198
Regional Environmental Changes: Application to Soil Erosion Modeling
population and urban industrial pollution. Such changes are coupled with a global
warming due to increasing levels of carbon dioxide and other greenhouse gases.
Remote sensing technology through a repetitive coverage of data is a good tracer
for the evolution of ecosystems at regional and global scales. The Landsat MSS
data acquired since 1972 represent an important basis of temporal, spectral and
spatial information for the study of global change phenomena in terrestrial
ecosystems. In this paper, which is part of a current research project, we illustrate
the usefulness of multitemporal remotely sensed data in deciphering the land
cover changes of the Mediterranean region. In this study, the normalized
vegetation index (NDVI) is used as an indicator for land cover changes. The area
under investigation is part of the Eastern Mediterranean Basin and concerns
maritime Syria located North-West of this country and with a typical
Mediterranean climate, characterized by dry summers and moist winters (Khreim
and Lacaze, 1991). The studied zone precisely pertains to the humid to sub-humid
bioclimatic stage according to the Emberger climatogram applied to Lebanon by
Abi-Saleh (1978). It constitutes the more vegetative region in Syria (Quezel and
Barbero, 1985).
2 M UL TITEMPORAL NOVI ANALYSES
The techniques used in land cover change detection are based on the comparison
of remotely sensed data and map data, or the comparison of remotely sensed data
taken at two or more different times. Numerous methods are available, each one
offering its own advantages and drawbacks: univariate image differencing, image
regression, image rationing, vegetation index differencing, principal component
analysis, post-classification comparison, direct multi date classification, change
vector analyses ... (Singh, 1989). Among these methods, we chose to consider the
vegetation index for multi temporal comparison. Since vegetation is considered to
be the functional equivalent of terrestrial ecosystems (Graetz, 1990), then
vegetation is a prime indicator of the type of ecosystem present (i.e. grasslands,
forests, deserts). The characterization of vegetation generally offers the observer a
great amount of information about its geographical location, current ecological
conditions and may even yield information on previous changes that have
occurred in a given region. The vegetation index which is the ratio between the
reflected radiation in near-infrared (NIR) and red (R) bands measured by the
satellite sensor, is directly linked to the density and state of green vegetation (LAI:
Leaf Area Index). The index used in this study is the normalized difference
vegetation index: (NDVI) = NIR - R / NIR + R (Rouse et al. 1974 ) as it can be
easily calculated and allows the relief effects of illumination to be reduced. Land
cover changes over time may be distinguished through the analysis of
multitemporal vegetation index imagery.
3 DATA PROCESSING
The data we used are 800x800 pixel extracts from three Multi Spectral Sensor
(MSS) scenes of Landsat 1 and 2 satellites, acquired on March 21, 1977
(Landsat 2), April 5, 1986 and March 5, 1993. Produced bX ESA (European
Space Agency, Fucino, Italy), these scenes allow a 2200 km 2 coverage of the
region. The absolute radiometric correction can be accomplished by converting
the different grey levels to physical values of radiance and then to apparent
reflectance at the sensor.
Regional Environmental Changes: Application to Soil Erosion Modeling
population and urban industrial pollution. Such changes are coupled with a global
warming due to increasing levels of carbon dioxide and other greenhouse gases.
Remote sensing technology through a repetitive coverage of data is a good tracer
for the evolution of ecosystems at regional and global scales. The Landsat MSS
data acquired since 1972 represent an important basis of temporal, spectral and
spatial information for the study of global change phenomena in terrestrial
ecosystems. In this paper, which is part of a current research project, we illustrate
the usefulness of multitemporal remotely sensed data in deciphering the land
cover changes of the Mediterranean region. In this study, the normalized
vegetation index (NDVI) is used as an indicator for land cover changes. The area
under investigation is part of the Eastern Mediterranean Basin and concerns
maritime Syria located North-West of this country and with a typical
Mediterranean climate, characterized by dry summers and moist winters (Khreim
and Lacaze, 1991). The studied zone precisely pertains to the humid to sub-humid
bioclimatic stage according to the Emberger climatogram applied to Lebanon by
Abi-Saleh (1978). It constitutes the more vegetative region in Syria (Quezel and
Barbero, 1985).
2 M UL TITEMPORAL NOVI ANALYSES
The techniques used in land cover change detection are based on the comparison
of remotely sensed data and map data, or the comparison of remotely sensed data
taken at two or more different times. Numerous methods are available, each one
offering its own advantages and drawbacks: univariate image differencing, image
regression, image rationing, vegetation index differencing, principal component
analysis, post-classification comparison, direct multi date classification, change
vector analyses ... (Singh, 1989). Among these methods, we chose to consider the
vegetation index for multi temporal comparison. Since vegetation is considered to
be the functional equivalent of terrestrial ecosystems (Graetz, 1990), then
vegetation is a prime indicator of the type of ecosystem present (i.e. grasslands,
forests, deserts). The characterization of vegetation generally offers the observer a
great amount of information about its geographical location, current ecological
conditions and may even yield information on previous changes that have
occurred in a given region. The vegetation index which is the ratio between the
reflected radiation in near-infrared (NIR) and red (R) bands measured by the
satellite sensor, is directly linked to the density and state of green vegetation (LAI:
Leaf Area Index). The index used in this study is the normalized difference
vegetation index: (NDVI) = NIR - R / NIR + R (Rouse et al. 1974 ) as it can be
easily calculated and allows the relief effects of illumination to be reduced. Land
cover changes over time may be distinguished through the analysis of
multitemporal vegetation index imagery.
3 DATA PROCESSING
The data we used are 800x800 pixel extracts from three Multi Spectral Sensor
(MSS) scenes of Landsat 1 and 2 satellites, acquired on March 21, 1977
(Landsat 2), April 5, 1986 and March 5, 1993. Produced bX ESA (European
Space Agency, Fucino, Italy), these scenes allow a 2200 km 2 coverage of the
region. The absolute radiometric correction can be accomplished by converting
the different grey levels to physical values of radiance and then to apparent
reflectance at the sensor.