Journal of Global Positioning Systems (2005)
Vol. 4, No. 1-2: 184-191
A Demonstrative GPS-aided Automatic Landslide Monitoring System
in Sichuan Province
Pinggen Zhou, Baishen Zhou
China Institute of Geo-environmental Monit oring, Beijing, 100081,China
e-mail: [e-mail]; Tel: [phone]; Fax: [phone]
Jingjun Guo, Donghang Li, Zhigang Ding
Tsinghua University, Beijing, 100084, China
e-mail: [e-mail]; Tel:[phone]; Fax: [phone]
Yanming Feng
Cooperative Research Centre for Satellite Systems, Queensland Univ e r sity of Technology, GPO 2 4 34,Q4001
e-mail: [e-mail]; Tel:[phone]; Fax: [phone]
Received: 15 November 2004 / Accepted: 12 July 2005
Abstract. In China, the geological disasters of landslide
and mud-rock flows cause losses of over 1000 lives and
total economic losses of over 10 billions of RMB each
year. There have been about 90,000 identified landslide
sites, mostly distributed in several southern and north-
western provinces of China. In the reservoir area of the
Three-Gorge project only, over 1 000 landslide sides h ave
been identified. A joint research was launched involving
authors from a number of scientific institution s to explore
technologies and methodologies for landslide monitoring
with focus on the characteristics of the geological
disasters at the up and middle reaches of Yangtze River.
This paper studies the combined technologies for
landslides monitoring and presents a demonstrative
automatic landslide monitoring system in a chosen
ancient landslide site, where the creeping movement
process continues since its latest large sliding on August
25, 1981. The landslid e test-bed is 500 m lon g and 300 m
wide located in the Ya’an-Xiakou area in Sichuan
province. To study the mechanism of the sliding process,
15 permanent GPS monuments were built in the area for
regular observations. Automatic ombrometers, digital
thermometers, underground water-level-meter and ground
fissure-displacement-meter were set-up as well. The data
from these automatic sensors are collected and
automatically sent to the data process centre in Beijing
via the Beidou-1 communicat ion satellite. The paper also
compares the landslide results from three GPS observing
campaigns, demonstrates the feasibility to identify the
displacements at the accuracy level of 2 mm using the
dual-frequency GPS receivers. The results are
encouraging and further analyses will be conducted,
considering influences of non-GPS measurements.
Key words: Augmented Reality, GPS, Inertial Sensors,
Kalman Filter, Integrated Systems
1 Introduction
China is a country frequently suffered from various
geological hazards. With the boom of economy and a
great deal of ongoing engineering projects such as
transportation, water conservancy and resource
development, as well as the influence of environmental
changes, damages caused by landslide are sharply
increasing. According to Tang (2004), in recent years,
rock falls, landslide and debris flow have caused an
economic loss of about 10 billion RMB and over 1000
lives per annum. In China, about 700 Counties (local
governments) have suffered from geologic hazards for
long and tens of millions of residents living in these areas
face threatens of serious geologic hazards, but lack of a
sense of safety. This sometimes causes serious social
problems. Therefore, it is particularly important to
develop cost-effective reliable landslide monitoring
systems and technologies.
Landslide monitoring is a complex of technological
means. Deformation monitoring is one of the most

Zhou et al.: A Demonstrative GPS-aided Automatic Landslide Monitoring System in Sichuan Province 185
important parts, providing an important basis for
identifying landslide. Conventional geodetic surveying
methods for deformation monitoring include transit
traverse survey, triangulation method, levelling survey,
total station methods. These methods provide reasonable
accuracy, but requiring skilled professionals to conduct
the work in situ, resulting in heavy workload, high
personnel risk and low efficiency. Monitoring and timely
alarms in case of hazard cannot be realized at night or in
continuous rain. Use of GPS can improve the situations
considerably (Xu et al, 2003, Xu et al.,2003, Gili and
Corominas, 2000). First, GPS can operate in all-weather
conditions and inter-station visibility is not required;
second, GPS static relative positioning can achieve
millimetre accuracy to meet the requirements (Guo et al,
2004 and Zhao et al, 2001). Third, if the satellite
communication is available to transfer the observation
data from the remote site lacking of landline
communication to an urban data processing centre, thus
realizing automatic monitorin g is more feasible.
Addressing the characteristics of geologic hazards in the
middle and upper reach of Yangtze River in China, China
Institute of Geo-Environment Monitoring and Tsinghua
University jointly constructed GPS-based landslide
monitoring demonstrative network at the Ya’an-Xiakou
landslide site in Sichuan Province. Regular monitoring
results since 2003 have demonstrated that GPS can
provide required accur acy for landslide monitoring in the
demonstrative case, which can lead to completely
replacing conventional geodetic surveying instruments,
showing also advantages in terms of speed and efficiency
(Guo et al, 2004). The test-bed is also equipped other
monitoring devices such as automatic pluviometer, digital
thermometer, underground water gauge and geofracture
displacement meter, to form a complete set of automatic
landslide monitoring system. All the monitoring data are
transmitted to Beijing Monitoring Analysis Centre
through the China Beidou No 1 navigation/satellite
communication channel. The system is 24/7 automatic
unmanned monitoring system.
2 GPS landslide monitoring experiments and results
2.1 Description of the experimental landslide area
As shown in Figure 1, the Ya’an- Xiakou, also called
Wujia Mountain, landslideis situated in the east bank of
the gorge in the middle reaches of Longxi River in
Longxi Village, north of Ya’an City, Sichuan Provin ce. It
is the largest landslide in the local area. The whole
landslide area consists of an ancient rock fall and
landslide deposit of about 10 million M2. (unknown
time), a reactivation landslide of 2.60 million M2 (1981)
as well as a deformable body of 0.75 million M2 (the part
deformed in late 1990s after sliding in 1981). Overall
sliding is direction I is from East to West. The sliding
mass is the ancient rock fall deposit, gravel mixed with
mauve clay, and the stone’s diameter is 0.2to 2m. The
sliding mass features favourable conditions for influent
seepage and shallow layer of underground water. The
region enjoys warm and wet climate, and that is why
Ya’an is also called Rain City. The mean annual
temperature there is 16ºc, and the annual rainfall is
1,800mm, mostly in the season June to September.
On the night of August 19th, 1981, a storm broke out, and
a large amount of rain flowed into the slope, triggering
off a large-scale slide and resulting in collapse of houses
and ruin of roads and channels. Different degrees of
deformation occurred in the creep deformation body of
the reactivation landslide following the rain season in
1995, and formed a potential sliding mass(Zhao).
Experimental area is located at the reactivation body,
500m long and 300m wide, running from east to west
slightly southwards about 25.50º.
Fig. 1 Sliding mass at the X iakou landslide 2.1 Reference Frames
2.2 Ya’an-Xiakou GPS landslide monitoring network
Reference points. Two points at the stable rock mass
outside the sliding mass were chosen as reference sites.
One is in the east and the other is in the northern ridge,
numbered as ya15 and ya17, respectively
Monitoring points. We set up monitoring points
according to the features of the local sliding mass. These
points should reflect the general deformation direction of
the sliding mass, magnitude of the deformation, as well
as range of the sliding mass and deformation speed.
Reception of signals from the satellite was considered at

186 Journal of Global Positioning Systems
each point. There should be no large-area screening
objects over the monitoring point (Liu et al., 1996). There
are in total fifteen monitoring points placed in the test
area, of which there are five points placed from top to
bottom crossing the section of landslide, each being close
to the inclinometer and crack displacement measuring
device. Three monitoring points are set up next to the
pluviometer and the underground water level measuring
device. The remaining seven points are scattered around
the sliding mass. The average distance between the points
is 120 m, with the maximum of 229 m and the minimum
of 17 m. At each monitoring point, the GPS monument
was cast with reinforced concrete with 1.5 m underneath,
and 1.5 m above the ground. A forced centring device is
also placed on the top of the observation pillar. Figure 2
illustrates the layout of the monitoring points of the GPS
network.
↑25.5 south of due east
Scale 1:5000
Fig. 2 Layout of GPS observation po ints
2.3 Collection and proce ssi ng of GPS data
Four Novatel PROPAK-II-TR2 dual-frequency GPS
receivers were used for GPS data collection and
observation conducted in March and October of 2003 in
the area. The data collection with a PDA-based device
developed by the researchers from Tsinghua University
GPS data were recorded at 5 seconds interval with 1 hour
session length. The elevation cut-off angle of 10º was set.
Meanwhile, Activesync software was used to enable PDA
and computer communication. The observation data were
converted to the standard RINEX format for processing.
The baseline positioning was completed using Trimble
Geomatics Office software. The results show the
maximum RMS values of 8 mm, the minimum value of 2
mm and the average value of 3.4 mm. 2.2.1 Kalman filter
functional model for the GPS observations
2.4 Results and Analysis
Baseline consistency check
To ensure accuracy of the observation results, in each
round of experiments, several baselines were observed
twice, allowing to conduct internal consistence checks
between the solutions. Table 1 gives the results from the
test conducted in March 2003. As seen, the maximum
difference between the two baseline solutions is 4 mm,
while the RMS is 3 mm for all the baselines.
Tab. 1 Baseline consistency Check
FromTo Baseline length
(m)
Computation
method Ratio RMS (m)
ya04ya05142.184 L1 fixed 4.5 .003
ya04ya05142.188 L1 fixed 5.7 .003
ya08ya11136.728 L1 fixed 23.5.003
ya08ya11136.732 L1 fixed 53.6.003
ya11ya1284.006 L1 fixed 22.2.003
ya11ya1284.003 L1 fixed 31.8.003
ya11ya14133.899 L1 fixed 11.4.003
ya11ya14133.899 L1 fixed 33.2.002
Close- loop consistence checks
The zero theoretical value for sum of the coordinate
differences in a loop formed by three or more baseline
vectors can be used to check the results as well. Referring
to Figure 2, each triangle in the network can form a close
loop. There are totally 4 0 close loops, of which 39 passed
the test. The failed one related to the monitoring point
ya06, which is close to a house and same threes. The
quality of GPS signals might be affected.
GPS network ad j ustment
The baseline solutions from the test in March 2003 was
adjusted using rank defect free network adjustment (Cui
et al., 2000). Table 2 shows the positional RMS accuracy
of the network adjustment results .
Gauss proje ct i on
To minimize the deformation of the coordinates in the
landslide monitoring, Gauss map projection is made at
the arbitrary zone closest to the survey area. That is, the
longitude 103 º 01’ E is taken as the central meridian and
projection is made at north latitude 30º as latitude of
pedal with the projection height of 820 m (geodetic
height), thus minimizing the projection error and ensuring

Zhou et al.: A Demonstrative GPS-aided Automatic Landslide Monitoring System in Sichuan Province 187
that the projection error will not affect the accuracy of
monitoring. To ensure that Y coordinate is the positive
one, we moved Y coordinate eastwards to the longitude
103 ºE. Table 3 gives the coordinate projection results.
Tab. 2 Positional accuracy after network adjustment process
Points Error in vertical
axis（m） Error in horizontal axis（m）
Ya01 0.009 0.004
Ya02 0.003 0.002
Ya03 0.003 0.003
ya04 0.002 0.002
ya05 0.003 0.002
ya06 0.003 0.003
ya07 0.002 0.002
ya08 0.002 0.002
ya09 0.002 0.001
ya10 0.003 0.002
ya11 0.002 0.002
ya12 0.002 0.002
ya13 0.002 0.003
ya14 0.002 0.002
ya15 0.003 0.003
ya16 0.003 0.003
ya17 0.002 0.002
Maximum 0.009 0.004
Minimum 0.002 0.001
Average 0.003 0.002
Tab. 3 Results of Gauss Map Projection
Points North
coordinate
（m）
Error in
vertical
axis（m）
East
coordinate
（m）
Error in
horizontal
axis（m）
ya01 7840.813 0.009 1463.268 0.004
ya02 7858.419 0.003 1462.534 0.002
ya03 7920.010 0.003 1509.314 0.003
ya04 7860.236 0.002 1523.291 0.002
ya05 7718.432 0.003 1519.071 0.002
ya06 7753.409 0.003 1610.950 0.003
ya07 7725.563 0.002 1648.389 0.002
ya08 7656.025 0.002 1690.707 0.002
ya09 7793.487 0.002 1692.598 0.001
ya10 7723.173 0.003 1739.654 0.002
ya11 7781.867 0.002 1741.487 0.002
ya12 7700.098 0.002 1758.348 0.002
ya13 7662.237 0.002 1752.438 0.003
ya14 7675.655 0.002 1815.950 0.002
ya15 7680.316 0.003 1864.527 0.003
ya16 7767.043 0.003 1439.250 0.003
ya17 8022.828 0.002 1467.256 0.002
Conversion of coordin ates
To facilitate analysis on the sliding mass, we convert the
gauss projection coordinates to rectangular coordinate
system at the sliding mass direction. The main sliding
mass direction ya01-to-ya12 is the negative direction for
X- axis, and Y- axis is normal to X- axis to form the right
hand rectangular coordinate system. Table gives the
coordinates of each point after the coordinate conversion.
Tab. 4 Results of Coordinate Conversion
Points X after conversio n
(m) Y after conversio n (m)
ya01 917.109 846.902
ya02 908.869 831.326
ya03 924.583 755.597
ya04 962.927 803.534
ya05 1020.155 933.346
ya06 1088.032 862.227
ya07 1133.811 871.246
ya08 1201.940 915.798
ya09 1144.478 790.908
ya10 1217.218 834.120
ya11 1193.608 780.353
ya12 1244.024 846.902
ya13 1254.986 883.620
ya14 1306.538 844.170
ya16 927.183 923.826
ya17 842.364 680.894
Table 4 gives the results from the experiment conducted
in March 2003. The second experiment took place in
October. The same observation methods and data
processing strategies were taken as in March 2003, but
the coordinates of the reference point ya15 known from
the pervious time were used in the network adjustment.
In addition, no observ ation was taken for Point ya07, due
to the complete blockage of signals by crops.

188 Journal of Global Positioning Systems
Tab. 5 Displacement of points
dx(m) dy(m)
YA01 -0.040 -0.001
YA02 -0.031 0.008
YA03 -0.042 0.000
YA04 -0.042 -0.003
YA05 -0.028 0.016
YA06 -0.027 -0.008
YA08 -0.012 0.015
YA09 -0.042 -0.003
YA10 -0.025 -0.005
YA11 -0.021 -0.014
YA12 -0.015 0.004
YA13 -0.017 0.010
YA14 -0.003 -0.012
YA15 0.000 0.000
YA16 0.008 -0.002
YA17 -0.032 0.001
↓ Direction of main coast
Scale 1:5000 Slide Measure 1:5
Ta
Fig. 3 Map of points displacement
In Table 5, Dx and Dy are the coordinates differences
between two experiments for x and y components,
respectively.
As seen from Table 5 and Figure 3, sliding mass not only
moves along the main sliding direction of the landslide,
but in the direction normal to the main sliding direction.
This is because the sliding mass projects in the middle,
and sinks at the two sides. Results from GPS landslide
monitoring accurately reflect sliding tendency and
displacement of the sliding.
Fig. 4 Composition of landslide remote m onitoring system
3 An automatic landslide monitoring system remotely
controlled though Beidou-1 communication satellite
3.1 Compone nts of the syst em
As shown in Figure 4, the remotely controllable
automatic landslide monitoring syste m, built in the Ya’an
landslide area, has four major components: in-field data
collection & monitoring station, digital automatic
landslide monitoring point, Beidou-1 navigation satellite
communication system and “geologic hazard monitoring
& analysis center”. Detailed descriptions are given as
follows.
(1) Field data collection & monitoring station
The field data collection & monitoring station is a set of
unmanned integrated data collection system. It includes
an information monitoring and collection platform for
various geological measurements, such as absolute

Zhou et al.: A Demonstrative GPS-aided Automatic Landslide Monitoring System in Sichuan Province 189
displacement monitoring for landslide ground,
monitoring for displacement of the deeper parts of the
landslide, dynamic monitoring for landslide underground
water, landslide relative displacement monitoring, and
monitoring for inducing factors of landslide. These
elements are consolidated and integrated into one unit to
collect, store, compress and transfer various data. Figure
5 is the integrated rack .
Fig. 5 Field data collection & monitoring station
Fig. 6 Underground water level monitoring point
(2) Digital automatic landslide monitoring points
These digital automatic landslide monitoring points are
scattered everywhere in the landslide area in order to use
various digital monitoring methods to collect data
concerning the geological environmental parameters,
such as underground water level, water temperature,
rainfall, displacement of the earth surface, deformation of
deeper parts and so on. The monitoring devices include
borehole tiltmeter, geofracture displacement meter,
digital automatic pluviometer, automatic water level
meter and so on. Figure 6 indicate one of the monitoring
points-un dergroun d wat e r le vel monitoring point .
(3) Beidou-1 navigation/ communication satellite system
“Beidou-1 navigation communication satellite system”is
a regional navigation system in China. It can provide all-
weather, around-clock satellite navigation information,
time service and two-way communication service (Ha,
2004). Communication signal of this system is stable, and
high intensively encrypted measures are designed to
ensure security and reliability, so it is very suitable for
monitoring landslide hazard in harsh and complicated
field conditions. This system is composed of space
segment, a ground controlling center and Beidou-1 user
terminal. Figure 7 illustrates the concept of the Beidou-1
system and services.
The space segment of Beidou-1 navigation
communication satellite system includes two earth
synchronous orbit satellites (GEO), namely BDSTAR-1
and BDSTAR-2. The signal transmission device on the
satellite is available to complete the relay task for two–
way wireless signals between ground controlling central
station and the client terminal.
Fig. 7 Diagram for comprehensive informatio n s e r v ice structure of
Beidou-1 navigation communication satellite system
(4) Geologic hazard m oni toring & an alysis center
The geologic hazard monitoring & analysis center
comprises a Beidou-1 communication user receiver, a
database server and an analysis unit. In terms of its
functions, the system includes the controlling software
for landslide remote monitoring system operation a
database and monitoring data analysis software. Through
Beidou-1 communication client receiver, the center uses
satellite signal channel to receive data collected and sent
from the field, and process and store the data and results
in the database server. The system controlling software
can automatically feed back information, and meanwhile

190 Journal of Global Positioning Systems
telecontrol field data collection & monitoring station
according to users’ requirements for field data collection.
The data analysis software can read data from the
database any time, analyse and process the data collected,
obtain the changing and alarm information of geologic
feature, and thus providing valuable reference to the
researchers and identifying landslide hazard, and make
accurate early-warning. Figure 8 shows the components
of the experimental system for monitoring and analysis
center.
Figure 8. Geo-hazard monitoring & analysis system (experimental
center)
3.2 Operation and monitoring function of the system
The remote automatic landslide monitoring system, based
on Beidou-1satellite communication, operates as follows:
(1) Field data collection, storage and processing of the
geological environmental features, including
underground water level, water temperature, rainfall,
the earth’s surface displacement, deformation of the
deeper parts collected by the various digital
automatic landslide monitoring points scattering in
the landslide monitoring region;
(2) Regular data packaging and compression, or via
telecontrol according to the needs, using the
transmission and communication function of
Beidou-1 satellite nav igation system to directly send
the data to the geologic hazard monitoring &
analysis center, where the data are analyzed and
processed. Meanwhile it can send feedback
information and control instructions to the field
system through Beidou-1 system as well.
The system enables the center facility to complete the
remote monitoring and control over the landslide
monitoring sensor through Beidou-1 satellite
communication link; while the landslide monitoring
sensor regularly transmits the data processed primarily
and locally stored by the field data collection &
monitoring station to the central facility. The system is
also equipped with satellite communication receiver for
cross checking and mutual checking. It can also display
the sensor status, check the sensor data; query the current
record file in the monitoring station, request for sending
the data on the specified date and time, request for
sending all the data of designated file, telecontrol runn ing
of the monitoring station from the starting point, display
in real time the sensor data; time mark, the received data;
display the particular running status of the program the
instructions ID length of received information; telecontrol
the normal exit (save the setting of the current state) and
cancellation of the program (exit of the program without
change of the original state). Meanwhile the central
facility has the capacity of automatically processing
remote monitoring data and pub licizing the early-warning
information promptly.
4 Conclusion s
GPS monitoring experiment results from the Ya’an-
Xiakou testbed have demonstrated that GPS can provide
sufficient accuracy to meet landslide displacement
monitoring requirements, leading to the replacement of
the conventional geodetic surveying methods. Use of
GPS-based methods can significantly improve the
efficiency of landslide monitoring and reduce the work
intensity of labour. In addition, with means of the
Beidou-1 satellite communication system, remote
automatic monitoring of the landslide becomes feasible.
The demonstrative GPS-aided automatic landslide
monitoring system developed has performed well,
providing an excellent technological platform for the
academics and experts engaged in the research of
geological and environmental hazards to timely,
accurately and conveniently obtain the real-time
monitoring data at each dangerous landslide region. As a
result, it saves costs for labour and data acquisition in
filed and avoid potential hazard to personnel in the field.
Meanwhile, the comprehensive information volume
obtained through the system is far greater than the
effective data obtained through the conventional means.
Moreover, a database with rich and detailed resources can
be set up, which would be of great significance to the
landslide hazard investigation analysis, forecast, early-
warning and hazard mechanism research in the future.
ACKNOWLEDGEMENTS
This work was carried out with financial support from the
Ministry of Science and Technology (MOST, P. R. China
and partially from Australia-China Special Funds for
Scientific and Technological Cooperation bilaterally

Zhou et al.: A Demonstrative GPS-aided Automatic Landslide Monitoring System in Sichuan Province 191
offered by MOST and Department of Education, Sciences
and Technology (DEST), Australia.
References
Yin Y. (2004): Initial study on the hazard-relief strategy of
geological hazard in China, The Chinese Journal of
Geological Hazard and Control, 15 (1): 1-8
Zhou P. (2004): Indicators and techniques of landslide
monitoring, Journal of Geomechanics (in Chinese),
10(1):19-26
Xu, S, W. Cheng, X. Huang, D. Xu (2003): The investigation
of landslide monitoring in the Three Gorges Reservoir
region by applying GPS, Journal of Hydraulic Engineering
(in Chinese). (1):114-118
Gili J.A.; Corominas J. (2000): Using Global Position System
techniques in landslide monitoring, Engineering Geology
(special issue), 55:167-192
Guo J J.; Yang Z.; Ding P.; Zhou X Zhu. (2004): Application of
GPS on landslide monitoring-A case study of Xiakou
landslide, Journal of Geomechanics (in Chinese). 10(1):
63-70
Zhao Y. (2001): Mechanical analysis of sliding process of
Ya’an Xiakou landslide and prediction of its hazard
model, Journal of Engineering Geology(in Chinese). 9(2):
188-193
Liu, D Y Shi, J Guo (1996): The Elements and Data
Processing of Global Position System, Tongji University
Press, Shanghai
X Cui, Z Yu, B. Tao (2000): The Generalized Surveying
Adjustment, Survey and Mapping Press, Beijing
Ha B. (2004) Beidou Navigation and Position Satellites
System, China Metrology,(3):22-23