International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
Figure 2: Decollement gypsum levels on band 5, Landsat ETM+
image are highlighted in white.
2.1. Field studies
In this study, field-based data were formed with the aid of fault
kinematic studies collected from the southern parts of the basin.
These measurements were evaluated Carey method (Carey,
1979). According to this method, the generally active minimum
principal stress axis (03) for this region is in a vertical direction
and compresses a strike in a NNW-SSE direction (Figure 3).
a) b)
Strike — Plunge Strike Plunge
c! 153 17 at:
I 326
al 87
al 180 76
Figure 3: Fault measurement results. (Lower hemisphere
stereographic projections of striated fault planes measured in the
field. c1, o2 and 03 are respectively the maximum, mean and
minimum paleostress axes. Histogram shows that the
distribution of deviation angles between predicted slip vector (
1) and the computed slip vector (s))
3. IMAGE PROCESSINGS
Gypsum karst is also a geomorphologic phenomenon which
affects intensively Oligocene Hafik formation gypsum deposits
in Sivas Tertiary Basin (Gunay, 2002). Eastern parts of Sivas
Basin which where outcropped over large areas are under this
geohazard risk. Also, some collapse lakes such as Hafik,
Todurge, western and eastern Lota are located eastern part of
the basin. All lakes were formed by collapsing gypsum
occurrences in this region (Figure 4) similar to other worldwide
examples such as in Florida/USA.
^ H A
+ o NES nas 5 A : X.
Figure 4: Collapse lakes in the eastern section of Sivas Basin on
Landsat TM 3,2 and 1 bands.
In this study, image processing procedures were realized
Brovey transform, principal component analysis (PCA), and
unsupervised classification methods.
3.1. Image fusion
Because of availability of 15 meters panchromatic spatial
resolution feature of Landsat ETM+ system, Brovey transform
was chosen as one of the most robust fusion methods.
As is well known, 1, 4 and 7 th bands represent visible, near and
middle infra-red regions of electromagnetic spectrum. Due to
this basic fact that, their RGB combination display which were
composed of respectively 7,4 and 1 bands emphasize
differentiation between lithological contacts combining with
spectral information content (Drury, 2001).
This method is evaluated the most photo-interpretative method
according to the other types of data fusion techniques such as
high-pass filtered, multiplicative, principal component analysis
based and IHS (intensity, hue and saturation) methods (Carter,
1998; Bretschnider and Kao, 2000).
This transformation algorithm (ER Mapper, 1998) was used for
RGB display in this study as below:
Red= band 7/ (band 1+ band 4+ band 7) + Landsat ETM+
panchromatic band
Green= band 4/ (band 1+ band 4+ band 7) + Landsat ETM+
panchromatic band
Blue= band 1/ (band 1+ band 4+ band 7) + Landsat ETM+
panchromatic band
A healthy differentiation based on visual interpretation between
exposing lithological units of Sivas Basin with the aid of
combining spatial and spectral enhancements was provided by
Brovey transform in this study (Figure 5). Although vegetation
is an undesirable component in image processing studies, it has
been very helpful geologically to derive structural features in
this study. In Figure 5, dark blue colour shows realms of Inner
Tauride Ocean and characterizes allochthonous ophiolites.
Figure also shows a remarkable fold-thrust belt that was
developed between Lower-Middle Eocene Bozbel formation,
which was composed of laterally graded flysch, tuffites and
volcanic intercalations and Late Cretaceous ophiolitic mélange
rocks. Bozbel anticline, which was described firstly by
Kurtman, (1973), elongates in NWW-SEE direction compatible
with boundary of southeastern parts of the basin.
Bea ^ NR a s wt e e :
Figure 5: Spectacular fold expressions on Landsat ETM+ 7,4, 1
(RGB) combination with Brovey transform displays southern
parts of the Sivas Basin.
3.2. Principal Component Analysis (PCA)
PCA, which was also known as Karhunen-Loéve (Sabins, 1996)
method is a linear transformation technique in image processing
550
Inter
—
stud:
band
SPO
As 1
prov
regit
obse
appr
bout
An
geo
det
geo
sho
rest
diff
But
up |
in y
oft
