PE&RS June 2015 - page 502

of terrain derived from the DEM of Lebanon and following
the cross-mapping approach as per Table 2. Fire spread maps
included the classes: low, moderate, high, and very high.
Following fire spread classification, KBDI was selected for
the development of Lebanon’s climate-based wildfire poten-
tial index (Mitri et al. , 2014). The drought index ranges from
0 to 800, where a drought index of 0 represents saturated
soil (no moisture depletion), and an index of 800 represents
extreme or absolute dry conditions.
The spatial datasets of monthly maximum temperature,
monthly precipitation, and mean annual rainfall (1-km spatial
resolution) were employed for the calculation of KBDI using the
Model Maker module in ERDAS Imagine® software (Janis et al. ,
2002; Liu et al., 2010). The monthly values were first convert-
ed to daily values by assuming that no daily variations exist
within a specific month (Liu et al., 2010). The calculations
consisted of three steps (Liu et al., 2010): (1) values were cal-
culated for each day over the average period of the dataset, (2)
the same calculation was made starting from 01 January but
using KBDI on 31 December as the initial value (the steps were
repeated 30 times until the difference between two adjacent
years became negligible), and (3) the resulting daily maps were
converted into average monthly KBDI images, and subsequently
into an annual KBDI image for analysis. The different fire po-
tential classes, namely very low, low, moderate, high, and very
high, were thus based on the ranges of the annual KBDI values.
The final classes for the fire hazard map (low, moderate,
and high, and very high) involved the combined classification
of the annual KBDI and fire spread maps and contextual infor-
mation following the same cross-mapping approach of “class
1” and “class 2” as in the previous case (Table 2).
Mapping Wildfire Vulnerability
Wildfire vulnerability is defined by the extent of loss and/
or damage that may affect people, goods, services, and the
environment after a fire event. As a result, the overall wildfire
vulnerability of Lebanon was primarily evaluated based on
demographic and forest type vulnerabilities. Both were classi-
fied at the super-object level.
First, the demographic vulnerability was defined as the
population’s sensitivity level regarding a possible fire event.
Demographic vulnerability was evaluated with three differ-
ent indicators, (1) occupation, (2) boundary, and (3) scat-
ter indicators for each cadastral unit (i.e., village). First, all
urban areas and settlements at the cadastral unit level were
identified and mapped using information extracted from the
attribute tables of the urban and administrative maps of Leba-
non. Second, forest vulnerability was defined as the level of
losses that forest systems may suffer in the event of a wildfire.
It is calculated based on the environmental value index and
the replacement value index. The calculation of the different
indicators is described as follows.
Occupation Indicator
The Occupation Indicator estimated the total density of con-
structions/homes present in a fire hazard area (i.e., a higher
occupation implies a higher demographic vulnerability). Four
intervals and categories of the Occupation Indicator were de-
termined (Table 3). The range of values obtained for the coun-
try (maximum occupation-minimum occupation) was equally
divided into three groups and classified for all cadastral units.
Boundary Indicator
The Boundary Indicator was defined as the sum of the length
(meters) of a common perimeter between urban and fire
hazard area, representing the degree of contact between urban
and fire hazard areas. High values indicate a higher threat to
human lives. Four intervals and categories of the Boundary
Indicator were determined (Table 4). The range of values ob-
tained for the country (maximum value-minimum value) was
divided into three groups and classified for all cadastral units.
Scatter Indicator
The Scatter Indicator estimates dispersion of residences in fire
hazard areas. It was calculated as the sum distance between
the centers of the settlements located in a fire hazard area
over the number of settlements. A high scatter meant a high
distance between homes, and therefore, increased resources/
time needed to protect people and their individual homes.
Four intervals and categories of the Scatter Indicator were
determined (Table 5). The range of values obtained for the coun-
try (maximum value-minimum value) has been analyzed and
divided into three groups and classified for all cadastral units.
Environmental Value Index
The environmental value index integrates the ecological and
recreational values of forest systems, and is identified based
on the value that society gives to their ecosystems. In Leba-
non, these areas generally correspond to protected areas such
as forests (usually by a ministerial decision) or forest natural
reserves (created by a National law). Thus, to determine the
environmental value of forest systems that are present in each
of Lebanon’s cadastral units (i.e., villages), the occurrence
of protected areas was taken into consideration. The inter-
vals which determined the environmental value index were
defined as follows: low (no occurrence of protected areas),
moderate (either existence of protected valleys or existence of
T able 2. Combined Use of C ombustibility and
Slope for Mapping Wildfire Spread
Combustibility
(class 1)\
slope (class 2)
Low
(0-10%)
Medium
(10-30%)
High
(30-50%)
Very high
(>50%)
No Combustibility No risk No risk No risk No risk
Low Low Low Moderate High
Medium Low Moderate High High
High Moderate Moderate High Very High
Very high Moderate High Very High Very High
Table 3. Occupation Intervals
Occupation Interval (ha) Occupation category Index
0 Zero 0
]0-57] Low 1
]57-116] Moderate 2
>116 High 3
Table 4. Boundary Intervals
Boundary Interval (m) Boundary category Index
0 Zero 0
]0-9716] Low 1
]9716-19432] Moderate 2
>19432 High 3
T able 5. Scatter Indicator
Scatter Interval (m) Scatter category Index
0 Zero 0
]0-3232] Low 1
]3232-6464] Moderate 2
>6464 High 3
502 June 2015 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING