2004 
  
ird — 
Data 
thout 
ence, 
Tigh- 
nials. 
. 69, 
gery. 
aska, 
'ation 
vithin 
hens, 
dt 
nual. 
cation 
joints, 
67, n° 
'nsing 
journal 
Validated Spectral Angle Mapper Algorithm for Geological Mapping: Comparative Study 
between Quickbird and Landsat-TM 
pis 1 . 
G.Girouard . A.Bannari 
' A. El Harti ” and A. Desrochers ? 
| Remote Sensing and Geomatics of the Environment Laboratory 
Department of Geography, University of Ottawa, Ottawa-Carleton GeoScience Center 
P.O. Box 450, Ottawa (Ontario) KIN 6N5 
Tel: (613) 562-5800 (ext. 1042), Fax: (613) 562-5145 
E-mail: abannari@uottawa.ca 
? Remote Sensing and GIS Applied to Geosciences and the Environment 
Faculty of Science and Techniques Beni-Mellal, Morocco, University of Cadi Ayyad 
? Department of Earth Science, University of Ottawa 
Ottawa-Carleton GeoScience Center 
KEY WORDS - Geology, Mapping, Exploration, Landsat, Quickbird, DEM, GIS 
ABSTRACT: 
The aim of this study is to validate the Spectral Angle Mapper (SAM) algorithm for geological mapping in Central Jebilet 
Morocco and compare the results between high and medium spatial resolution sensors, such as Quickbird and Landsat TM respectively. 
The geology of the study area is dominated by the Sahrlef schist, which is intruded by numerous bodies of acidic and basic magmatic 
rocks and lens-shaped pyrrhotite stratabound massive deposits. The results show that SAM of TM data can provide mineralogical maps 
that compare favorably with ground truth and known surface geology maps. Even do, Quickbird has a high spatial resolution compared 
to TM; its data did not provide good results for SAM because of the low spectral resolution. 
1. Introduction 
Multispectral remote sensing techniques have been widely used 
in the past decades to discriminate different materials based on 
the dissimilarity that exist among their spectral properties (Hunt 
et al, 1971). Geological applications could greatly take 
advantage of this technology because it allows observing and 
mapping the surface of the Earth over large areas, but the 
generally low spectral resolution of multispectral sensors 
obstruct geological research. Digital image classification is a 
technique that is widely used to classify each individual pixel in 
an image based on the spectral information in order to create 
lithological or geological maps. Various classification 
algorithms have been used in the past decades in a variety of 
applications for mapping: forestry, agriculture, land use, 
geology etc. The main disadvantage of these algorithms is that 
each pixel in the image is compared to the training site 
signatures identified by the analyst and labeled as the class it 
most closely "resembles" digitally. Spectral Angle Mapper 
(SAM) is different from these standard classification methods 
because it compares each pixel in the image with every 
endmember for each class and assigns a ponderation value 
between 0 (low resemblance) and 1 (high resemblance). 
Endmembers can be taken directly from the image or from 
signatures measured directly in the field or laboratory. The 
main disadvantage of these algorithms, including SAM, is that 
they do not consider the sub-pixel value. The spectral mixture 
problem can become problematic because most of the Earth's 
surface is heterogeneous, thus for a medium space resolution 
sensor like Landsat TM, false assumptions can be made. 
Nonetheless, SAM has been used successfully in the past for 
geological mapping and for identifying potential mineral 
exploration sites with the use of the *USGS Spectral Library" 
599 
as reference spectrum (Crosta ef al., 1998). McCubbin et al. 
(1998) used SAM to map alteration minerals associated with 
mine waste and Van der Meer e/ al. (1997) used this technique 
for mapping different ophiolite lithologies in Cyprus. The 
purpose of this paper is to discuss and to validate the results for 
the geological mapping of Central Jebilet in Morocco using 
SAM algorithm with two different optical sensors such as 
Landsat TM and Quickbird. These two sensors possess spatial 
and spectral characteristics that are opposite to each other. 
Landsat TM has a medium spatial resolution (30 x 30 meters) 
versus the high spatial resolution (2.5 x 2.5 meters) for 
Quickbird. Nonetheless, the spectral dimension of TM is richer 
for geological mapping than Quickbird, because it has 6 bands 
that cover the spectrum between 0.45-2.35 um in opposition to 
4 bands for Quickbird that only cover the 0.45-0.90 um range. 
2. Materials and Methods 
2.1. Geological Setting 
The study area is located in the Paleozoic terrain of central 
Jebilet in Morocco, which is about 8 Km north of Marrakech 
(8°20'W - 7°40'W; 31°40'N - 32°N). The topography of the area 
is formed of small-elongated hills, which vary from 350 to 750 
meters in elevation. The total area covered by this study is 100 
Km/ and is characterized by a lithological diversity, semi-desert 
environment, significant mineral potential and consistent 
geological setting. The geology of central Jebilet is composed 
of four main rock units of which the Sahrlef schist is the 
dominant rock unit. It is intensively deformed and intruded by 
numerous acidic and basic magmatic bodies. The economic