04
due
ard
to
ER
and
nds
15-
ER
on.
)-m
)-m
gth
but
ack
the
the
low
VIR
the
etic
lon
"ed
ited
udy
ron
and
it 7
M+
son,
data
sely
tral
rion
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
2. CONCEPTION AND METHODOLOGY
2.1 Used Data
We have used on-demand product ASTER Surface Reflectance
(AST 07) from the EROS Data Center (EDC), which contains
surface reflectance for each of the nine VNIR and SWIR bands
at 15-m and 30-m resolutions, respectively. The results are
obtained by applying an atmospheric correction to radiances
reported by the ASTER sensor. The atmospheric correction
removes effects due to changes in satellite-sun geometry and
atmospheric conditions. The atmospheric correction algorithm
is applied to clear-sky pixels only and the results are reported as
a number between 0 and 1 (Abrams, ASTER User Handbook).
As additional information for support the ASTER Digital
Elevation Model has been used.
2.2 Workflow of Study
The processing of ASTER Surface Reflectance data (AST 07)
consists of four steps. In the first stage all remotely sensed data
were geometrically pre-processed (so as to have the same
geodetic reference). In the second stage, the different false color
composite images and image enhancement techniques were
applied to the data and geological photo interpretation was
carried out. In the third stage, different band ratio techniques
were applied to the data and finally, in the fourth stage, the
spectral mapping techniques (as Matched Filtering, ENVI
software technique and Variable Multiple Spectral Mixture
Analysis, Garcia-Haro — 2001) in order to map different mineral
groups were accomplished.
However Russian authors discuss in detail geochemical
characteristics, the information about geospatial distribution of
phenomena is insufficient and can be just supposed (no
geographic information as maps or coordinates is mentioned).
This fact does not allow us to get precise knowledge of the area
and thus to link spatially maps in Dvorov, 1975 or Lebedev,
1983 with satellite imagery.
2.3 Image Interpretation
Different band combinations have been tested to provide best
image interpretation using false-colour composite image. Other
RS techniques (contrast stretches, histogram equalization,
image transformations) was applied for image enhancement
before interpretation.
Following false-colour composite images has been used to carry
out interpretation of study area:
e 321 false-colour composite to obtain the best spatial
resolution images and to get basic information about
vegetation across the area
e 631 false-colour composite has showed the most clear
differences in the geological and geographical information
mentioned in the literature
Bands 321 false-colour composite is commonly used in order to
highlight the presence of vegetation, which appears in red
shades whereas the other land cover types are decreased. In the
ASTER imagery can be used 321 bands instead of Landsat 432
bands. The whole area is shown in brown shades except the
small areas close to settlements (Cheleken City and unknown
village) which are most likely irrigated gardens. The rest of
peninsula should be avoid of vegetation or sparsely covered.
Bands 631 false-colour composite has proven to be the most
worthwhile for mapping of geological phenomena. Roughly, the
647
area is divided into three parts. The north part represents flat
plain covered by eolic sand covers (dunes) appearing light
yellow or red-brown, the central part (fig. 4) is formed by
various types of phenomena — Red Series — shown in light
yellow and highly disrupted by system of erosion gullies; mud
volcanoes shown in pink colour (situated in the neighbourhood
of Red Series — Aligul respectively Rozovyj Porsugel, and in
the W part of image (Zapadnyj Porsugel); solonchaks are
situated mostly along fault on the SE slope of Chochrak Ridge
and between Aligul and Zapadnyj Porsugel or as a result of
recent volcanic activity of Zapadnyj Porsugel mud volcano. The
brines affected areas appears from dark green to dark brown
shades. Man affected areas NW from Chochrak, which
surrounds rectangular reservoirs and appear in dark grey or
black. The south part represents flat plain gently sloped towards
South Cheleken Bay. Most of the area appears in light blue,
cyan or white colours, in some areas disrupted by solonchaks
(dark green shades).
Figure 4. Bands 631 colour composite approved best
capability to map geological phenomena across whole
peninsula. Red series appear as triangle in the centre, mud
volcano deposits appear as pink areas westwards (Aligul) and
eastwards (Rozovyj Porsugel) from Red series.
2.4 Band Rationing
Ratio images designed to display the spectral contrast of
specific absorption features have been used extensively in
geologic remote sensing and vegetation mapping.
ASTER band 1 / band 2 ratio image. Ratio of these
wavelengths has been successfully used by Segal, 1989, who
have used Landsat TM bands 3/2 ratio to map iron-bearing
rocks in Lisbon Valley, Utah. According to Hunt, 1977, ferric-
iron (Fe’") rich rocks exhibit a sharp fall-off in reflectance
approximately from 0.8 micrometers to shorter wavelengths,
thus ferric-iron rich exposures exhibit very low ASTER band 1/
band 2 values.
When band rationing has been applied to Cheleken data, the
lowest band ratio values clearly represent Red Series situated in
the middle of the scene and confirms supposition, that red-
colouring of beds is caused by higher iron content of rock
layers. Besides the Red Series, low ratio values appear in some
areas of the North Plain located most likely at summits of sand
dunes, which are one of landscape-forming feature mentioned
by Dvorov, 1975.
The newest studies of geochemical characteristics of Holocene
sediments in the Aral Sea region have found out significant
difference between Syr Darya and Amu Darya sediments. The
