175
dendritic drainage system to the south combined with
its typical texture and physical resistance of the
surface materials clearly indicates the presence of
the Southern Mountain zone (Figure:5). The Bandung
zone, on the other hand, occupies most of the valleys
surrounding the recent volcanoes and volcanic
materials, This zone is occupied by Tertiary
sedimentary rocks with the surface topography showing
bedding, folds, faults and some fractures. The water
divide lying along the center of the island coincides
with the radial drainage pattern of the active
volcanoes. The water divide divides the flowage of
the water to the South and to the North except in the
SW section which shows a southwesterly water flowage
and in the NE-section where the water north of the
water divide flows to the South instead of to the
North (Figure:6). The occurence of these flowage
direction anomalies may be related to geologic
disturbance in the SW (Sukabumi) section and the NE
(Leirbang) section where the image processing and
ground investigation latter were concentrated.
Figure 6: Regional drainge system, west Java, from
Landsat imagery.
The investigation in Sukabumi and the Leirbang area
was particularly directed toward determining the
distribution of surface materials and the structural
features interpreted from several kinds of Landsat
digital processed imagery.
COMPARISON OF VARIOUS TYPES OF DIGITALLY PROCESSED
LANDSAT IMAGERY
Systematic and comprehensive analysis and
interpretation of standard product Landsat full-scene
images rapidly define the regional geologic setting of
the West Java area and delineate two areas with some
broadly distinctive geologic terrains. However,
because of the spatial and spectral resolution
characteristics of Landsat MSS, the results lack
detail, particularly concerning the variations in
sedimentary rock lithologies and of some of the
geologic structures in the region. Interpretation of
digitally enhanced Landsat MSS imagery irrproves
discrimination of lithologic variation and enables one
to extract more structural geologic details and
patterns, which in turn, lead to more reliable
tectonic synthesis.
Three types of digital image processing were
conducted in the Sukabumi and Leirbang test site areas
: Principal Component Transformation, Band Ratioing
and Image Edge Enhancement. The image processing and
manipulation in the Sukabumi area were particularly
intended to discriminate lithologies, to find some
lithologic contrast to support the structural geologic
interpretation and to find the relationship between
its drainage system to the structural geology. For
the Lembang area, however, these processings were
conducted with the intention of discriminating
volcanic materials erupted by different events and to
establish a firm understanding regarding the
development of the Lenbang fault.
The Principal Couponent Transformation (PCT) is a
linear transformation technique used to minimize data
redundance, a characteristic of highly correlated
data, as well as to enhance an area of subtle spectral
disparity. By rotation and translation, a set of
eigenvectors and eigenvalues, representing new or
transformed axes, was formed (Jenson, 1982). Using
these data, a new imagery - Principal Conponent
Imagery - can be developed (Figure:7). The
eigenvectors lie on the new transformed axes and the
length of the eigenvectors are eigenvalues. The
longest eigenvector, containing equal loading of Band
4,5,6,7 and totalling 95% of the total scene, was
chosen as the first Principal Coprrponent (PC-1)
(Colwell,1983). The PC-2 covers 4.3% of the total
scene variance while the PC-4 contains approximately
0.2% of the total scene variance.
ORIGINAL
Figure 7: Model of rotation and translation of Prin
cipal Conponent Transformation in two dimensions.
(Jenson & waltz, 1979)
The Principal Component imagery offers information
which previously was undetected on the Landsat
original data. The PC-1 imagery of the Sukabumi area
(Fig:8) shows very bright tones and coarse textures
on the recent Quarternary volcanic products (Qr2),
while the older volcanic products, the Qr4, show a
darker tone and smoother texture. These older volcanic
products are mostly weathered with high moisture
content and are now mostly agricultural fields. The
limestone of Bojonglopang (Te7) however, shows
brighter tone which indicates an area with low
vegetation density and lower moisture content -
characteristics of a carbonate terrain -, as compared
to the older Quarternary volcanic materials (Qr4)
(Figure:8b). The volcanic breccia of the Janpang
Formation (Te6) is distinguishable from the texture
and tone on the imagery. It is a highly resistant
rock formation with some bedding and fractures and is
easily discriminated from the surrounding materials.
The tuffaceous sandstone of Bentang Formation (Tpt) is
separable from the rest of the surface materials by
its tone as well as by their linear, tilted and folded
structures.
The PC-2 imagery (Figure :.8c) of same area shows
various grey-tones less distinct as conpared to the
PC-1 imagery. However, this PC-2 imagery is capable of
discriminating the marl of the Janpang Fm.(Te3) and
the occurence of the most recent volcanic materials
(Te8). In PC-2 imagery, with 4.3% of total variance
are included, the digital data of Band-4 and 5 are
contrasted against the digital data of Band 6 & 7,
resulting in tonal variations based on the variations
in vegetation density.
Band Ratioing is a digital processing technique
where pixel brightness values in one band are divided
by the corresponding pixel brightnes values in a
different band and from which a new ratio image then
may be created (Colwell,1983). The ratioing technique
minimizes the effects on scene brightness that are
caused by variations in topographic slope and aspect,