THE GEOMORPHOMETRIC DESCRIPTION OF CLUTER MAPS
J. G. Rodopoulos * *, G. Ch. Miliaresis"
* Technological Educational Institute of Athens, Department of Topography, 21 Souliou Str., Athens 154-51, Greece -
jrodopou@hotmail.com
? Dept. of Geology, University of Patras, Rion, 26500, Greece — miliaresis@email.com
Commission PS, WG IV/6
KEY WORDS: Geomorphology, Landslides, Interpretation, Extraction, Landscape, Segmentation, Spatial
ABSTRACT:
The aim of this paper was to describe from the geomorphometric point of view a cluter map derived by classification. More
specifically an ASTER image of the study area was radiometrically and geometrically corrected. Maximum likelihood classification
defined the landcover classes of the study area (various types of forest, cultivated land and bare ground). The digital elevation model
of the study area was derived by contour lines digitization from topographic maps. The landcover classes were parametrically
represented with attribute-value pairs. The attributes corresponded to the following geomorphometric parameters, mean elevation,
maximum elevation, roughness, local relief, meant gradient. It was found that the fir forest presents the greatest mean elevation
while the mixed forest and the bare ground class follows. The cultivated landcover class occupies areas with lower mean elevation.
The greatest mean gradient values are observed for both the fir forest and the mixed forest classes. The cultivated lands also present
the lowest mean gradient values while the barren class to present high mean gradient value. Then, a connected component labeling
algorithm identified distinct objects in the barren class. Each object was described by geomorphometric parameters and the objects
were ranked on the basis of landslides susceptibility. The most interesting finding was the greatest in size objects of the barren class
occupy the highest in elevation places and presents the greatest gradient values. Additionally the spatial distribution of these
polygons is along the main road connecting the capital with the main port of the island.
1. INTRODUCTION
The landcover mapping is of great significance in change
detection, urban planning and environmental protection. On the
other hand due to the severe climatic change and abrupt
meteorological incidences that take place recently, the
correlation between landcover and geomorphometry is of great
importance (Elumnoh and Shrestha, 2000). The risk and hazard
assessment studies for flash floods, landslides caused by heavy
rainfalls etc. (Miliaresis, 1999a; 1999b; 2001; White, 1993)
require knowledge of landcover type (Panagou and Miliaresis,
2003) and the geomorphometry (Pike, 2002) of the local
landscape neighbourhood (Treitz and Howarth, 2000). The
aim of the study is to a)parametrically represent the landcover
map derived by image classification techniques on the basis of
geomorphometrical parameters, and b)interpret the spatial
distribution of landcover types within the landscape.
Kefallinia Island
| Study Area
Ionian Sea
2. METHODOLOGY
pA RT
A terrain partition framework was composed on the basis of the
landcover types and objects (pixel arrangements with the same
landcover) were defined. Then, each object was parametrically
described on the basis of geomorphometric attributes and Figure 1. The study area.
mapped.
The island stands in the sea between Greece and Italy. The
2.1 Study Area study area is in the south east part of the island (Figure 1) and
enclosed by rectilinear coordinates as follows: X minimum =
The study area was Kefalonia Island in Ionian Sea. 200,137, X maximum = 222,337, Y minimum = 4,215,922and
Y maximum = 4,233,802.
* Corresponding author.
494
M Ct o 4 ANAND nr u]
-— NY NN
fu^ Fr. SS jl iig