In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
plays an important role in landslide body delineation. These 
morphometrical parameters were used also by Tapas et al., 2010 
to create rules for objects classification 
Weather data and mainly the rainfall data is a feature we 
proposed for this analysis. As landslidings depends on the 
rainfall thresholds we used the opportunity to collect data with 
the help of an automatic weather station. This was installed 
eastern from Breaza town, in Cornu village since 2007 in a 
characteristic position in order to obtain reliable data for rainfall 
every 15 minutes. From the total amount of data we selected 
only the records corresponding to the most relevant landslidings 
episodes around the area of the station. They feature mainly the 
torrential rainfall periods during the months of June and July. 
Another data set was that from Cimpina weather station data, 
which is situated southemwards from Breaza town. This rainfall 
data (since 2005) was selected according to the above 
mentioned principle (for the torrential rainfall episodes with 
landslidings re-activation). This digital data was the subject of 
interpolation in order to generate a special layer for image 
segmentation. The main target is to extend the analysis to other 
spatial data in order to improve the landslide inventory. 
Other data is represented by geological features in digital 
format. This layer was derived from 1:50000 scale geological 
maps on paper (edited by the Geological Institute). The role of 
this data within the analysis is quite limited because landslide 
inventory made between 2004-2009 in the Breaza town 
administrative area showed that landslide bodies occurs on 
almost every geological formations. Even sandstones and 
conglomerates alternating with clay and marls are featured by 
landslides in combinations with rockfalls. The lithological data 
was used only as ancillary data in the image segmentation and 
classification steps 
Materials 
Spatial resolution/Acquisition time 
Orthophotos 
0.5m 
2005 
Oblique air-photos 
0.1m 
2008 
DEM 
2.5m 
1980 
Lithology 
30m 
1980 
Precipitation 
10m 
2000-2009 
Table 1. Materials 
2.2 Methods 
For OBIA analysis we used ENVI EX product, developed by 
ITT (ENVI EX). The steps necessary to complete the feature 
extraction module from ENVI EX are: choosing the scale 
parameter; merging the object primitives; refine the objects 
using a threshold value for just one band of the image and it is 
an optional step; extraction of the attributes; object 
classification based on rules or examples. The segmentation 
algorithm used by ENVI EX is developed by ITT and is a 
patented product (Xiaoying, 2009). Chossing the right scale for 
the segmentation process is an important step in the OBIA 
analysis, which can lead to important errors in the final result. 
The scale parameter is related with the size of the objects that 
have to be identified and in ENVI EX the scale can take values 
from 0 to 100. Choosing a low scale value leads to small size 
objects and sometimes to over-segmentation on the other hand, 
choosing high scale can lead to under segmentation and large 
objects. It is always recommended that the scale parameter 
should be choosing after several trials should be combined with 
the merge objects step from the segmentation process. In ENVI 
EX the merge option presented in the second step is developed 
based on Robison et al 2002 
For the morphometrical analysis it was used ArcGIS Desktop 
9.3.1 - ArcINFO version with Spatial Analyst extension. The 
spatial analyst extension provides advanced tools for generation 
of DEM (the ANUDEM, Hutchinson, 1989) and for terrain 
analysis. From the DEM the slope, plan curvature and profile 
curvature were generated. An image with the amount of 
precipitation fallen in one week was generated for the entire 
area. The image was generated based on the measurements from 
the automatic meteo stations located in Cornu and on a linear 
relation between altitude and the amount of precipitation. This 
linear relation allowed us to spatialize the precipitation, even 
though the area has only ~350m difference in altitude 
In order to use the orthophotos and the morphometrical 
parameters for the segmentation and classification process, it is 
first required that all the data should be fusion. The fusion 
process was carried out in the ENVI 4.7 image analysis 
software and the file was saved and ENVI standard format. The 
schema of the analysis is presented in figure 2. 
Figure 2. Schematic representation of the analysis process 
3. RESULTS AND DISCUSSION 
The first step of the process was to choose the appropriate scale 
for the segmentation process. Several trials were performed with 
different scale values ranging from 20 to 65 and the best 
segmentation was obtained with a scale value of 35 
The second step was the objects merge and like for the first 
step, several trials were made. The best merge values for a scale 
parameter of 35 for our study area was 95