LANDSLIDE MONITORING THROUGH KALMAN FILTERING: 
A CASE STUDY IN GÜRPINAR 
M. Acar *, M. T. Özlüdemir, R. N. Celik, S. Erol, T. Ayan 
ITU, Civil Engineering Faculty, Division of Geodesy, 34469 Maslak Istanbul, Turkey 
(acarmusta, tozlu, celikn, erol, ayan)@itu.edu.tr 
Commission VII, WG VII/5 
KEY WORDS: Landslide monitoring, deformation analysis, Kalman Filtering, GPS 
ABSTRACT: 
Many regions in Turkey are located on earthquake zones. The latest major earthquakes that hit eastern Marmara on August 17th and 
November 12th 1999 left tens of thousands of deaths. Another natural disaster that is a common in Turkey, mostly in Black Sea 
region and in some particular locations, is the landslides. In recent years several landslides happened after heavy rains and the 
resulting floods. This makes the landslide monitoring and mitigation techniques an important study subject for the related 
professional disciplines in Turkey. In the periphery of Istanbul, which is a metropolis with a population of more than 10 million, 
there are also some landslide regions. In this study, a landslide monitoring project in such a landslide area and the results obtained 
are discussed. The area is near Gürpinar village, located along the Marmara Sea cost to the northwest of Istanbul. The area that had 
been formerly an agricultural place was opened up for the settlements starting from the mid 80's. After than, several houses, mostly 
designed as weekend houses, were built. However, the effects of the landslides on the buildings were terrible. Many of the 
constructed buildings were damaged. The landslide monitoring project was carried out to identify the characteristics of the landslides 
and their potential risks. The geodetic measurement campaigns consist of two phases. In the first part of the monitoring, terrestrial 
geodetic observations were done in 6 epochs between 1990 and 1991. By the time, some network points were lost. In the second 
phase, the remaining network points were measured using the GPS technique in 4 epochs between 1996 and 1998. The results of the 
terrestrial works had been published before. In this study, only the campaigns with GPS observations are evaluated. The deformation 
analysis procedure applied is based on the Kalman Filtering technique. The results indicate the seriousness of the landslides. The 
displacements detected reach up to 5 meters. In the study, the approach for the deformation analysis and the overall displacements 
are presented. 
1. INTRODUCTION 
The basic objective of the science of geodesy is to determine 
the shape of the Earth and its gravity field. Besides, it deals 
with many applications related to positioning and surveying. 
One of these fields is the determination of crustal movements. 
Crustal movements or ground movements sometimes suddenly 
happen and these sudden movements are generally caused by 
earthquakes or landslides. The actual crustal movements, 
however, are slow that are difficult to be determined. Due to the 
slowness of these movements and the differences between the 
reference systems chosen for comparison, direct determination 
of these movements is normally impossible. Therefore, relative 
movements are determined using repeated observations 
(Yalginkaya, 1994). The advances in measurement technology 
in the last three decades have made great contributions to 
deformation monitoring studies. As a result of these 
developments, deformation monitoring on large structures like 
dams, bridges, towers and the areas subjected to landslides and 
earthquakes have become widely applied. 
For the analysis and the interpretation of deformations, different 
deformation models have been developed. These models consist 
of static, kinematic and dynamic models. Static model that is 
not dependent on time provides the determination of 
deformations on the characteristic points of the area or the 
structure, which is monitored. In time and position dependent 
  
* Corresponding author. 
kinematic model, the movements of the characteristic points 
and the motion of these movements are investigated. In 
dynamic model, in addition to the kinematic model, the relation 
between deformations and the influencing factors are also taken 
into consideration. 
Different deformation analysis algorithms have been applied so 
far. One of these mathematical approaches is the Kalman 
filtering technique. In this study, a deformation analysis 
procedure that has been carried out through a kinematic 
deformation model based on Kalman filtering technique is 
discussed. The geodetic data used for the analysis had been 
collected in a landslide area. In the following sections, 
characteristics of the landslide area, deformation analysis 
procedure and numerical applications are given. 
2. CHARACTERISTICS OF LANDSLIDE AREA 
The study area is a landslide region located nearby Gürpinar 
village that is to the northwest of Istanbul. In the area, without 
doing proper geotechnical investigations some buildings, 
mostly weekend houses had been built. But after the 
construction work had completed, many damages on the 
constructions took place as a result of landslides. In order to 
investigate the effects of landslides in and around the settlement 
area, a multidisciplinary project has been realised. Throughout 
the project geotechnical investigations and geodetic 
582 
Inte 
defi 
20€ 
Ac 
arti 
mo' 
und 
the 
hea 
Cari 
was 
sma 
area 
(Fig 
perf 
Cur! 
ou 
Figu 
(Alt: 
Tim 
velo 
form 
7 (Ka4 
X; 
7 (k+] 
Y; 
Zt 
J: 
wher 
7 (k+ 
X 
xo 
a? 
y? 
k= 
j 1, 
Kalm 
prese 
motio 
at per 
positi 
accel