ChapterPDF Available

Training of Sensors for Early Warning System of Rainfall-Induced Landslides: IEREK Interdisciplinary Series for Sustainable Development



Content may be subject to copyright.
Training of Sensors for Early Warning
System of Rainfall-Induced Landslides
Naresh Mali, Pratik Chaturvedi, Varun Dutt, and Venkata Uday Kala
Landslides have been a major issue in the Himalayan
region where slopes are cut and reformed for construction
practices for infrastructure development, deforestation,
and many other human activities. In lieu of the mitigation
measure for rainfall-induced landslides to improve the
factor of safety against failure, several warning techniques
have been suggested. However, they are quite expensive,
resulting in an only limited application for innite slopes.
In lieu of the existing conditions, early warning systems
(EWS) for detecting slope failure using the sensors have
been found to be handy to control the fatality of the
disaster. But, the various sensors have been used for these
warning systems are not unique. Hence, they need to be
trained for each type of soil and other favorable
conditions. For the proposed study, Micro-Electro-
Mechanical Systems (MEMS) based sensors have been
used to predict the slope failures under rainfall conditions
at controlled laboratory scale prototype and to perform a
series of ume tests in order to develop the threshold for
moisture levels and movement that can trigger the slope
Slope-instability Flume test Sensors
Early warning system
1 Introduction
In Himalayan region of India; slope failures predominantly
occur during or immediately after rainfall [14], which leads
to increase in piezometer levels such as rainwater inltration,
thereby, triggering slope failures. Several mitigation tech-
niques have been proposed in order to decrease the effect
due to landslides. However, where the slopes are steep and
extend to greater heights, most of these may be neither
applicable nor economical. In such situations, the early
warning systems for detecting slope failure using the sensors
have been found to be handy to control the fatality of the
disaster. But these warning systems employing various
sensors are not unique. Hence, efforts have been made to
train the sensors for each type of soil and other conditions to
retrieve the data from the eld. During the course of the
study, the multi-disciplinary involvement for bringing out
the sensors, assembly, applications, calibration, testing,
placing, data retrieving and model-based predictions were
developed [1,3,4].
In the proposed study, Micro-Electro-Mechanical Sys-
tems (MEMS) based sensors were used to predict the slope
failures under rainfall conditions at controlled laboratory
scale prototype and to perform ume tests in order to
develop the threshold for moisture levels and movement that
can trigger a slope failure.
2 Materials and Methodology
Most of the slope failures in and around Mandi region of
Himachal Pradesh are shallow failures. The index properties
of soil (Table 1) were determined (IS-2720). The present
methodology developed by performing ume tests (Fig. 1.)
considering various conditions such as dry state and also by
increasing the amount of moisture levels [1,3,4] within the
slope. However, while performing the tests, the amount of
moisture content, displacement, acceleration and velocity
N. Mali (&)V. U. Kala
School of Engineering, Indian Institute of Technology Mandi,
Kamand Campus, Mandi, 175005, Himachal Pradesh, India
P. Chaturvedi
Scientist Dwith DTRL, DRDO, New Delhi, India
V. Dutt
School of Computing and Electrical Engineering, Indian Institute
of Technology Mandi, Mandi, Himachal Pradesh 175005, India
©Springer Nature Switzerland AG 2019
A. Kallel et al. (eds.), Recent Advances in Geo-Environmental Engineering, Geomechanics and Geotechnics,
and Geohazards, Advances in Science, Technology & Innovation,
were monitored using sensors and identifying the response
of soil slope failure. Based on the experimental threshold
sensors data, the SMS alert will be generated.
3 Results and Discussion
Tri-axial accelerometer: It is capable of measuring accel-
eration forces (static and dynamic) by providing simulta-
neous measurements in three orthogonal directions x, y and
z. Thus, the sensor could be used for the analysis of dif-
ferent vibrations experiencedbyastructure.Thissensor
expends two capacitors formed by a moveable plate held
between two xed plates. Under zero net force, the two
capacitors are equal but a change in force causes the
moveable plate to shift closer to one of the xed plates,
increasing the capacitance.
Soil-moisture sensor: It is usually used to detect the soil
humidity. When the soil is wet, the output voltage decreases
but it increases when the soil is dry.
Force sensor: It is a piezo-resistive conductive polymer,
which changes resistance in a predictable manner following
application of force to its surface. Like all resistive sensors,
this requires a relatively simple interface and can operate
satisfactorily in moderately hostile environments.
Tilt sensor: Tilt sensors are devices that produce an
electrical signal, which varies with angular movement.
These sensors are used to measure slope and tilt within a
limited range of motion. They are usually made by a cavity
and a conductive free mass inside, such as a blob of mercury
or rolling ball. One end of the cavity has two conductive
elements (poles). When the sensor is oriented so that its end
is downwards, the mass rolls onto the poles and shorts them,
acting as a switch.
3.1 Soil Moisture Sensor
For sensing the soil-moisture content in percentage, we used
YL-69 module (Fig. 2a). Analog readings for soil-moisture
sensors in dry and completely wet states were 395 (0%
moisture) and 1022 (100% moisture), respectively. To rep-
resent the analog values from the soil-moisture sensor in
percentage, and hence the moisture percentage was calcu-
lated by the following equation.
Moisture Percentage ¼1022 Analog value
1022 395 100 ð1Þ
The results depicted that, based on the performing of
ume tests, initially the tests were conducted for more than
Table 1 Physical properties of
sand Symbol Description Value
Specic gravity 2.62
%Gravel size 4
%Sand size 49
%Silt size 44
%Clay size 3
USCS Soil Classication SP (poorly graded sand)
C (kPa) Cohesion 5
UAngle of internal friction 28°
Fig. 1 Line diagram of
experimental setup (Not to scale)
450 N. Mali et al.
the critical angle the failure is observed at around 12%
moisture contents whereas, the tests conducted for less than
the 17% moisture levels. According to [1,3,4], moisture
content after reaching the threshold value causes the failure
of the slope leading to debris ow.
The failure of slope also depends upon several factors
such as soil type, angle of inclination, ramp surface texture
and others.
3.2 Accelerometer
For the determination of acceleration, GY-61 accelerometer
module was employed. It measures the tri-axial accelerations
in three orthogonal axes (Fig. 2b). This accelerometer is
capable of measuring acceleration in the range of ±3g
(where, g = 9.81 m/s
), in each of the three orthogonal axes.
For the accelerometer placed at the top and base of the pipe,
, and a
refers to the axis perpendicular to the base of
the ramp (pointing upwards), along the width of the ramp
(from right to left), and sloping at an angle with the soil
away from the ramp (his the angle of the ramp with hori-
zontal), respectively. Based on the predened threshold
values, the modules outputs LOW, otherwise, it outputs
HIGH. The threshold value for the digital signal can be
adjusted using the built-in potentiometer. The integration of
different sensors to a microcontroller and then the micro-
controllers connection to an Internet cloud was carried out.
The microcontroller receives data from sensors and it is
connected to a GSM modem for transmitting the sensors
readings to the web server. The values from sensors were
logged onto an in-house developed web server http://www. with the help of GSM Module.
After the successful reception of the data from soil
moisture sensors and accelerometer sensors, the landslide
probability was computed with the weights assigned to
individual sensor values. When the probability value crosses
a prexed threshold (in this study, 85), an alert would be
triggered to the registered users.
3.3 Experiences During the Training Sensors
While Performing the Tests
1. Soil moisture sensor needs to be calibrated frequently for
calculating the soil-moisture percentage.
2. The probes of the soil moisture sensor are at a distance of
37 mm apart, hence the resistance will be created at the
probe, but not in the gap.
3. Calibration should be achieved before placing these
sensors in the soil at the time of testing.
4. Calibration values may not be the same for all the sensors
5. Flex sensors should be supported to avoid getting
detached from the one end of the node.
6. Any change in voltage would affect the tilt sensor.
4 Conclusion
Early Warning System (EWS) architecture is in place for the
people to be alerted about the oncoming disaster. The
threshold values are evaluated from the analysis of logged
0 5 10 15 20 25 30 35 40 45 50 55 6 0 65 70 75 80 85 90 95 100
Moisture Content (%)
Time (min)
Tests condcucted for less than critical angle
Tests condcuted for more than critical angle
Accleration (m/s2)
Timer (min)
Plot showing the moisture levels with
respect to time period
Plot showing acceleration (m/s2) with respect
(a) (b)
to time period
Fig. 2 aPlot showing the moisture levels with respect to time period. bPlot showing acceleration (m/s
) with respect to time period
Training of Sensors for Early Warning System of 451
soil-moisture and soil-movement values, by performing the
set of ume tests on the ramp. Once the soil-moisture or
soil-movement thresholds are reached, the above-mentioned
alert generation unit generates landslide alerts. As soon as
any activity in the area under surveillance of sensors crosses
a pre-determined limit, an alert would be triggered to inform
the concerned people to take necessary steps. Hence, as soon
as any value of the database crosses the threshold level, an
alert is sent via an SMS.
Acknowledgements The authors would like to thank the State
Council for Science, Technology & Environment, Himachal
Pradesh, India, for providing the nancial support to pursue this study.
1. Huang, C.-C., Lo, C.-L., Jangand, J.-S., Hwu, L.-K.: Internal soil
moisture response to rainfall-induced slope failures and debris
discharge. Eng. Geol. 101, 134145 (2008)
2. Fell, R.: Landslide risk assessment and acceptable risk. Can.
Geotech. J. 31, 261272 (1994)
3. Lourenco, S.D.N., Sassa, K., Fukuoka, H.: Failure process and
hydrologic response of a two layer physical model: implications for
rainfall-induced landslides. Geomorphology 73, 115 130 (2006)
4. Wu, L.Z., Huang, R.Q., Xu, Q., Zhang, L.M., Li, H.L.: Analysis of
physical testing of rainfall-induced soil slope failures. J. Environ.
Earth Sci. 73(12), 85198531 (2014)
452 N. Mali et al.
... The porosity values were lying found between 0.30 and 0.45. The results revealed that the fine content was higher for slope failure areas than no-slope failure areas (Giannecchini and Pochini 2003;Kim and Song 2015;Mali et al. 2019), and their range of the soil is represented in Table. 2. The geotechnical parameters were determined through insitu and laboratory testing procedures (refer to Table 2.). ...
... Hence, it is required to monitor the continuous measurement of the in-situ moisture content at different depth levels along with the soil properties of the desired site locations. The monitoring based on the in-situ moisture content at various depths at desired locations subjected to rainfall helps estimate the critical time for evacuation and defines a warning threshold (Dikshit and Satyam 2017;Chaturvedi et al. 2018;Mali et al. 2019). Therefore, it is opined that monitoring in-situ moisture content is essential alongside scientific instruments such as tensiometers, piezometers, rain gauges, inclinometers, tiltmeters and acoustic devices for developing a landslide hazard warning system. ...
Full-text available
The slope failures cause significant damage, and the slope failure assessment can help understand the underlying factors contributing to these disasters. These disaster risks can be reduced through landslide monitoring and generating early warning alerts at vulnerable sites. Furthermore, the parameters to ascertain the instability for a given region were not explored. Therefore, criteria can be evaluated based on the understanding developed from local conditions, as it might be tedious to perform slope stability analysis for such large areas. Thus, an extensive site investigation was conducted in a study area subjected to numerous slope failures in Mandi, India. The primary goal of this study was to investigate the factors contributing to the slope failures. The present study was planned in the area anticipated for anthropogenic landslides due to construction activities. Hence, a series of geotechnical tests were performed, and the corresponding data from 26 soil samples obtained from different sites (with and without slope failures) of the Mandi region were employed for the current study. The geotechnical characteristics were determined using field and laboratory investigations from the collected soil samples, and thereby, their relationship with the slope failure occurrences was evaluated using multivariate correlation analysis. The results revealed that the most influential parameters for slope instability in this study area are saturated permeability, porosity, suction, in-situ density, in-situ water content and the angle of internal friction. Further, a comparison of these parameters, their critical values indicating failures have been presented with different study areas from the literature.
Full-text available
Rainfall is a significant factor that triggers slope failures around the world. This paper reports a series of physical tests, which were conducted to simulate rain-induced slope failures. The experiments dealt with two scenarios including (1) rainwater infiltration into the slope and (2) slope failures induced by artificial rainfall with different initial conditions. Slope deformation and slope failures were observed and possible mechanisms were interpreted based on the experimental results. The results confirm the hypothesis that pore-water pressure and water content in a loose soil slope change rapidly and that water infiltration into cracks in the slope has a great impact on landslide development. The observed slope failures can be divided into three types: overall sliding failure, partial sliding failure and flow slide. The effect of slope gradient, rainfall intensity and distribution of initial suction on the slope deformation and failure process are also summarized for possible applications under the similar conditions.
Field observations and theoretical analysis have been used in the literature to assess slope instability caused by permeability variations. This investigation aims to study the influence of permeability variations on slope behaviour by experimental means. It focuses particularly on the pore water pressure generation in the vicinity of soils with different permeabilities, and the corresponding failure mode. A series of generated failures in a model with 2 soil layers was performed by means of a flume device. The soil layers were made of a medium-sized sand and a fine sand, placed in horizontal layers. A combination of photography and pore water pressure measurements was used to examine the relationship between the pore water pressure generation and failure modes. Experiments were conducted for different arrangements of soil layers (by changing the soil layer position), and infiltration direction (downward infiltration by sprinkling water on the soil, and upward infiltration from the bottom of the lower soil layer).
Predictions of rainfall-induced fast-moving mass flow and/or debris flows require better knowledge of the mechanism controlling the debris discharge of slopes in debris source areas. A series of rainfall tests on 0.32 m-deep, 0.7 m-high, 1.35 m-wide sandy slopes resting on a bi-linear impermeable rigid base was performed. Soil moisture content and solid discharge measurements were performed to gain insights into the rainfall-induced retrogressive slope failure. The solid (or debris) discharge is a result of the wash-out of the fluidized slope toe by the interflow along the soil–bedrock interface. Characteristics of the failure process for the slopes are represented by mass wasting curves or ‘solid discharge (Qs) vs. time (t)’ curves which are functions of the rainfall intensity and/or the cumulative rainfall. The mass wasting curves have inflection points representing transitions from minor toe failures into remarkable retrogressive failures. The first inflection point of the soil moisture (ω) vs. t curve measured at the soil–bedrock interface signaling the arrival of the descending ‘wet front’, may serve as a precursor for predicting the onset of an abrupt solid discharge induced by shallow slope failures. The time of peak water content measured at the soil–bedrock interface may approximate the time of 5% total solid volume discharge. Up to the time of 5% of total slope volume discharge, a fully saturated state (Sr ≒ 100%) was never observed at the 0.2 m-below-surface zone; however, it was observed along the soil–bedrock interface at near-toe zone of the slope, regardless of the intensity of rainfall investigated. Retrogressive failures were essentially associated with nonuniformly distributed water content in the slope. For both the 0.2 m-below-surface zone and the soil–bedrock interface, a more uniform distribution of Sr along the full height of the slope was found for slopes subjected to high rainfall intensities of 47 and 65 mm/h than that for the slope subjected to a low rainfall intensity of 23 mm/h. At the inflection point of the Qs vs. t curve and 5% of total solid volume discharge, values of Sr at a certain distance from the toe for the soil–bedrock interface were higher than those measured at the same distance from the toe for the 0.2 m-below-surface zone, indicating the effect of infiltration-induced interflow along the soil–bedrock interface and its effects on the fluidization of the slope toe and the retrogressive failure of the slope.
Definitions for risk and hazard which are suited to landslide risk assessment are presented. Acceptable risk is discussed in relation to other risks accepted by the community, and acceptable specific risks are proposed, depending on whether the landsliding is natural or sliding of a man-made slope. Methods for quantifying the risk are discussed, and qualitative definitions are suggested for use when these are desirable. Examples are given of use of risk assessment in areas affected by landsliding and debris flows.