Assessment of Wildfire Risk in Lebanon Using
Geographic Object-Based Image Analysis
George Mitri, Mireille Jazi, and David McWethy
Abstract
During the past decade, Lebanon has experienced a large number
of severe wildfires that have had significant social and ecological
consequences. In this context, the assessment of wildfire risk is
important to support planning of fire prevention measures and
risk mitigation. The purpose of this study was to assess the spatial
distribution of wildfire risk in Lebanon. The objectives were to iden-
tify and map (a) wildfire hazard, (b)wildfire vulnerability, and (c)
wildfire risk. We developed a model using geospatial biophysical
and climatic data and Geographic Object-Based Image Analysis
(
GEOBIA
). Development of the wildfire hazard map included clas-
sification of forest fuel type, combustibility, and fire spread whereas
the vulnerability map included classification of demographic vul-
nerability (i.e., boundary, occupation and scatter indicators) and
forest vulnerability (i.e., environmental and replacement values).
The resulting geospatial map of wildfire risk provided important
information for potential use in fire risk management.
Introduction
Wildfire is an important disturbance process that has shaped
the structure, composition, and function of Mediterranean for-
ests for centuries (Liu
et al
, 2010). However, in recent decades,
increases in wildfire occurrence, size and intensity are having
profound and perhaps unprecedented impacts on Mediter-
ranean ecosystems (Chuvieco
et al.
, 2010; Dimitrakopoulos
et
al.
, 2010; FAO, 2013; Liu
et al.
, 2010).
Like other Mediterranean countries affected by wildfires,
Lebanon’s forested ecosystems have been severely impacted
the increase in the number and size of fires (Salloum
et al.
,
2013). Research has shown that recent fire occurrence in Leba-
non was positively correlated with mean monthly tempera-
tures and negatively correlated with mean monthly precipita-
tion (Salloum and Mitri, 2014). Increasing fire occurrence was
also observed in association with high maximum temperatures
and long dry seasons. In addition, average length of the fire
season extended over 146.6 days and was negatively correlat-
ed with mean annual precipitation (Salloum and Mitri, 2014).
The social and ecological impacts resulting from increasing
number of large fires in Lebanon is a National concern (Mitri
and Gitas, 2011). Most recently, a National Strategy for Forest
Fire Management (Decision No. 52/2009) was endorsed by
the Lebanese Council of Ministers (MOE/AFDC, 2009). This
strategy highlighted the need to reduce risk of intense and
frequent forest fires while allowing for fires that are socially,
economically, and ecologically sustainable. Important compo-
nents of this strategy include: (a) developing improved mod-
els of wildfire risk assessment, and (b) outlining strategies for
responding to and adapting to projections of future climatic
conditions that will likely further promote large and intense
fires in the Mediterranean. These tasks will be critically nec-
essary for implementing effective fire prevention policies.
Research demonstrates that fire risk models represent char-
acteristics of fire behavior under specific climatic conditions
(Sharples, 2009). Climate models have been used to assess
fire risk, anticipating potentially dangerous conditions (Liu
et
al.
, 2010). Several fire danger rating systems are used around
the world, and one of the most widely employed indices is
the Fire Weather Index (
FWI
) used by the Canadian Forest Fire
Danger Rating System (Dimitrakopoulos
et al.
, 2010). The
FWI
uses surface daily temperature, precipitation, relative humid-
ity, and wind speed to represent fuel moisture changes and
their effects on fire behavior (Dimitrakopoulos
et al.
, 2010;
Karali
et al.
, 2012). The Keetch-Byram Drought Index (
KBDI
) is
another fire danger index, which is a part of the National Fire
Danger Rating System (
NFDRS
) in the United States (Keetch
and Byram, 1968; Melton, 1989).
KBDI
is a cumulative esti-
mate of fire potential based on meteorological input param-
eters and an empirical approximation for moisture depletion
in the upper soil and surface litter levels (Keetch and Byram,
1968).
KBDI
uses temperature and precipitation to estimate fire
potential (Janis
et al.
, 2002).
Many of these models rely solely on climatic variables
even though biophysical factors such as fuel type and to-
pography play a major role in determining fire spread and
behavior. The complex interactions between these compo-
nents directly influence fire spread and flammability of fuels.
Moreover, demographic variables are critical for predicting
risk and hazard of fires on communities (Bühler
et al.
, 2013;
Chuvieco
et al.
, 2010). To improve the assessment of both
social and ecological aspects of fire risk it is important to
incorporate spatial distributions of populations and settle-
ments and the presence and distribution of protected areas
and biodiversity hotspots.
Overall, fire risk is a combination of likelihood, intensity,
and effect (Miller and Ager, 2013). Understanding the com-
ponents that comprise wildfire risk is important for predict-
ing when and where a fire is more prone to occur and spread
and the degree to which it will have negative impacts on
communities and ecosystems. Our study explicitly estimates
biophysical and climatic spatial characteristics for fire occur-
rence (Pinol
et al.
, 1998), expressed here as “fire hazard,” and
the potential damage that it may cause, expressed here as “fire
vulnerability” (Chuvieco
et al.
, 2010).
In this context, this study evaluated the spatial distribution
of wildfire risk in Lebanon. The specific objectives were to
identify and map (a) wildfire hazard, (b) wildfire vulnerabil-
ity, and (c) wildfire risk.
To address these objectives we used geospatial datasets
that included biophysical and climatic data. Because we used
multi-resolution data in addition to multi-source bio-physical
and climatic data, it was necessary to develop new methods
George Mitri and Mireille Jazi are with the Institute of the
Environment, University of Balamand, Kelhat-El Koura,
Lebanon (
).
David McWethy is with the Department of Earth Sciences,
Montana State University, Montana.
Photogrammetric Engineering & Remote Sensing
Vol. 81, No. 6, June 2015, pp. 499–506.
0099-1112/15/499–506
© 2015 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.81.6.499
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
June 2015
499
