Combined Uses of Supervised Classification and Normalized Difference Vegetation Index Techniques to Monitor Land Degradation in the Saloum Saline Estuary System - The Land/Ocean Interactions in the Coastal Zone of West and Central Africa - page 50

Study Area

sensed data obtained from Earth-orbiting satellites because
of repetitive coverage at short intervals and consistent
image quality (Anderson 1977 ; Ingram et al. 1981 ; Nelson
1983 ). Change detection using images has been traditionally
performed by comparing the classification of multitemporal
data sets or by image processing techniques such as dif-
ferencing and rationing. Change detection is useful in such
diverse applications as land-use change analysis, monitor-
ing of shifting cultivation, assessment of deforestation,
changes in vegetation phenology, seasonal changes in pas-
ture production, damage assessment, crop stress detection,
disaster monitoring snow-melt measurements, daylight
analysis of thermal characteristics, and other environmental
changes (Singh 1989 ).
In change detection processes (Singh 1989 ; Coppin and
Bauer 1994 ; Lu et al. 2003 , 2004a , b ; Coppin et al. 2004 ;Pu
et al. 2008 ; Pan et al. 2011 ; Datta and Deb 2012 ; Petropoulos
et al. 2012 ), time series images acquired from different dates
are compared to analyze the spectral difference, caused by
land-use/land-cover change (LULC) over time while trying
to normalize other conditions to similar levels during that
period. Therefore, it is necessary to confirm the estuary
dynamic for mapping and monitoring the land-cover change
with different techniques. Satellite remote sensing has been
widely applied and recognized as a powerful and effective
tool for detecting land-use and land-cover changes. How-
ever, according to the change's indices adopted and the
methods of detection applied, results obtained show signif-
icant differences that can be evaluated both quantitatively
(importance of the changes over time) and qualitatively
(types of changes observed). Works of Smits et al. ( 1999 )
and Coppin et al. ( 2004 ) identified ten types of detection
methods that are based on different techniques of image
processing, including image subtraction, crossing classifi-
cations, principal component analysis (PCA: statistical
analysis multivariate), vector calculating change, or neural
networks.
In this study, two change detection techniques are eval-
uated: a classification method and a normalized remote
sensing technique—normal difference vegetation index
(NDVI) differencing method, focusing on a comparison
between the two techniques and also on the determination
of the threshold of the NDVI differencing method. Both
techniques are common and effective in change detection of
LULC (Gong and Howarth 1992 ; Kontoes et al. 1993 ; Fo-
ody 2004 ; San Miguel-Ayanz and Biging 1997 ; Aplin et al.
1999 ; Stuckens et al. 2000 ; Franklin et al. 2002 ; Pal and
Mather 2004 ; Gallego 2004 ; Lu et al. 2004a , b ; Pu et al.
2008 ; Datta and Deb 2012 ). Classification and NDVI dif-
ferencing change detection methods were adopted in this
study

The Saloum estuary system, located approximately between
longitudes 1401 0 and 1656 0 W and latitudes 1331 0 and
1457 0 N (Fig. 1 ), shrank after the last pluvial episode in
around 10,000 BC and represents one of the largest African
reverse estuaries. It consists of an extensive network of fossil,
dried secondary channels (so-called thalwegs) stretching
north and eastward. The terrain in the study area is generally
flat with altitudes ranging from below sea level in the estu-
arine zone to about 40 m above mean sea level (a.m.s.l.)
inland; the longitudinal slope of the river course is corre-
spondingly low as well as the shallow bathymetry of the
river. The climate is Sudano-Sahelian type with a long dry
season from November to June and a 4-month rainy season
from July to October. The regional annual precipitation,
which is the main source of freshwater recharge to the
superficial aquifer, increases southward from 600 to
1,000 mm. The average temperature is 28-29 C, and the
average annual evaporation varies from 1,500 to 2,500 mm
(source: meteorological data). The geomorphology consists
of a gently sloping plain that extends toward the coast,
ranging in elevation from 0 m in the estuary system to 40 m
a.m.s.l. inland (Barusseau et al. 1985 ;Diop 1986 ). Sand dune
deposits occur near the coast with an altitude of 1 m in the
northern part and between 2 and 8 m a.m.s.l. in the southern
part of the region. The hydrologic system of the region is
characterized by the river Saloum, its two tributaries (Ban-
diala and Diomboss), and numerous small streams locally
called ''bolons.'' Downstream, it forms a large low-lying
estuary bearing tidal wetlands, a mangrove ecosystem, and
vast areas of denuded saline soils called ''tan'' locally.

Methodology

Landsat data were selected to generate time series of land-
cover changes in Saloum estuary. The regular revisit times
and spatial resolution of the Landsat mission are well suited
for regional, national, and global land-use changes. Four
images were selected for this study dated October 17, 1984,
October 31, 1992, November 01, 1999, and November 26,
2010, respectively. Accordingly, the study period covered
about the last three decades.
The methodology applied in this work consists of three
major steps: (1) collect and clean training samples; (2)
automatic classification for LULC (land-cover-land-cover)
mapping; and (3) NDVI differencing analysis. The images
were selected with respect to resolution, number of bands,
and season. Although the four scenes were already geore-
ferenced to the UTM Zone 28 North and WGS 84 projec-
tion, they were geomatching. The outputs of the second and

to analyze land-cover changes associated with

salinization.