1997 drought was significantly correlated mainly to area burned. It became obvious
To the memory of my beloved father Panagiotis A. Dimitrakopoulos (1926–2011).
A. P. Dimitrakopoulos ·M. Vlahou ·I. D. Mitsopoulos
Faculty of Forestry and Natural Environment,
Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
A. P. Dimitrakopoulos (B)
Laboratory of Forest Protection, School of Forestry and Natural Environment,
Aristotle University of Thessaloniki, P.O. Box 228, 54124 Thessaloniki, Greece
e-mail: alexdimi@for.auth.gr
Ch. G. Anagnostopoulou
Department of Geology, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
332 Climatic Change (2011) 109:331–347
that drought episodes, although they are not solely responsible for fire occurrence
and area burned, they exert an increasingly significant impact on wildfire activity
in Greece.
1 Introduction
An increase in the number of wildfires and area burned has been reported during
the last decades in many parts of the world by several authors (Stocks et al. 1998;
Flannigan et al. 2003; Brown et al. 2004; European Commission 2004). Many factors
are considered to influence wildland fire activity in a region, including weather
conditions (Balling et al. 1992b), human activities (Vélez 1993), fuel characteristics
(Schoenberg et al. 1991;DeLuísetal.2001; Mouillot et al. 2002), fire management
activities (Badia et al. 2002; Fried et al. 2008), land use changes (Rego 1992)and
climatic change (Davis and Michaelsen 1995; Flannigan et al. 2000; Fried et al.
2004). Recently, there is much consideration about the relationship between cli-
matic change and fire occurrence, globally. One way to analyze this relationship
is by using raw meteorological data (Vázquez and Moreno 1993;Pausas2004).
Moreover, drought indices (Schlobohm and Brown 2001; Hall and Brown 2003;
Dimitrakopoulos and Bemmerzuk 2003; Collins et al. 2006; Maingi and Henry 2007),
fire weather index systems (Piñol et al. 1998) and Global Circulation Models (GCMs)
(Flannigan and Van Wagner 1991; Torn and Fried 1992; Wotton and Flannigan 1993;
Flannigan et al. 2000; Flannigan and Wotton 2001; De Groot et al. 2003; Brown et al.
2004) have also been used for this purpose.
In 1978, Greece entered a prolonged period of drought as a result of a change in at-
mospheric circulation or possible climatic change (Dalezios et al. 2000; Maheras et al.
2004; Pnevmatikos and Katsoulis 2006;Baltas2007; Livada and Assimakopoulos
2007; Repapis et al. 2007). Maheras et al. (2000,2004) analyzed the rainfall variability
in relation to 500 hPa atmospheric circulation, and concluded that the increased
drought could be attributed to an abrupt change in the frequency of cyclonic and
anticyclonic circulation types in Greece after 1976–1977. The cyclonic types induce
rainfall events, whereas the anticyclonic types are connected with the absence of
rainfall events. Collacino and Conte (1993) attributed the phenomenon to the
greenhouse effect, as a part of the progressive climate change in the Mediterranean
Basin (Bolle 2003).
During the same period, fire statistics show a significant increase in both the
number of wildfires and area burned in Greece. The number of fires doubled and
the area burned tripled during the period 1978–1997, compared to the period 1961–
1977. Several reasons have been speculated for this augmentation in wildfire activity
such as changes in population activities, socioeconomic conditions, land use, fuel
accumulation, drought frequency and duration, etc. (Dimitrakopoulos 1995,2005;
Dimitrakopoulos and Mitsopoulos 2006).
The main scope of this study was to investigate the impact of drought on wildland
fire occurrence and area burned in Greece. The objectives were: (a) to examine the
correlation between drought (as expressed by the Standardized Precipitation Index,
SPI) and wildfire occurrence and area burned, and, (b) to investigate the hypothesis
that the increased fire activity in Greece after 1977 could be a consequence of the
Climatic Change (2011) 109:331–347 333
increased frequency of the drought episodes that have been observed during the
same period, in the context of a possible climatic change.
2 Materials and methods
2.1 Study area
The study area covers the whole Greece and is comprised of 17 prefectures grouped
in five geographical districts (Northern, Western, Central, Eastern and Southern),
which correspond roughly to the climatic variety of Greece (Fig. 1). The Northern
and Western regions are characterized by continental-type climate with higher
annual precipitation and lower temperatures, as opposed to the drier and warmer
Southern regions (Mediterranean-type climate). Regarding the fire regime, the
Southern and Central regions are more fire-stricken (both in terms of fire occurrence
and area burned) than the more humid, colder and less populated Northern and
Western districts. The prefectures used in this study were chosen according to the
spatial distribution of the official (class A) meteorological stations in Greece and the
availability of meteorological data in each station. Table 1shows the specific climatic
type of each prefecture (meteorological station), based on annual precipitation
(Flokas 1994).
19
20
21 22 23 24 25 26 27 28
34
35
36
37
38
39
40
41
42
19
20
21 22 23 24 25 26 27 28
34
35
36
37
38
39
40
41
42
Fig. 1 Geographical distribution of the meteorological stations of Greece, used in this study
334 Climatic Change (2011) 109:331–347
Table 1 Districts, prefectures and climatic types of the meteorological stations in Greece
Districts Prefectures Meteorological Mean annual Climatic
stations precipitation (mm) type
Northern Greece Evros Alexandroupolis 553.2 Semi-wet
Thessaloniki Thessaloniki 448.3 Semi-dry
Kozani Kozani 507.6 Semi-wet
Central Greece Larissa Larissa 423.2 Semi-dry
Southern Greece Attiki Athens 364.8 Semi-dry
Arkadia Tripolis 780.6 Semi-wet
Messinia Kalamata 780.3 Semi-wet
Chania Chania 621.5 Semi-wet
Heraklion Heraklion 483.2 Semi-dry
Lasithi Ierapetra 661.8 Semi-wet
Western Greece Ioannina Ioannina 1081.5 Wet
Kefalonia Argostoli 820.0 Semi-wet
Corfu Corfu 1017.3 Wet
Aitoloakarnania Agrinio 931.2 Semi-wet
Eastern Greece Lesvos Mitilini 648.1 Semi-wet
Rhodes Rhodes 703.0 Semi-wet
Samos Samos 709.0 Semi-wet
2.2 Climate and fire data
In this study, the drought was expressed as SPI (Standardized Precipitation Index)
values, calculated from the precipitation data taken from the 17 official meteorologi-
cal stations (one for every prefecture) of the Greek National Meteorological Service,
during the period 1961–1997 (37 years, the longest time series available in Greece). In
addition, the annual number of wildland fires and area burned data for each of the 17
prefectures were used as measures of wildfire activity. During the period 1961–1997,
in total 17,926 wildland fires occurred in Greece which burned 603,615 ha.
Standardized precipitation is simply the difference of precipitation from the mean
for a specified time period divided by the standard deviation, where the mean and
standard deviation are determined from past records. Since precipitation is not
normally distributed for accumulation periods of 12 months or less, a transformation
is first applied, so that the transformed precipitation values (namely SPI values)
follow a normal distribution. SPI is normalized by transforming the cumulative
probability of a given rainfall into a normalized variant z. Hence, SPI values for a
given region and a given period have a mean of 0 and a standard deviation of 1. For
a region where monthly precipitation is given, SPI can be calculated for any month
considering the precipitation during i previous months, where i =1, 2, 3, . . . 12, . . . 24,
. . . 48, etc. For example, the 6-month SPI (SPI6) for October 1961 is calculated using
the precipitation total of May 1961 through October 1961.
The Standardized Precipitation Index (SPI) is calculated in the following se-
quence. A monthly precipitation data is prepared for a period of mmonths, ideally
a continuous period of at least 30 years. A set of averaging periods are selected to
determine a set of time classes of period imonths where iis 3, 6, 12, 24, or 48 months.
These represent arbitrary but typical time scales for precipitation deficits. The data
set is moving in the sense that each month a new value is determined from the
previous imonths. Each of the data sets is fitted to the Gamma function to define the
Climatic Change (2011) 109:331–347 335
relationship of probability to precipitation. Once the relationship of probability to
precipitation is established from the historic records, the probability of any observed
precipitation data point is calculated and used along with an estimate of the inverse
normal to calculate the precipitation deviation for a normally distributed probability
density. This value is the SPI for the particular precipitation data point (McKee et al.
1993). McKee et al. (1993) assigned the criteria of a drought event at any time-scale
according to the range of the SPI values: >2.0 Extremely wet, 1.5 to1.99 Very wet,
1.0 to1.49 Moderately wet, −0.99 to 0.99 Near normal, −1.0 to −1.49 Moderately dry,
−1.5 to −1.99 Severely dry, <−2.0 Extremely dry.
In this study, the SPI6_October and the SPI12_September time series for the
period 1961–1997, were calculated for every year (Anagnostopoulou 2003). SPI6 of
October values were calculated using the cumulative monthly rainfall data of May
through October and they represent the summer drought (Briffa et al. 1994). SPI12
of September values were calculated using the cumulative monthly rainfall data of
October (previous year) through September (present year), thus representing the
annual drought (Wilhite and Glantz 1985; Estrela et al. 2000).
2.3 Data analysis
Wildfire activity in Greece presents a non-stationary time series whose variation
increases as time progresses (Shatfield 1995). In order to detect a possible fire
regime shift and to identify a threshold change in the fire statistics (number of
fires and area burnt) of Greece, the change point analysis was used as a method
of statistical analysis. Change point analysis is a method for identifying thresholds
in relationships between variables and attempts to find a point along a distribution
of values where the characteristics of the values before and after the point are
different (Andersen et al. 2009; Radionov 2004; Chen and Gupta 2000; Csörgö and
Horváth 1997; Bhattacharya 1994). The analysis was performed by using the Change
Point Analyzer software (Taylor 2000). Change Point Analyzer iteratively uses a
combination of cumulative sum charts and bootstrapping to detect the regime shifts.
Simple linear regression on the fire parameters (annual number of fires and
annual area burned) and SPI values of each district was applied for the whole period
1961–1997, and the two sub-periods 1961–1977 and 1978–1997, separately. Possible
differences in the rate of increase of fire statistics among the five different districts
was investigated by testing the assumption of homogeneity of the regression slopes
with Analysis of Covariance (ANCOVA) (Milliken and Johnson 2001). Mann–
Whitney Utest at p<0.05 was applied to fire and meteorological data, to investigate
the hypothesis that fire occurrence, area burned and drought have significantly
increased from 1978 and after, compared to the previous years. F-test was used
as criterion for the significance of the trends at significance levels of p<0.05 and
p<0.01, with 184 df. The two sub-periods were analyzed separately in order to test
the hypothesis that after 1977, the increased drought episodes are correlated to the
increased fire activity that was observed all over in Greece during this period.
The number of fires and area burned were correlated to the SPI drought index
values (SPI6 and SPI12) by using a non-parametric correlation test (Spearman’s
coefficient). The statistical analysis was conducted with the SPSS 12.0 statistical
software package (Norusis 1997).
336 Climatic Change (2011) 109:331–347
3Results
3.1 Fire occurrence and area burned changes
Figure 2shows the results of the change point analysis for the fire statistics in Greece,
during the period 1961–1997. The shaded areas denote a regime shift in the data. It
was detected a threshold change for both the number of fires and area burned in
Greece beyond the year 1977.
The number of fires showed an increasing trend during the period 1961–1997, all
over Greece (Fig. 3). Specifically, fire occurrence increased in Northern (F=22.8,
p<0.001, R2=0.39), Southern (F=39.3, p<0.001, R2=0.52) and in Western
Greece (F=54.2, p<0.001, R2=0.60). A weaker but also clear increase was
observed in Eastern (F=3.9, p=0.05, R2=0.10) and Central Greece (F=5.6,
p<0.05, R2=0.13).
The number of fires showed an important increase during the sub-period 1978–
1997 compared to the previous sub-period 1961–1977, in almost all Greece (except
for the Central region). Specifically, during the sub-period 1978–1997, the mean
annual number of fires in Northern Greece was significantly higher than the sub-
period 1961–1977. In Southern Greece, during the sub-period 1978–1997, there was
also a significant increase in the annual number of fires, compared to the sub-period
1961–1977. Similar trends were observed also in Western Greece and in Eastern
Greece (Table 2).
Area burned also followed an increasing trend all over Greece during the period
1961–1997 (Fig. 4). Area burned increased in Northern Greece (F=8.9, p<0.05,
R2=0.20), in Southern Greece (F=6.9, p<0.05, R2=0.16), in Western Greece
(F=5, p<0.05, R2=0.12), in Eastern Greece (F=5.3, p<0.05, R2=0.13) and in
Fig. 2 Change point analysis
of the number of fires and area
burned in Greece during the
period 1961–1997. Shaded
areas denote regime shifts.
A threshold change was
determined in 1977.
Horizontal lines denote the
control limits for the values
used in the analysis. Points
outside the control limits
indicate that a change has
occurred. (Confidence level
for candidate changes =50%,
Confidence level for inclusion
in table =90%, Confidence
interval =95%, Bootstraps =
1000, without replacement,
MSE estimates)
Climatic Change (2011) 109:331–347 337
Northern Greece
y = 1.9898x - 3886.3
R2 = 0.39
0
20
40
60
80
100
120
140
160
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Number of fires
Western Greece
y = 9.0019x - 17665
R2 = 0.60
0
100
200
300
400
500
600
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Number of fires
Central Greece
y = 0.7933x - 1537.9
R2 = 0.13
0
20
40
60
80
100
120
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Number of fires
Southern Greece
y = 5.2295x - 10139
R2 = 0.52
0
100
200
300
400
500
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Number of fires
Eastern Greece
y = 0.4137x - 776.85
R2 = 0.10
0
20
40
60
80
100
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Number of fires
Fig. 3 The increasing trend of the number of fires for each district of Greece, for the period 1961–
1997
Central Greece (F=5.15, p<0.05, R2=0.12). Additionally, the mean area burned
was significantly higher during the sub-period 1978–1997 compared with the period
1961–1997 in almost all Greece (except for the Eastern region; Table 2).
Table 2 Mann–Whitney Utest for the number of fires and area burned in Greece during the two
sub-periods (1961–1977, 1978–1997)
Number of fires Northern Greece Western Greece Southern Greece Eastern Greece
Period 1961–1977 29.11 (17.46)a64.6 (39.9) 154 (47.5) 36.7 (7.8)
Period 1978–1997 70.6 (33.72) 221.3 (128) 257.2 (66.4) 46.2 (16.6)
Change of mean (%) +142.5 +242.6 +67 +25.9
U33.5 18 35.5 92.5
Significance level p<0.001 p<0.001 p<0.001 p=0.017
Area burned (ha) Northern Greece Western Greece Southern Greece Central Greece
Period 1961–1977 4784.6 (17.4)a13866.7 (13681.7) 35411.12 (26405.3) 9665 (11743)
Period 1978–1997 20211.8 (21341.7) 50127.4 (56457.1) 100433.23 (85579.1) 28928 (37799)
Change of mean (%) +322.43 +261.49 +183.62 +199.31
U56 57 57 81
Significance level p<0.001 p<0.001 p<0.001 p=0.011
aStandard Deviation in parenthesis
338 Climatic Change (2011) 109:331–347
Northern Greece
y = 766x - 2E+06
R2 = 0.20
0
20000
40000
60000
80000
100000
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Burned area
(stremata)
Western Greece
y = 1509.3x - 3E+06
R2 = 0.12
0
50000
100000
150000
200000
250000
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Burned area
(stremata)
Central Greece y = 998.96x - 2E+06
R2 = 0.12
0
50000
100000
150000
200000
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Burned area
(stremata)
Southern Greece
y = 2720.3x - 5E+06
R2 = 0.16
0
100000
200000
300000
400000
500000
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Burned area
(stremata)
Eastern Greece y = 983.05x - 2E+06
R2 = 0.13
0
50000
100000
150000
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Burned area
(stremata)
Fig. 4 The increasing trend of the area burned for each district of Greece, for the period 1961–1997.
(1 stremma =0.1 ha)
The rate of increase was significantly different among the five districts for the
number of fires (F=13.9, p=0.001, 184 df) but not for the area burned (F=1.59,
p=0.178, 184 df). Southern and Western Greece regarding the number of fires and
only Southern Greece regarding the area burned had a higher increase rate than the
other districts.
3.2 Drought change
The SPI12 (September) values followed a decreasing trend in almost the whole
Greece (except for the Central region), which means that there was an increase
in the annual drought episodes during the period 1961–1997. Specifically, SPI12
(September) values showed a decreasing trend in Northern Greece (F=13.8, p<
0.001, R2=0.28), in Southern Greece (F=7.04, p<0.05, R2=0.16), in Western
Greece (F=12, p<0.01, R2=0.25) and in Eastern Greece (F=7.01, p<0.05,
R2=0.16), during the study period (Fig. 5). The SPI6 (October) values did not show
any statistically significant trend for the same period.
Comparing the two sub-periods, the SPI6 (October) values and the SPI12
(September) values decreased significantly only in Northern and Western Greece
during the sub-period 1978–1997 (M−0.25, SD =0.73), as compared to the sub-
period 1961–1977 (M =0.29, SD =0.89; Table 3). This indicates that both the
Climatic Change (2011) 109:331–347 339
Western Greece
y = -0.0427x + 84.522
R2 = 0.25
-3
-2
-1
0
1
2
3
1950 1960 1970 1980 1990 2000
Year
SPI12
Southern Greece
y = -0.0283x + 55.938
R2 = 0.16
-3
-2
-1
0
1
2
3
1950 1960 1970 1980 1990 2000
Year
SPI12
Eastern Greece
y = -0.0317x + 62.748
R2 = 0.16
-3
-2
-1
0
1
2
3
1950 1960 1970 1980 1990 2000
Year
SPI12
Northern Greece
y = -0.0407x + 80.649
R2 = 0.28
-3
-2
-1
0
1
2
3
1950 1960 1970 1980 1990 2000
Year
SPI12
Fig. 5 The trend of the annual drought episodes (SPI12 – September) in Greece, for the period
1961–1997
summer and annual drought episodes increased significantly in Northern and West-
ern Greece after the 1977.
3.3 Correlation of drought and fire occurrence
The correlation between drought and number of wildland fires was statistically
significant and positive all over Greece, except for the Eastern region, during the
period 1961–1997. In Northern Greece, both SPI6 and SPI12 values were significantly
correlated with the number of fires during the period 1961–1997. The same was
observed in Western Greece, where both SPI6 and SPI12 were also significantly
correlated with fire occurrence during the same period. On the contrary, in Southern
and Central Greece only the summer drought (SPI6) was correlated significantly with
the number of fires for the period 1961–1997 (Table 4).
Table 3 Mann–Whitney Utest for the SPI of Northern and Western Greece during the study periods
Mann–Whitney Utest for SPI Northern Greece Western Greece
Period 1961–1977 SPI6 0.29 (0.8962)a–
SPI12 0.4 (0.64) 0.33 (0.76)
Period 1978–1997 SPI6 −0.257 (0.7395) –
SPI12 −0.33 (0,82) −0.38 (0.91)
Change of mean (%) SPI6 −212.8–
SPI12 −221.2−215.15
USPI6 243 –
SPI12 255 242.5
Significance level SPI6 p=0.026 –
SPI12 p=0.009 p=0.028
aStandard Deviation in parenthesis
340 Climatic Change (2011) 109:331–347
Table 4 Spearman’s correlation coefficient between number of fires, area burned and SPI values for
each district in Greece, during the period 1961–1997
Number of fires Area burned
Northern Greece 1961–1997 SPI6_October −0.611a(p=0.000) −0.535a(p=0.001)
SPI12_September −0.648a(p=0.000) −0.510a(p=0.001)
Southern Greece 1961–1997 SPI6_October −0.377b(p=0.021) −0.514a(p=0.001)
SPI12_September −0.319 (p=0.055) −0.154 (p=0.362)
Western Greece 1961–1997 SPI6_October −0.436a(p=0.007) −0.358b(p=0.029)
SPI12_September −0.519a(p=0.001) −0.602a(p=0.000)
Eastern Greece 1961–1997 SPI6_October −0.212 (p=0.208) −0.168 (p=0.321)
SPI12_September −0.251 (p=0.134) −0.586a(p=0.000)
Central Greece 1961–1997 SPI6_October −0.358b(p=0.029) −0.474a(p=0.003)
SPI12_September −0.193 (p=0.254) −0.156 (p=0.355)
aCorrelation is significant at the 0.01 level (two-tailed)
bCorrelation is significant at the 0.05 level (two-tailed)
Analyzing the two sub-periods separately, the results suggest a significant corre-
lation in Southern and Western Greece between SPI6 and fire occurrence during
the sub-period 1961–1977 (Table 5). In Northern Greece, there was a statistically
significant correlation between both SPI6 and SPI12 and fire occurrence during the
same sub-period. During the sub-period 1978–1997 (Table 6), in Northern Greece
there was a significant correlation between SPI12 values and fire occurrence and in
Western Greece the same was true with the SPI6 values. No significant correlations
were found for the other districts.
3.4 Correlation of drought and area burned
The correlation between drought and area burned was statistically significant and
positive for all Greece during the period 1961–1997 (Table 4). Specifically, in
Northern and Western Greece there was a significant correlation between area
burned and both SPI6 and SPI12 values, for the period 1961–1997. In Southern and
Central Greece the correlation was significant only between the SPI6 values and the
area burned. In Eastern Greece the correlation was significant only with the SPI12
values.
Table 5 Spearman’s correlation coefficient between number of fires, area burned and SPI values for
each district in Greece, during the period 1961–1977
Number of fires Area burned
Northern Greece 1961–1977 SPI6_October −0.683a(p=0.003) −0.417 (p=0.096)
SPI12_September −0.612a(p=0.009) −0.299 (p=0.243)
Southern Greece 1961–1977 SPI6_October −0.499b(p=0.042) −0.326 (p=0.202)
SPI12_September −0.313 (p=0.221) −0.064 (p=0.808)
Western Greece 1961–1977 SPI6_October −0.632a(p=0.007) −0.459 (p=0.064)
SPI12_September −0.424 (p=0.090) −0.326 (p=0.202)
Central Greece 1961–1977 SPI6_October −0.445 (p=0.073) −0.529b(p=0.029)
SPI12_September −0.224 (p=0.388) −0.211 (p=0.417)
aCorrelation is significant at the 0.01 level (2-tailed)
bCorrelation is significant at the 0.05 level (2-tailed)
Climatic Change (2011) 109:331–347 341
Table 6 Spearman’s correlation coefficient between number of fires, area burned and SPI values for
each district in Greece, during the period 1978–1997
Number of fires Area burned
Northern Greece 1978–1997 SPI6_October −0.394 (p=0.086) −0.421 (p=0.064)
SPI12_September −0.582a(p=0.007) −0.451b(p=0.046)
Southern Greece 1978–1997 SPI6_October −0.074 (p=0.757) −0.598a(p=0.005)
SPI12_September −0.138 (p=0.560) 0.098 ( p=0.682)
Western Greece 1978–1997 SPI6_October −0.632a(p=0.007) −0.459 (p=0.064)
SPI12_September −0.424 (p=0.090) −0.326 (p=0.202)
Eastern Greece 1978–1997 SPI6_October −0.147 (p=0.536) −0.233 (p=0.323)
SPI12_September −0.198 (p=0.402) −0.623a(p=0.003)
Central Greece 1978–1997 SPI6_October −0.382 (p=0.096) −0.555b(p=0.011)
SPI12_September −0.263 (p=0.262) −0.227 (p=0.335)
aCorrelation is significant at the 0.01 level (2-tailed)
bCorrelation is significant at the 0.05 level (2-tailed)
Considering the two sub-periods separately, the correlation was significant only
during 1978–1997 (Table 6). Specifically, during this sub-period there was a sig-
nificant correlation between annual drought (SPI12 values) and area burned in
Northern and Eastern Greece. In Southern Greece, the correlation was significant
only between the summer drought (SPI6) and area burned. In Central Greece there
was a significant correlation between SPI6 and area burned for both sub-periods
1961–1977 and 1978–1997. No other significant correlation for the first sub-period
(1961–1977) was found.
4 Discussion
Wildfire occurrence and area burned show a significant increasing trend in Greece
during the last decades (1961–1997). During 1978–1997 both the area burned and
the number of fires increased in comparison to the period 1961–1977: the mean
area burned almost quadrupled in Northern Greece and almost tripled in the other
districts, whereas the number of fires almost doubled in Northern and Western
Greece. An increase in the number of wildfires was also reported in Spain (European
Commission 2004;Pausas2004) and in the USA (Brown et al. 2004). Piñol et al.
(1998) did not find an increasing trend in area burned in the eastern coast of Spain,
during the period 1968–1994. Pausas (2004) reported that during the last three
decades area burned did not show a clear trend in the Valencia region (Spain).
Although there was a clearly increasing trend in annual drought episodes in
Greece during the period 1961–1997, the same was not true for the summer drought.
This suggests that the overall increase in drought was derived from the deficiency
of precipitation during the whole year and not only the summer months. This is in
accordance with Dalezios et al. (2000) who concluded that the rate of precipitation
decrease in Greece was higher during the rainy period (October to March),especially
in Northern Greece. Piñol et al. (1998) did not find a remarkable change in annual
precipitation in the eastern coast of Spain and suggest that the increase in drought
was caused by an increase in mean temperature and diurnal temperature range
which, in turn, resulted in a higher estimated potential evapotransiration. De Luís
342 Climatic Change (2011) 109:331–347
et al. (2001) report a decreasing trend in mean annual rainfall in Valencia, during
the period 1961–1990. For the same area, Pausas (2004) report a decreasing trend in
summer rainfall, whereas annual rainfall did not show a clear trend with time.
Comparing the two sub-periods, a significant increase in both annual (SPI12) and
summer (SPI6) drought episodes was observed in Northern and Western Greece,
during the sub-period 1978–1997. These districts are characterized by higher annual
precipitation and the lower mean temperatures, compared to the other districts of
Greece. Southern and Central Greece (with less annual precipitation and higher
mean temperatures) did not show any significant change in drought events during
the same period. Dalezios et al. (2000) concluded that the observed decrease in
precipitation in Greece during the period 1951–1990, was higher in regions which
are characterized by the highest amounts of precipitation and, gradually, reduced
southwards at the more arid regions.
There was a significant positive correlation between drought, wildfire occurrence
and area burned all over Greece, during the period 1961–1997. However, the number
of fires and area burned at Northern and Western Greece were significantly corre-
lated both to summer (SPI6 of October) and annual drought (SPI12 of September),
whereas at Southern and Central Greece they were correlated only to the summer
drought. This could be explained by the duration of the dry period required to cure
the forest fuels in order to lose sufficient moisture content and become ignitable.
It is expected that in the more humid and colder environments of Northern and
Western Greece, longer drought periods are needed in order to reduce the fuel
moisture content to ignitable levels (Castro et al. 2003; Pellizzaro et al. 2007). Thus, at
Southern and Central districts (that present the highest fire activity in Greece) only
the summer drought is a determinant factor for fire occurrence and severity. On the
other hand, in the more humid and less fire prone districts of Northern and Western
Greece, the annual drought becomes a determining factor for the fire load as well
(Clark 1989; Veblen et al. 2002; McKenzie et al. 2004; Verbesselt et al. 2007;Yebra
et al. 2008). Carvalho et al. (2010) also projected larger increases in fire activity in
the more humid and cold Northern and Central regions of Portugal as compared
to the drier Southern regions. Pausas (2004) found a weak correlation between the
number of fires and rainfall in Valencia and a stronger relation between area burned
and annual, but mostly summer, rainfall. He suggests that fire ignition in the region
is determined mainly by human activity. Piñol et al. (1998) reported a significant
correlation between the number of very high-risk days and the number of wildfires
and area burned in coastal Eastern Spain and attributed the increased fire activity
to the climatic change. A significant correlation between drought (as expressed by
SPI and PDSI Palmer-z indexes) and the number of fires and area burned was also
reported in the USA (Hall and Brown 2003).
Comparing the two sub-periods, drought was significantly correlated only to fire
occurrence during 1961–1977 (except Northern Greece, where the correlation was
significant during 1978–1997 too). On the contrary, during 1978–1997, drought was
significantly correlated only to the burned area (except Central Greece where the
correlation was significant during the period 1961–1977 as well). These apparently
contradictory results could be explained by analyzing the main fire causes, which
differ during the two sub-periods. During 1961–1977, the rural population of Greece
was high and most wildland fires were caused by negligence. The situation changed
after the late 1970s, when the urbanization of a large part of the rural population
Climatic Change (2011) 109:331–347 343
created pressure for forest land use change for urban development. This, in combi-
nation with other socioeconomic changes, resulted in a dramatic increase of arson
fires after 1977 (Dimitrakopoulos 1995).
Drought is a determinant factor of wildfire occurrence (Keetch and Byram 1968;
Balling et al. 1992a; Hall and Brown 2003). Considering that forest fuel ignition
mainly depends on the moisture content, it is expected that during drought periods
fuels become drier, more ignitable, and hence, more fires occur by negligence. Var-
ious predictive models of people-caused fire occurrence have found strong relation-
ships between dead fine fuel moisture and wildfire occurrence by negligence (Martell
et al. 1989; Todd and Kourtz 1992; Vega-Garcia et al. 1995; Wotton et al. 2003).
Although drought severely deteriorates burning conditions, wind velocity is the
environmental factor that primarily influences area burned by wildfires (Schroeder
and Buck 1970; Rothermel 1983; Balling et al. 1992a; Cheney et al. 1993; Millan
et al. 1998; Johnson and Miyanishi 2001;Nelson2002; Collins et al. 2006; Anderson
et al. 2007; Maingi and Henry 2007; Moritz et al. 2010). In the Mediterranean
countries, most large fires (and area burned) burn under a combination of strong
winds with drought and are caused by arsons (European Commission 2004;Moritz
et al. 2010). Hence, while drought significantly influences the incidence of fires
caused by negligence, arson fires and strong winds cause the most area burned in
the Mediterranean Basin (Millan et al. 1998; Dimitrakopoulos and Mitsopoulos 2006;
Leone et al. 2009).
In Northern Greece, where negligence remained the main cause of wildfires dur-
ing the sub-period 1978–1997, the correlation between drought and fire occurrence
was significant during this period too. On the other hand, in Central Greece where
most fires were traditionally set deliberately by shepherds for pasture management,
there was a strong correlation between drought and area burned during the first sub-
period (1961–1977), as well.
The future climatic scenarios, based on GCMs (Global Circulation Models),
predict increased fire occurrence, area burned, fire severity and fire danger in
general, as a result of the changing climatic conditions in several areas around the
world (Wotton and Flannigan 1993; Davis and Michaelsen 1995;Stocksetal.1998;
Flannigan et al. 2000; De Groot et al. 2003; Brown et al. 2004; Fried et al. 2004,2008).
On the other hand, many authors suggest that the increasing trend in fire incidence
and area burned constitutes not only the result of the climatic change, but also of
social, economic and land management changes (Vélez 1993;Pausas1999;Badia
et al. 2002). In Portugal and Spain after 1960, the industrial development resulted
in the decrease of rural population and the abandonment of pasture and forest
exploitation. This caused unnatural fuel accumulation in the wildlands (Rego 1992).
Pausas (2004) suggests that this can partly explain the increasing fire occurrence and
area burned during the 1970s. Badia et al. (2002) claim that this increase cannot be
attributed only to climatic conditions and individual human activity, but to land use
changes as well.
5 Conclusions
1. There was a statistically significant increasing trend and a positive correlation
between the measures of fire activity (number of fires and area burned) and
344 Climatic Change (2011) 109:331–347
the annual drought episodes in Greece, during the period 1961–1997 (37 years).
Summer drought episodes did not show any particular trend for the same period.
2. In the more humid and cold Northern and Western Greece fire activity was
influenced by both the annual and the summer drought events, whereas in the
drier and more hot Southern and Central Greece only the summer drought was
a determinant factor.
3. The average number of fires and area burned were significantly higher in Greece
during the sub-period 1978–1997 (when Greece entered a prolonged period of
drought, possibly due to climate change) compared to the previous sub-period
1961–1977. However, during 1978–1997, only in Northern and Western Greece
there was a significant increase in both the annual and summer drought episodes.
4. Comparing the two sub-periods, drought mostly affected fire occurrence during
the sub-period 1961–1977 and area burned during the sub-period 1978–1997.
Exceptionally, in Northern Greece there was a significant correlation between
summer and annual drought and fire occurrence during both sub-periods.
Conclusively, annual and summer drought had a significant impact on the number
of fires and area burned in Greece during 1961–1997. Climate change (in this case,
increased annual drought episodes) that is observed globally and, possibly, also in
Greece after 1977, seems to have increased the area burned by wildfires. This is
more profound in wet and cold regions (Northern and Western Greece), which
traditionally were characterized by fewer fire events and less area burned. It also
became obvious that drought has a progressively increasing impact on wildfire
activity in Greece. Thus, in the context of climatic change scenarios, more frequent
drought episodes are expected to deteriorate the wildland fire problem of Greece in
the future.
The incorporation of drought conditions into a predictive model of wildland fire
activity could be the subject of future research.
References
Anagnostopoulou GC (2003) Contribution to the drought analysis in Greece. PhD Dissertation,
Aristotle University of Thessaloniki, Thessaloniki, Greece (In Greek)
Anderson K, Reuter G, Flannigan MD (2007) Fire-growth modelling using meteorological data with
random and systematic perturbations. Int J Wildland Fire 16:174–182
Andersen T, Carstensen J, Hernandez-Garcia E, Duarte C (2009) Ecological thresholds and regime
shifts: approaches to identification. Trends Ecol Evol 24:49–57
Badia A, Saurí D, Cerdan R, Llurdés JC (2002) Causality and management of forest fires in Mediter-
ranean environments: an example from Catalonia. Environ Hazards 4:35–32
Balling RC, Meyer GA, Wells SG (1992a) Climate change in yellowstone national park: is the
drought-related risk of wildfires increasing? Climatic Change 22:35–45
Balling RC, Meyer GA, Wells SG (1992b) Relation of surface climate and burned area in Yellostone
National Park. Agr Forest Meteorol 60:285–293
Baltas E (2007) Spatial distribution of climatic indices in northern Greece. Meteorol Appl 14:69–78
Bhattacharya P (1994) Some aspects of change-point analysis. In: Carlstein E, Muller HG,
Siegmund D (eds) Change-point problems, IMS monograph 23. Institute of Mathematical Sta-
tistics, Hayward, pp 28–56
Bolle HJ (ed) (2003) Mediterranean climate—variability and trends. Springer, Berlin, p 372
Briffa RK, Jones DP, Hulme M (1994) Summer moisture variability across Europe, 1892–1991: an
analysis based on the Palmer drought severity index. Int J Climatol 14:475–506
Climatic Change (2011) 109:331–347 345
Brown JT, Hall LB, Westerling LA (2004) The impact of twenty-first century climate change on
wildland fire danger in the Western United States: an application perspective. Climatic Change
62:365–388
Carvalho A, Flannigan MD, Logan KA, Gowman LM, Miranda AI, Borrego C (2010) The impact
of spatial resolution on area burned and fire occurrence projections in Portugal under climate
change. Climatic Change 98:177–197
Castro FX, Tudela A, Sebastià MT (2003) Modeling moisture content in shrubs to predict fire risk in
Catalonia (Spain). Agr Forest Meteorol 116:49–59
Chen J, Gupta A (2000) Parametric statistical change point analysis. Birkhauser, Boston, p 196
Cheney NP, Gould JS, Catchpole WR (1993) The influence of fuel, weather and fire shape variables
on fire spread in grasslands. Int J Wildland Fire 3:31–44
Clark JS (1989) Effects of long-term water balances on fire regime, North-Western Minnesota. J Ecol
77(4):989–1004
Collacino M, Conte M (1993) Greenhouse effect and pressure patterns in the Mediterranean Basin.
IL NUOVO CIMENTO C 16:67–76
Collins BM, Omi PN, Chapman PL (2006) Regional relationships between climate and wildfire-
burned area in the Interior West, USA. Can J For Res 36(3):699–709
Csörgö M, Horváth L (1997) Limit theorems in change-point analysis. Wiley, Chichester
Dalezios RN, Loukas A, Vasiliadis L, Liakopoulos E (2000) Severity–Duration–Frequency Analysis
of Droughts and Wet Periods in Greece. Hydrol Sci J 45:751–768
Davis WF, Michaelsen J (1995) Sensitivity of fire regime in chaparral ecosystems to climate change.
In: Moreno MJ, Oechel CW (eds) Global change and Mediterranean-type ecosystems. Ecologi-
cal studies 117. Spinger, New York, pp 435–456
De Groot WJ, Bothwell PM, Carlsson DH, Logan KA (2003) Simulating the effects of future fire
regimes on Western Canadian boreal forests. J Veg Sci 14:355–364
De Luís M, García-Cano FM, Cortina J, Raventós J, González-Hildago CJ, Sánchez RJ (2001)
Climatic trends, disturbances and short-term vegetation dynamics in a Mediterranean shrubland.
For Ecol Manag 147:25–37
Dimitrakopoulos AP (1995) Analysis of the causes of forest fires in Greece. Options Mediterr 25:33–
40
Dimitrakopoulos AP (2005) Analysis of the wildland–urban interface problem of Greece. In:
Konijnendijk CC, Schipperijn J, Nilsson K (eds) Proceedings of the scientific conference
“Threats to urban forests and trees – The North and South perspective”. E.C., COST Action
E12: ‘URBAN FORESTS & TREES’, pp 157–164
Dimitrakopoulos AP, Bemmerzouk AM (2003) Predicting live herbaceous moisture content from a
seasonal drought index. Int J Biometeorol 47:73–79
Dimitrakopoulos AP, Mitsopoulos ID (2006) Global forest resources assessment 2005—report on
fires in the Mediterranean Region. FAO, Forestry Department, Fire Management Working
Paper 8. FAO, Rome, Italy. www.fao.org/forestry/site/fire-alerts/en,p38
Estrela JM, Penarrocha D, Millan M (2000) Multi-annual drought episodes in the Mediterranean
(Valencia region) from 1950-1996. A spatio-temporal analysis. Int J Climatol 20:1599–1618
European Commission (2004) Forest Fires in Europe—Report No. 4. Official Publication of the
European Commission S.P.I.04.124, p 49
Flannigan MD, Van Wagner CE (1991) Climate change and wildfire in Canada. Can J For Res 21:66–
72
Flannigan MD, Wotton BM (2001) Connections-Climate/weather and area burned. In: Johnson EA,
Miyanishi K (eds) Forest fires: behaviour and ecological effects. Academic, San Diego, pp 335–
357
Flannigan MD, Stocks JB, Wotton MB (2000) Climate change and forest fires. Sci Total Environ
262:221–229
Flannigan MD, Stocks JB, Weber M (2003) Fire regimes and climatic change in Canadian forests.
In: Veblen TT, Baker LW, Montenegro G et al (eds) Fire and climate change in temperate
ecosystems of the Western Americas. Springer, New York, pp 97–119
Flokas AA (1994) Climatology and meteorology, 2nd edn. Ziti, Thessaloniki p 480
Fried SJ, Torn SM, Mills E (2004) The impact of climate change on wildfire severity: a regional
forecast for Northern California. Climatic Change 64:169–191
Fried SJ, Gilless JK, Riley WJ, Moody TJ, de Bias CS, Hayhoe K, Moritz M, Stephens S, Torn M
(2008) Predicting the effect of climate change on wildfire behaviour and initial attack success.
Climatic Change 87(Suppl 1):251–264
346 Climatic Change (2011) 109:331–347
Hall BL, Brown JT (2003) A comparison of precipitation and drought indices related to fire activity
in the U. S. In: Proceedings of the 5th symposium on fire and forest meteorology. American
Meteorological Society, Orlando FL
Johnson E, Miyanishi K (2001) Forest fires. Behaviour and ecological effects. Academic, San Diego,
p 600
Keetch JJ, Byram GM (1968) A drought index for forest fire control. U.S. Department of Agri-
culture, Forest Service, Southeastern Forest and Range Experiment Station, Research paper
SE-38
Leone V, Lovreglio R, Martín M, Martínez J, Vilar L (2009) Human factors of fire occurrence in
the Mediterranean. In: Chuvieco E (ed) Earth observation of wildland fires in mediterranean
ecosystems. Springer, Berlin, pp 149–170
Livada I, Assimakopoulos VD (2007) Spatial and temporal analysis of drought in Greece using the
Standardized Precipitation Index (SPI). Theor Appl Climatol 89:143–153
Maheras P, Patrikas I, Karakostas T, Anagnostopoulou C (2000) Automatic classification of circula-
tion types in Greece: methodology, description, frequency, variability and trend analysis. Theor
Appl Climatol 67:205–223
Maheras P, Tolika K, Anagnostopoulou C, Vafiadis M, Patrikas I, Flokas H (2004) On the rela-
tionships between circulation types and changes in rainfall variability in Greece. Int J Climatol
24:1695–1712
Maingi JK, Henry MC (2007) Factors influencing wildfire occurrence and distribution in eastern
Kentucky, USA. Int J Wildland Fire 16:23–33
Martell DL, Bevilacqua E, Stocks BJ (1989) Modelling seasonal variation in daily people-caused
forest fire occurrence. Can J For Res 19(12):1555–1563
McKee TB, Doesken NJ, Kleist J (1993) The relationship of drought frequency and duration to time
scale. In: Proceedings of the 8th Conference on Applied Climatology. American Meteorological
Society, Boston, pp 179–184
McKenzie D, Gedalof Z, Peterson DL, Mote P (2004) Climatic change, wildfire, and conservation.
Conserv Biol 18:890–902
Millán M, Estrela M, Badenas C (1998) Meteorological processes relevant to forest fire dynamics on
the Spanish mediterranean coast. J Appl Meteorol 37:83–100
Milliken GA, Johnson DE (2001) Analysis of messy data. Volume III: analysis of covariance.
Chapman and Hall/CRC, Boca Raton
Moritz M, Moody T, Krawchuk M, Hughes M, Hall A (2010) Spatial variation in extreme winds
predicts large wildfire locations in chaparral ecosystems. Geophys Res Lett 37:L04801
Mouillot F, Rambal S, Joffre R (2002) Simulating climate change impacts on fire frequency and
vegetation dynamics in a Mediterranean-type ecosystem. Glob Chang Biol 8:423–437
Nelson RM Jr (2002) An effective wind speed for models of fire spread. Int J Wildland Fire 11:153–
161
Norusis MJ (1997) SPSS professional statistics. SPSS, Chicago, p 376
Pausas JG (1999) The Response of plant functional types to changes in the fire regime in
Mediterranean ecosystems. A simulation approach. J Veg Sci 10:717–722
Pausas JG (2004) Changes in fire and climate in the Eastern Iberian Peninsula (Mediterranean
Basin). Climatic Change 63:337–350
Pellizzaro G Duce P, Ventura A, Zara P (2007) Seasonal variations of live moisture content and
ignitability in shrubs of the Mediterranean Basin. Int J Wildland Fire 16:633–641
Piñol J, Terradas J, Lloret F (1998) Climate warming, wildfire hazard and wildfire occurrence in
coastal Eastern Spain. Climatic Change 38:345–357
Pnevmatikos JD, Katsoulis BD (2006) The changing rainfall regime in Greece and its impact on
climatological means. Meteorol Appl 13:331–345
Radionov S (2004) A sequential algorithm for testing climate regime shifts. Geophys Res Lett
31:L09204
Rego FC (1992) Land use changes and wildfires. In: Teller A, Mathy P, Jeffers JN (eds) Response of
forest fires to environmental change. Elsevier, London, pp 367–373
Repapis CC, Philandras CM, Kalabokas PD, Zanis P, Zerefos CS (2007) Is the last years abrupt
warming in the National Observatory of Athens records a climate change manifestation? Global
Nest J 9(2):107–116
Rothermel RC (1983) How to predict the spread and intensity of forest and range fires. U.S.
Department of Agriculture, Forest service, Intermountain forest and range experiment station,
Gen. Tech. Rep. INT – 143. Ogden, Utah, p 25
Climatic Change (2011) 109:331–347 347
Schlobohm PM, Brown JT (2001) Fire danger and the standardized precipitation index. In: Proceed-
ings of the 4th Symposium on Fire and Forest Meteorology. American Meteorological Society,
Orlando. pp. 220–222
Schoenberg PF, Peng R, Huang Z, Rundel P (1991) Detection of non-linearities in the dependence
of burn area on fuel age and climatic variables. Int J Wildland Fire 12:1–6
Schroeder MJ, Buck CC (1970) Fire Weather—a guide for application of meteorological information
to forest fire control operations. USDA, Forest Service, Agriculture Handbook 360. Washington,
p 229
Shatfield C (1995) The analysis of time series: an introduction, 4th edn. Chapman and Hall, London,
p 322
Stocks BJ, Fosberg MA, Lynham TJ, Mearns L, Wotton BM, Yang Q, Jin J-Z, Lawrence K, Hartley
GR, Mason JA, McKenney DW (1998) Climate change and forest fire potential in Russian and
Canadian boreal forests. Climatic Change 38:1–13
Taylor W (2000) Change-Point Analyzer 2.0 shareware program, Taylor Enterprises, Libertyville,
Illinois. Web: http://www.variation.com/cpa
Todd JB, Kourtz PH (1992) Predicting the daily occurrence of people-caused forest fires. Canadian
forest service, petawawa national forestry institute, Information Report PI-X-103, Chalk River,
Ontario,p16
Torn MS, Fried JS (1992) Predicting the impacts of global warming on wildland fire. Climatic Change
21:257–274
Vázquez A, Moreno MJ (1993) Sensitivity of fire occurrence to meteorological variables in Mediter-
ranean and Atlantic areas of Spain. Landsc Urban Plan 24:129–142
Veblen TT, Baker WL, Montenegro G, Swetnam TW (2002) Fire and climatic change in temperate
ecosystems of the Western Americas. Springer, New York, p 464
Vega-Garcia C, Woodard PM, Titus SJ, Adamowicz WL, Lee BS (1995) A logit model for predicting
the daily occurrence of human caused forest fires. Int J Wildland Fire 5(2):101–112
Vélez R (1993) High intensity forest fires in the Mediterranean Basin: natural and socioeconomic
causes. Disaster Manage 5:16–20
Verbesselt J, Somers B, Lhermitte S, Jonckheere I, van Aardt J, Coppin P (2007) Monitoring
herbaceous fuel moisture content with SPOT VEGETATION time-series for fire risk prediction
in savanna ecosystems. Remote Sens Environ 108(4):357–368
Wilhite AD, Glantz NH (1985) Understanding the drought phenomenon: the role of definitions.
Water Int 10:111–120
Wotton BM, Flannigan MD (1993) Length of the fire season in a changing climate. For Chron 69:187–
192
Wotton BM, Martell DL, Logan KA (2003) Climate change and people-caused forest fire occurrence
in Ontario. Climatic Change 60:275–295
Yebra M, Chuvieco E, Riano D (2008) Estimation of live fuel moisture content from MODIS images
for fire risk assessment. Agr Forest Meteorol 148 (4):523–536
