An early warning system for active landslides

Omar Husaini1 & M. Ratnasamy2

1
Director, Mountainous Terrain Development Research Centre, Faculty of Engineering, Universiti Putra Malaysia,
43400 UPM Serdang, Selangor, Malaysia. (e-mail: husaini@eng.upm.edu.my)
2
Associate Director, Road Safety Research Centre, Faculty of Engineering, Universiti Putra Malaysia,
43400 UPM Serdang, Selangor, Malaysia. (e-mail: ratnas@eng.upm.edu.my)

to threshold limits. These measurements then provided a

forewarning of any adverse effects on the safety of
lope failures are commonplace in tropical construction (Dunnicliff 1988).

S countries like Malaysia. Where they affect infra-

structure, an efficient and safe method for
monitoring of such collapsed slopes provides
an indispensable safeguard. This paper describes an
early warning system that was assembled and trialed
According to Abramson et al. (1996) in all cases
whether as part of construction or remediation, the
instruments should be used for monitoring purposes.
The data collected should be processed by the close of
work each day and shared with all of the interested
during the reconstruction of a collapsed slope at Gunung parties by the following morning.
Tempurung on the North South Expressway. The magni- This paper describes real-time monitoring using soil
tude of the landslide and the risk involved in the recon- strainmeters as an early warning system during recon-
struction of the active slope posed a direct challenge to struction of a collapsed slope. The threshold values for
researchers, academicians, consulting engineers and the warning system were selected based on the soil
contractors alike. The paper explains the procedures strainmeters background readings.
followed for safe installation of the instrumentation over
the failed slope. It also outlines the techniques used for
remote management of the instrumentation and the
associated early warning alarm system. Site geology
Keywords: failures, monitoring, safety
Boreholes drilled during the site investigations demon-
strated that the slope was underlain by firm, reddish
brown silty clay with some fine to medium sand about
Malaysia has a total land area of about 330 000 square 5 m thick, over firm to stiff, pale red brown clayey silt
kilometres. The length of the highway network totals approximately 20 m thick. The soil units rest on whitish
almost 61 000 km, of which about 15% traverses grey, weak to extremely weak fine grained dyke-rocks
mountainous terrain that features steep to very steep that vary up to 33 m thick. The dyke-rocks pass down-
slopes. Wherever possible roads are built on stable wards into bedrock of Quartz Schist that continues in
slopes but excavated slopes through steep terrain are depth beneath the toe of the slope. The recorded
frequently affected by slope failures especially along water level ranged from 19 to 20 m below the ground
newly cut slopes. surface. The geological profile of the site is illustrated
In January 1996, a soil slope collapsed on the North– diagrammatically in Figure 3.
South Expressway (one of the major highway in
Malaysia) at Gunung Tempurung (Figs 1 & 2). A
temporary diversion lane was constructed to minimize
disruption to the traffic. Realignment of the highway in The soil strainmeters
the vicinity of the landslide was then proposed as a
permanent solution to the problem. The site was The instrumentation system was designed to be capable
stabilized using a rock buttress. of giving early warning to the workers should the slope
During reconstruction of the collapsed slope instru- be expected to fail. In order to achieve this, electrical
mentation was installed on the unstable scarp primarily sensors were used in conjunction with dataloggers
to ensure the safety of the workers during the remedial capable of triggering an alarm in real time when pre-
works. The instrumentation enabled real time monitor- determined threshold values were exceeded. As a further
ing of ground movements and provided early warning safeguard, the instruments were monitored remotely
signs of instability that could be conveyed to the workers from the slope.
for safety purposes. A system was organized to evacuate The soil strainmeter measures the linear movement
the working area under the scarp in the event of across two points on a soil surface. Figures 4 and 5
reactivation of the landslide. shows the soil strainmeter installed at the proposed
Instrumentation programme provided the necessary location. The relative movement of the soil can be
safeguards by measuring ground behaviour with respect measured by installing the soil strainmeter across and
Quarterly Journal of Engineering Geology and Hydrogeology, 34, 299–305. 1470-9236/01 $15.00  2001 The Geological Society, London
300 OMAR HUSAINI AND M. RATNASAMY

180 170

N
190 160
250

150

GUNUNG TEMPURUNG
SS-5 SS-8
SS-6
240 SS-7 SS-10
SS-4
SS-9
SS-3
SS-1 SS-2
220

190
210
180

140

220

LEGEND:
140 SS - Soil Strainmeter
- Borehole
130
180
- North-South Expressway
200 170

130
140
- Contour

SCALE: 1 : 2000

Fig. 1. Site plan of the Grunung Tempurung slope failure and locations of soil strainmeters.

Fig. 2. Slope failure at Gunung Tempurung.

perpendicular to the existing tension cracks. According Two soil strainmeters were installed perpendicular to
to Dunnicliff (1988) the strainmeters are typically used tension cracks on both sides of the failed slope where
for monitoring tension cracks behind slopes and for movement was expected to occur and eight soil strain-
monitoring joint and faults in rock. In this study, the meters were installed at positions where no indications
readings from the soil strainmeters were used to trigger of tension cracks existed (Table 1). The configuration of
the alarm for the early warning monitoring system. The the total ten soil strainmeters is shown in Figure 6.
soil strainmeters are connected to a datalogger via a During installation, one of the anchor points was
multiplexer for a complete datalogging system. embedded in a stable location and the other was
 AN EARLY WARNING SYSTEM FOR ACTIVE LANDSLIDES 301

0 Firm, reddish brown silty CLAY with some fine to

medium sand.
5m

Firm to stiff, pale red brown clayey SILT.

20m

25m

Whitish grey, weak to extremely weak fine-grained

duke (highly weathered)

58m

Quart Schist

Fig. 5. The installation of soil strainmeter.

Table 1. The installation of soil strainmeters

Soil Length of Field features

116m strainmeter soil strainmeter
identifier (m) Tension crack No crack
Fig. 3. The geological profile.

SS1 4.97 √
SS2 4.98 √
SS3 4.99 √
SS4 3.00 √
SS5 4.98 √
SS6 4.97 √
SS7 4.99 √
SS8 5.00 √
SS9 5.00 √
SS10 4.99 √

emphasize the importance of careful installation to

avoid misleading, inaccurate, or false information, and
to ensure value for money.
Fig. 4. Soil strainmeters.

embedded in the place where movement was anticipated. Monitoring system

To avoid disturbance by animals, accidental movement,
and also to minimize temperature effects the sensors A datalogger connected to a computer comprised the
were buried at a shallow depth of about 200 mm. datalogging system (Fig. 7). The soil strainmeters were
Both Abramson et al. (1996) and Wrightman (1974) connected to the datalogger through a multiplexer.
302 OMAR HUSAINI AND M. RATNASAMY

4.990
2.780 4.970 2.190
Strainmeter No : SS1 (Tension Crack) Strainmeter No : SS7 (No Crack)
Date Installed : 08/07/97 Base Reading : 2.50 mm Date Installed : 09/07/97 Base Reading : 2.56mm
Anchor to Anchor : 4.970m Date Taken : 10/07/97 Anchor to Anchor : 4.990m Date Taken : 09/07/97

4.980
5.000
Strainmeter No : SS2 (No Crack)
Date Installed : 08/07/97 Base Reading : 1.95mm Strainmeter No : SS8 (No Crack)
Anchor to Anchor : 4.980m Date Taken : 10/07/97 Date Installed : 09/07/97 Base Reading : 2.80mm
Anchor to Anchor : 5.000m Date Taken : 09/07/97

4.990

Strainmeter No : SS3 (No Crack) 2.950 5.0 m 2.050

Date Installed : 10/07/97 Base Reading : 3.51mm Strainmeter No : SS9 (Tension Crack)
Anchor to Anchor : 4.990m Date Taken : 10/07/97 Date Installed : 09/07/97 Base Reading : 1.90mm
Anchor to Anchor : 5.000m Date Taken : 09/07/97

3.000 m

Strainmeter No : SS4 (No Crack) 4.990

Date Installed : 10/07/97 Base Reading : 2.91mm Strainmeter No : SS10
Anchor to Anchor : 3.000m Date Taken : 10/07/97 Date Installed : 09/07/97 Base Reading : 1.85mm
Anchor to Anchor : 4.990m Date Taken : 09/07/97

4.980

Strainmeter No : SS5 (No Crack)

Date Installed : 10/07/97 Base Reading : 3.06mm
Anchor to Anchor : 4.980m Date Taken : 10/07/97

4.970

Strainmeter No : SS6 (No Crack)

Date Installed : 10/07/97 Base Reading : 3.44mm Fig. 6. The schematic diagram and detailed of strainmeters
Anchor to Anchor : 4.970m Date Taken : 10/07/97 installed (SS1–SS10).

Early warning system

The early warning system consisted of flashing lights and
sirens that were connected to the datalogger. Figure 8
shows the flashing light at one of the warning sign sites
for the workers. The datalogger triggers the alarm as
soon as the readings exceed the set threshold values. The

SS SS

SS SS

Fig. 8. The flashing light.

Datalogger
flashing lights and sirens were installed at the toe of the
slope as well as on the slope itself.

Computer

Results and discussions

The background readings were collected over a period
Fig. 7. The schematic diagram of the monitoring system. of four weeks. Upon completion of background
 AN EARLY WARNING SYSTEM FOR ACTIVE LANDSLIDES 303

Table 2a. Background readings for week 1

Soil strain meter no. Max. reading (mm) Min. reading (mm) Average (mm) Range (mm) Std. deviation (mm)

SS1 3.37 3.25 3.33 0.12 0.031479

SS2 2.59 2.40 2.51 0.19 0.035982
SS3 5.04 4.88 4.94 0.16 0.047260
SS4 4.30 4.18 4.19 0.12 0.031583
SS5 4.19
14.68
9.76 18.87 0.826315
SS6 4.42 4.21 4.24 0.21 0.048406
SS7 495 4.31 4.74 0.65 0.135003
SS8 4.10 3.85 3.93 0.25 0.076749
SS9 2.77 2.72 2.75 0.21 0.047199
SS10 2.84 2.72 2.72 0.12 0.030022

Table 2b. Background readings for week 2

Soil strain meter no. Max. reading (mm) Min. reading (mm) Average (mm) Range (mm) Std. deviation (mm)

SS1 3.38 3.25 3.30 0.12 0.032360

SS2 2.69 2.44 2.51 0.25 0.056299
SS3 5.41 5.04 5.28 0.36 0.108334
SS4 4.79 4.22 4.60 0.57 0.208067
SS6 5.35 4.42 4.96 0.93 0.281098
SS7 4.55 4.22 4.38 0.32 0.086454
SS8 4.10 3.57 3.84 0.53 0.206440
SS9 2.81 2.64 2.77 0.17 0.044605
SS10 2.82 2.68 2.73 0.14 0.025313

Note: SS5 was removed due to the technical problem.

Table 2c. Background readings for week 3

Soil strain meter no. Max. reading (mm) Min. reading (mm) Average (mm) Range (mm) Std. deviation (mm)

SS1 3.41 3.05 3.14 0.37 0.052806

SS2 2.48 2.20 2.34 0.29 0.062842
SS3 2.40 1.95 2.19 0.45 0.106432
SS4 5.04 4.79 4.94 0.25 0.078291
SS6 5.67 5.31 5.47 0.36 0.120563
SS7 4.46 3.98 4.01 0.48 0.034676
SS8 3.78 3.55 3.69 0.23 0.052329
SS9 2.81 2.56 2.66 0.25 0.033069
SS10 2.92 2.72 2.78 0.20 0.036726

Note: SS5 was removed due to the technical problem.

Table 2d. Background reading for week 4

Soil strain meter no. Max. reading (mm) Min. reading (mm) Average (mm) Range (mm) Std. deviation (mm)

SS1 2.84 2.60 2.70 0.24 0.079089

SS2 151.93 151.85 151.90 0.08 0.011206
SS3 2.93 2.64 2.8 0.28 0.054377
SS4 5.16
15.52 2.54 20.68 6.775624
SS6 5.67 5.31 5.44 0.36 0.115421
SS7 4.14 4.02 4.06 0.12 0.030906
SS8 3.57 3.33 3.46 0.25 0.067211
SS9 2.40 2.11 2.30 0.29 0.060666
SS10 2.88 2.68 2.75 0.20 0.057822

Note: SS5 was removed due to the technical problem.

304 OMAR HUSAINI AND M. RATNASAMY

Background Reading for Week 1

6.00

5.00
Reading (mm)
SS1
4.00 SS2
SS3
3.00 SS4
SS6
2.00 SS7
SS8
1.00 SS9
SS10

0.00
14 15 16 17 18 19 20
July 1997
Time (day)

Fig. 9. The background readings.

monitoring the soil strainmeters were zeroed and the (1) Yellow Level(a) ROD of 0.05 mm/min(b) Move-
alarm system activated. The main purpose of the back- ment of 5 mm
ground readings is to establish the degree of noise in (2) Red Level(a) ROD of 0.10 mm/min(b) Movement
instrument signals. By zeroing the soil strainmeters, of 10 mm
relative movements can be readily identified. It is also
important to establish the degree of noise in the soil During the Yellow Level, the alarm audio signal in the
strainmeter signals to ensure threshold levels for alarm control room was set at a continuous low volume and a
triggering represent significant actual movements rather yellow flashing light operated. At the working area no
than simple signal noise, and thereby also prevent false alarm warning was given but the Safety Officer and
alarm triggering. Resident Engineer were informed promptly and work
The background readings results are tabulated in below the slope ceased. The engineering geologist then
Table 2. Values range from 0.12 mm to 0.93 mm and carried out visual inspections of the slope and a site
standard deviation is 0.030022 mm to 0.281098 mm. The meeting was convened to discuss the safety implications.
background reading results are also illustrated in Figure During the Red Level, the audio signal alarm in the
9. Unfortunately, soil strainmeter No 5 (SS5) was control room was set at continuous low volume and the
damaged during installation as a result of site works and red flashing light operates. At the working area a
had to be removed from the system. For all of the other continuous siren and flashing light system was activated
soil strainmeter instruments except soil strainmeter No 6 automatically. The Safety Officer and Resident Engineer
(SS6), no apparent significant movement was recorded. were again informed immediately and all work on the
In the case of SS6 an apparent movement resulted in the site stopped immediately and the workers evacuated to
recording of a cumulative movement of around 2.5 mm. designated safe area. In the remote control room the
Based on the background reading for a period of one engineering geologist continued to monitor the ground
week, a rate of displacement (ROD) of between 0.05 to movements at closely spaced intervals. A site meeting
0.1 mm/min was adopted. Two levels of warning were was eventually called for briefing purposes and to decide
then proposed, the Yellow Level and the Red Level. The further action. The client was also kept informed of all
lower threshold value was set to the Yellow Level and Red Level events. Table 3 shows detailed of the two
triggered the alert alarm only. The higher threshold levels of the warning system and the action to be taken.
value was set to the Red Level and initiated the danger All the soil strainmeters were continuously monitored
alarm. On the Red Level alarm all the workers had to for about six months throughout the reconstruction of
evacuate the site immediately. the collapsed slope. The datalogger was set to log and
The threshold values determined for the two proposed stored data continuously with a scanning rate of 5
levels are as follows: second. The data were transferred to the computer hard
 AN EARLY WARNING SYSTEM FOR ACTIVE LANDSLIDES 305

Table 3. The warning levels and action to be taken

Warning level Criteria Alarm system Action

Yellow (i) Rate of change of displace- (i) In control room: Continuous (i) Report to Safety Officer and
ment of 0.05 mm/min. (ii) Move- low volume audio signal and Resident Engineer. (ii) Stop work
ment of 5 mm at any strainmeter flashing light. (ii) Outside: No below the slope immediately. (iii)
station. Alarm Inspect the slope visually. (iv)
Site meeting to discuss safety
measures.
Red (i) Rate of change of displace- (i) In control room: Continuous (i) Report to Safety Officer and
ment of 0.10 mm/min. (ii) Move- low volume audio signal and Resident Engineer. (ii) Stop All
ment of 10 mm at any flashing light. (ii) Outside: Con- work immediately. (iii) Evacuate
strainmeter station. tinuous siren and flashing light. workers to designated safe area.
(iv) Site meeting for briefing of
further action. (v) Continue
monitoring at shorter interval.
(vi) Inform the client. (vii) Write
report and recommendation.

disk where the readings at ten minutes intervals were workers on a difficult and dangerous site during recon-
stored. struction of the failed slope. Continued operation of the
alarm system is therefore a prerequisite for safe working
conditions.
Conclusions
Acknowledgements. The authors would like to thank the
MTD-RC staff of their support through out this project.
Two levels of warning, Yellow and Red, were proposed
for the monitoring system using soil strainmeters. The
threshold values for the two proposed levels were deter-
mined based on the background readings of week 1. It is References
proposed that for the Yellow Level, the rate of displace-
ment of change is 0.05 mm/min and movement of 5 mm. A, L. W., L, T. S., S, S. & B, G. M.
For the Red Level, the rate of displacement of change is 1996. Slope Stability and Stabilization Methods. John
0.10 mm/min and movement of 10 mm. Wiley & Sons Inc., New York.
The early warning system consisted of flashing lights D, J. 1988. Geotechnical Instrumentation for Moni-
and sirens connected to a datalogger and monitored toring Field Performance. John Wiley & Sons Inc., New
York.
through the computer in the control room. The system
W, J. 1974. Some organization and benefit aspects
served its purpose well as an alert for the workers of instrumentation In: Field Instrumentation in Geo-
should an imminent failure be expected. The real time technical Engineering Symposium. Butterworths, London,
monitoring system is found useful to ensure safety of the 488–496.

Received 24 July 2000; accepted 17 November 2000.