Invited Lecture. National Seminar Slope 2000, 27 April 2002, Bandung, Indonesia. In Paulus P Rahardjo (edt.) Slope 2002.

Proceedings of the National Seminar, Slope 2002. 27-April 2002. Bandung, Indonesia. 1529.


15

1 INTRODUCTION
The occurrence of rainfall-induced landslides in re-
sidual soil slopes is a problem encountered in many
tropical regions with abundant rainfalls. Further-
more, the increasing rate of urbanisation has in-
creased hillside developments for engineered and fill
slopes in many regions in the tropics. The analyses
of the stability of these slopes involve unsaturated
soils because the water table is usually deep. Cli-
matic changes directly affect the unsaturated soil
zone. It is important to note that rainfall-induced
slope failure involves infiltration through the unsatu-
rated zone above the ground water table. Therefore,
a slope should be considered as an integral system of
unsaturated-saturated soils in the stability analyses.
The objective of this paper is to outline the basic
procedure for assessing stability of residual soil
slopes involving saturated-unsaturated soils. The
relevant theory and measurements associated with
the unsaturated soil for slope stability assessment are
briefly introduced. Characterization of soil proper-
ties, flux boundary conditions, field instrumentation
and interpretation of field measurements for slope
stability assessment are discussed through example
applications.


2 THEORY
2.1 Stress State Variables
Stress state variables are stress variables that define
the stress condition in a soil and it allows the trans-
fer of theory between saturated and unsaturated soil
mechanics. There are two independent stress state
variables for an unsaturated soil; namely net normal
stress, ( u
a
), and matric suction, (u
a
u
w
), where
is total stress, u
a
is pore-air pressure, and u
w
is
pore-water pressure. Under a special condition,
when a soil is saturated, the pore-water pressure
equals the pore-air pressure and the matric suction,
(u
a
u
w
), diminishes. The stress state variables then
revert to a single effective stress, ( u
w
). The stress
state variables control the shear strength and volume
change behavior of the soil.
2.2 Soil-Water Characteristic Curve
The soil-water characteristic curve (SWCC) relates
the water content of a soil to matric suction and it is
an important relationship for an unsaturated soil.
The SWCC essentially shows the ability of an un-
saturated soil to retain water under various matric
suctions. It has a similar role as the consolidation
curve of a saturated soil that relates void ratio or wa-
ter content to effective stress. The SWCC of a soil
dictates the manner by which the permeability, shear
strength and volume change of the soil will behave
Studies of rainfall-induced slope failures
H. Rahardjo, E.C. Leong & R.B. Rezaur
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore






ABSTRACT: Rainfall-induced slope failures are common problems in steep residual soil slopes in the trop-
ics. The characteristics of water flow, pore-water pressure changes, and shear strength of soils are the main
parameters associated with slope stability analysis involving unsaturated soils which are directly affected by
the flux boundary condition (infiltration, evaporation) at the soil-atmosphere interface. Procedures for slope
stability analyses considering unsaturated-saturated soils as an integral system are presented. The paper also
highlights procedures and the importance of characterizing soil properties, flux boundary conditions, spatial
and temporal variability of pore-water pressures and instrumentation for slope stability assessment. It has also
been shown through examples how these procedures can be used to generate stability assessment charts for
local slopes under local environmental considerations.




16
at different matric suctions upon drying and wetting
(Fredlund and Rahardjo, 1993a).
The SWCC of a soil can be obtained using a pres-
sure plate extractor as shown in Figure 1 and the
principle of pressure plate test involving surface ten-
sion of water is explained in Figure 2.



2.3 Shear Strength
The shear strength of an unsaturated soil can be ex-
pressed in terms of the independent stress state vari-
ables as follows (Fredlund et al., 1978):
( ) ( )
b
f w a
f
a f ff
u u u c + + = tan tan (1)
where: c' = intercept of the 'extended' Mohr-
Coulomb failure envelope on the shear stress axis
where the net normal stress and the matric suction
are equal to zero (It is also referred to as 'effective
cohesion'); (
f
u
a
)
f
= net normal stress at failure;
(u
a
u
w
)
f

= matric suction at failure; ' = angle of
internal friction associated with the net normal stress
(
f
u
a
); and
b
= an angle indicating the rate of in-
crease in shear strength relative to matric suction,
(u
a
u
w
).
The extended Mohr-Coulomb failure envelope is
defined as a three-dimensional surface tangent to the
Mohr circles at failure with the shear stress, , as the
ordinate and the two stress state variables, ( u
a
)
and (u
a
u
w
) as abscissas. At saturation, the matric
suction becomes zero because the pore-water pres-
sure, u
w
approaches the pore-air pressure, u
a
. As a
result Equation 1 reverts to the shear strength equa-
tion for saturated soils. The extended Mohr-
Coulomb failure envelope can be obtained by con-
ducting triaxial tests using a modified triaxial cell
(Figure 3) for testing unsaturated soils.


2.4 Permeability
The flow of water through an unsaturated soil can be
described by Darcys law (Childs and Collis-George,
1950):
y
h
k v
w
w w

= (2)
where: v
w
= flow rate of water; k
w
= coefficient of
permeability with respect to the water phase;
h
w
/y = hydraulic head gradient in the y-direction;
h
w
= hydraulic head (i.e., the sum of the elevation
and pore-water pressure heads or y + u
w
/(
w
g);
w
=
density of water; g = acceleration due to gravity.
Because the ability of an unsaturated soil to retain
water varies with the matric suction, unlike the satu-
rated zone the coefficient of permeability k
w
, in an
unsaturated zone is not a constant, it is rather a func-
tion of matric suctions. The unsaturated water coef-
ficient of permeability can be indirectly estimated

Figure 1. Schematic diagram of a pressure plate extractor (from
Rahardjo et al., 2001)

Figure 2. Schematic diagram of air-water interface in a
pressure plate test (from Aung et al., 2001)
Figure 3. Modified triaxial cell for testing unsaturated soils
(from Rahardjo et al., 2001)



17
from the SWCC and the saturated coefficient of
permeability, k
s
. As SWCC can be determined with
greater reliability and in a shorter time this indirect
method of obtaining the permeability is attractive.
Since water can only flow through the water-filled
pores, the SWCC therefore, essentially indicates the
space available for the water to flow through the soil
at various matric suctions. Therefore, the shape of
the SWCC can be used to estimate the variation in
the water coefficient of permeability with respect to
matric suction or the permeability function. Indirect
method of obtaining the permeability function is de-
scribed in detail by Fredlund and Rahardjo (1993b).
A permeability function estimated using the SWCC
and k
s
is shown in Figure 4. Relationship between
soil-water characteristic curve and shear strength for
a sand and a clayey silt is shown in Figure 5.


2.5 Seepage Analysis
Seepage analyses are performed to calculate the
pore-water pressure changes in a slope due to rain-
fall. The governing equation for the flow of water
through an isotropic soil can be formulated using
Darcy's law as follows (Fredlund and Rahardjo
1993b):
t
h
g m
y
h
k
y x
h
k
x
w
w
w w
w
w
w

=
|
|
.
|

\
|

+ |
.
|

\
|

2
(3)
where: x and y = cartesian coordinates in the x- and
y-direction, respectively;
w
= density of water; g =
gravitational acceleration;
w
m
2
= coefficient of water
volume change with respect to a change in matric
suction (u
a
u
w
), or the slope of the SWCC.


A soil slope can be considered as an unsaturated-
saturated soil system. The changing pore-water pres-
sures in the slope under the influence of infiltration
can be analyzed using Equation (3). The computed
pore-water pressures can then be input to a slope
stability analysis, in order to assess the variation in
the factor of safety of the slope under different rain-
fall conditions.
2.6 Stability Analysis
Once the possible range of pore-water pressure
changes in a slope during a rainfall is established
either from seepage analyses or from field monitor-
ing, the stability of a slope can be assessed using the
limit equilibrium analyses. The factors of safety with
respect to moment equilibrium, F
m
and force equilib-
rium, F
f
can be written as follows (Fredlund and Ra-
hardjo 1993b):


Figure 4. Relationship between soil-water characteristic curve
and coefficient of permeability for a sand and a clayey silt
(from Fredlund & Rahardjo, 1993b)



Figure 5. Relationship between soil-water characteristic curve
and shear strength for a sand and a clayey silt (from Fredlund
& Rahardjo, 1993b)



18
Nf Wx
R u N R c
F
b
w
m

(
(

+
=
' tan
' tan
tan
'
(4)

(
(

+
=
sin
cos ' tan
' tan
tan
cos '

N
u N c
F
b
w
f
(5)
where: c' = effective cohesion, = sloping distance
across the base of a slice, R = the radius for a circu-
lar slip surface or the moment arm associated with
the mobilised shear force on the base of each slice, N
= the total normal force on the base of the slice, W =
the total weight of a slice of width b and height h,
x = the horizontal distance from the centreline of
each slice to the centre of rotation or to the centre of
the moments, f = the perpendicular offset of the
normal force from the centre of rotation or from the
centre of moments, =the angle between the tan-
gent to the centre of the base of each slice and the
horizontal.
The slope is again treated as an unsatu-
rated-saturated soil system. The
b
angle describes
the change in shear strength as a result of matric suc-
tion changes that can occur during a rainfall. The
b

angle is equal to ' when the soil is saturated or the
pore-water pressure is positive. The
b
angle gener-
ally decreases when the soil is unsaturated or the
pore-water pressure is negative. Both Equations (4)
and (5) indicate a decrease in factor of safety as the
pore-water pressure, u
w
, increases during rainwater
infiltration.
3 CHARACTERISATION OF SOIL
Soil properties effect the SWCC, permeability, shear
strength and hence the stability of a residual soil
slope. Therefore, it is essential to characterize the
soil properties for stability assessment. A typical six-
grade soil profile characterization for residual soil
based on bedrock information and weathering pro-
file condition, proposed by Little (1969) is explained
in Figure 6. A comprehensive characterization for a
residual soil is illustrated in Figure 7.
Six residual soil slopes were selected for rain-
fall-induced slope failure studies in Singapore. The
locations of the slopes are shown in Figure 8.
Yishun, Mandai, NTU-CSE, NTU-ANX and
NTU-IHPT93 slopes were intended for comprehen-
sive instrumentation while NTU-FYP slope was se-
lected for small scale point measurement studies.
Yishun and Mandai slopes were located in the Bukit
Timah granitic formation. The other four slopes
were located in the Jurong sedimentary formation.
(Figure 8). For illustration, only results from the Ju-
rong sedimentary formation are presented in the sub-
sequent sections.




Figure 6. Typical weathering profile of residual soil (from
Little, 1969)

Figure 7. Porous saprolite soil from basalt near Londrina, Bra-
zil (from Vargas, 1985)

Figure 8. Location of the study slopes, generalized geological
map of Singapore, schematic diagram of relative position and
arrangement of field instruments (from Rezaur, et al. 2002)



19
Ground investigations were performed in these
slopes by drilling boreholes to establish the strati-
graphy of the slope, locate the groundwater table and
obtain undisturbed soil samples for laboratory test-
ing. At least one borehole was located at both the
crest and toe of the slopes. Additional boreholes
were drilled on the slope face depending on the
length of the slope face and the uniformity of the
soil in the slope. At least one of the boreholes was
advanced to the depth of the groundwater table or to
the bedrock interface to identify the bedrock forma-
tion. Continuous undisturbed sampling was per-
formed in each borehole using a Mazier sampler.
By combining the information of both field inves-
tigations and laboratory tests, simplified soil profiles
were produced to provide convenient reference for
soil layering and soil properties at the research
slopes. An example of a simplified soil profile for
the NTU-CSE slope is shown in Figure 9.


3.1 Ground investigation results
The variation of index properties with depth and
hence with the degree of weathering of residual soils
from the Jurong sedimentary formation are shown in
Figure 10. This section illustrates how the ground
investigation results are assessed for use in the seep-
age analyses.
3.1.1 Specific Gravity
Specific gravity depends on the mineralogy of soil
and it can reflect the history of weathering (Tuncer
and Lohnes, 1977). The mineralogy of residual soils
varies considerably depending on parent rocks and
weathering process. Specific gravity of residual soils
from the Jurong sedimentary formation ranges from
2.65 to 2.75 and its value increases slightly with
depth (Figure 10). This could be accounted for by
the presence of minerals other than Quartz in deeper
layers that have a higher specific gravity than Quartz
(Aung, 2001).


3.1.2 Void Ratio and Total Density
Weathering leads to a porous structure due to con-
siderable leaching of minerals in the soil. Void ratio
appears to decrease with depths (Figures 10) reflect-
ing the variation in the degree of weathering. Water
and air replace the soluble minerals resulting in a
porous structure. In the upper layers of residual
soils, porosity and void ratio are higher therefore
water and air phase occupy more space as compared
to the lower layers. As a result total density is lower
near the surface. At deeper depths, porosity de-
creases resulting in an increase in total density.
Therefore, the variation in total density as well as
variation in dry density reflects the variation in de-
gree of weathering. The total density of residual
soils from the Jurong sedimentary formations ranges
from 1.9 to 2.5 Mg/m
3
.
3.1.3 Atterberg Limits
In general, plasticity index (liquid limit minus plas-
tic limit) of residual soils decreases with depth as the
degree of weathering decreases. In addition, natural
water content is very close to or less than plastic
limit throughout the depth indicating the unsaturated

Figure 9. Idealized soil profile of the NTU-CSE slope (from
Lee et al., 1999)

Figure 10. Results of index property tests on residual soils from
different depths of the sedimentary Jurong formation (from
Rahardjo et al. 2002)



20
condition of the residual soils. However, the de-
creasing trend of plastic limit and water content with
depth is not obvious in the Jurong sedimentary for-
mation (Figure 10). This could be due to less weath-
ering in the soil profiles of the Jurong sedimentary
formation at this particular site.
3.1.4 Soil-water characteristic curve & permeabil-
ity function
The soil-water characteristic curve (SWCC) is a
measure of the water storage capacity of soil at vari-
ous suction values. The ability of a soil to retain wa-
ter varies with soil suction. Therefore, the coefficient
of permeability is not a constant in unsaturated soils
but is a function of soil suction because water can
only flow through water-filled pores. Suction at
which a soil starts to become unsaturated is called
air-entry value. The air-entry value of a soil depends
on the soil structure. Its value varies with grain size
distribution as well as pore size distribution. The
coarser the grain size and the larger the inter-particle
and intra-particle pores of a soil, the lower is the
ability of the soil to retain the saturated condition.
When a soil becomes more weathered, more clay
minerals are formed and the ability of the soil to re-
tain water under high matric suctions increases or
the air-entry value of the soil increases. Therefore,
the SWCC of a soil reflects its pore size distribution
and hence the degree of weathering.


The SWCCs of the residual soils from different
depths of the Jurong sedimentary formation are
shown in Figure 11. Figure 11 shows that the satu-
rated volumetric water content and the air-entry
value decrease with depth as the degree of weather-
ing decreases or the pore sizes decrease except for
the soil from 13-14m depth. This inconsistency may
be due to the difference in the parent rock type. The
air-entry value of the residual soils from the Jurong
sedimentary formation was found to be around 300
kPa. The void ratio varies with depth from 0.5 to
0.2. The permeability functions for residual soils
from different depths of the Jurong sedimentary
formations estimated using the SWCC of Figure 11
and k
s
are shown in Figure 12.


3.2 Shear strength
A summary of shear strength parameters of residual
soils from the Jurong sedimentary formations is
shown in Table I. The shear strength parameters in
general show an inverse relationship with the degree
of weathering. While the degree of weathering gen-
erally decreases with depth, the shear strength pa-
rameters of residual soils appear to increase with
depth.


The effective cohesion c' of Grade IV residual
soils at 3 m to 4 m depths was found to be very high
(i.e., 125 kPa) and the corresponding effective fric-
tion angle is 42 (Table I). This could be attrib-
uted to the very high bond strength among individ-

Figure 11. Soil-water characteristic curves of residual soils
from different depths of the Jurong sedimentary formation
(from Rahardjo et al. 2002)

Figure 12. Permeability functions for residual soils from the
Jurong sedimentary formation at different depths (from
Rahardjo et al. 2002)
Table I. Summary of shear strength parameters for residual
soils from different depths of the Jurong sedimentary
formation (from Rahardjo et al. 2002)
Depth
(m)
Soil type Grade c
(kPa)

()

b
()
Jurong Sedimentary Formation
3-4 Purple Silty Sand IV 125 42 N.A
4.5-5.5 Purple Silty Sand IV 55 51 40
9-10 Orange Silty Sand IV 35 45 35
22-24 Purple Silty Sand III 225 50 N.A



21
ual particles of the weathered rocks of Grade IV. For
the weathered rocks between 4.5 m to 5.5 m depths,
the effective cohesion decreases to 55 kPa and the
effective friction angle increases to 51. This de-
crease in effective cohesion and corresponding in-
crease in effective friction angle could probably be
due to a change in grain size between these two lay-
ers (34 m and 4.55.5 m depths). The shear
strength parameters of Grade IV highly weathered
sandstone at 9 m to 10 m depths are lower than those
at 4.5 m and 5.5 m depths, perhaps due to a weaker
bond strength. The shear strength of Grade III mod-
erately weathered sandstone is very high. The effec-
tive cohesion for this weathered sandstone is as high
as 225 kPa and the effective friction angle is 50.
4 FLUX BOUNDARY CONDITIONS
Flux boundary conditions control the response of a
slope to climatic changes. Infiltration, runoff, evapo-
ration and subsequent changes in pore-water pres-
sure, shear strength of soil, groundwater table fluc-
tuation and subsurface water movement are influ-
enced by the flux boundary conditions. Therefore, it
is important to evaluate these parameters for slope
stability assessment.



4.1 Infiltration
Prior to the full-scale instrumentation of a slope it is
desirable to have a preliminary qualitative under-
standing of the infiltration capacity and transient re-
sponse of matric suctions to an advancing wetting
front in the instrumented slopes. This could be
achieved by performing a double-ring infiltration
test in conjunction with a few quick draw tensiome-
ters as shown in Figures 13 and 14. The layout of the
infiltrometer and tensiometers is shown in Figure 15.
The study can be performed on a vegetated and a
bare surface simultaneously. Results from such ex-
periments on a flat grass-covered surface are shown
in Figure 16. Figure 16a shows the cumulative infil-
tration rate at various times obtained from the dou-
ble-ring infiltrometer test. Figure 16b shows the re-
sponse of the tensiometers at different depths to the
advancing wetting front. Such tests are useful to as-
sess the infiltration rate and permeability of the soil
at shallow depths. For example, from the response
time of the tensiometers to the infiltrating water at
two different depths (Figure 16b) it is possible to
calculate the velocity of water movement in the ver-
tical direction. The permeability of the soil can be
calculated from Darcy's law.






Figure 13. Schematic diagram of an open double-ring infiltra-
tion test (from Agus et al., 1999)

Figure 14. Double-ring infiltrometer and quick draw tensiome-
ter assembly to study infiltration capacity and transient re-
sponse of matric suction to an advancing wetting front in
NTU-CSE slope (from Kumar & Ganesh, 1996)

Figure 15. Plan view and sectional view of layout of
double-ring infiltrometer and quick draw tensiometers for a
synchronised measurement of infiltration capacity and matric
suction response to an advancing wetting front (from Kumar &
Ganesh, 1996)



22




It is not always feasible to conduct a detailed and
full-scale comprehensive field instrumentation for
long-term monitoring because of budget constraints.
Under such circumstances point measurements with
simulated rainfalls can be performed. Such studies
although are not comprehensive and lack the fea-
tures of a full-scale study can still provide meaning-
ful results from small plots within a short time. Such
rainfall simulations speed up the process of data col-
lection, as it is not necessary to wait for a natural
rainfall. The installation of tensiometers can be con-
fined to a small plot (1 1 m
2
) as shown in Fig-
ure 17 rather than covering the entire slope. Results
from a point measurement study on a small plot with
simulated rainfalls at the NTU-FYP slope are shown
in Figure 18.


4.2 Runoff
Field tests can also be carried out on the instru-
mented slopes in order to measure the infiltration
characteristics on a slope surface. The rainfall-
runoff-infiltration characteristics of a slope can be
studied on an isolated plot of the slope (Figure 19).
The runoff is measured using a calibrated flume with
a capacitance water depth probe (Figure 20). The
study can be conducted under a simulated rainfall
using sprinklers (Figure 19) or a natural rainfall. The
uniformity in rainfall application (simulated) can be
assessed by collecting rainfall in a number of small
cans spread randomly over the area and comparing
the rainfall collected in the cans for a fixed period of
rainfall application.


(a) Cumulative infiltration versus time

(b) Matric suction versus time
Figure 16. Cumulative infiltration rates from a double-ring
infiltration test on a flat grass-covered surface and tensiometer
responses at various depths to the advancing wetting front due
to infiltration (from Kumar & Ganesh, 1996)

Figure 17. Small scale point measurements with simulated
rainfall (from Teo & Yap, 1993)

Figure 18. Typical results from small scale point measurement
studies at NTU-FYP slope. Illustration shows total head varia-
tions with time at different depths subjected to a simulated
rainfall of 196 mm h
1
for a period of 32 minutes (from Teo &
Yap, 1993)



23


Figures 21 and 22 show the runoff hydrograph
from a simulated and a natural rainfall, respectively.
Such runoff measurements and hydrographs allow
for indirect assessment of total infiltration into the
slope resulting from rainfall events. The infiltration
rates shown in Figure 21 were derived as the differ-
ence between rainfall and runoff rates assuming in-
terception losses to be negligible. Such hydrographs
resulting from several rainfall events can then be
used to assess infiltration into a slope as a percent-
age of rainfall amounts as shown in Figure 23. In
Figure 23, infiltration (as a percentage of total rain-
fall) is plotted against rainfall amount records from
natural rainfalls monitored in the NTU-CSE slope. It
appears from Figure 23 that rainfall events which
produce small total amount of rainfall may contrib-
ute fully to infiltration.
Figure 23 also suggests the existence of a thresh-
old rainfall amount. Any rainfall below this amount
will not produce any runoff and the whole rainfall
may end up as infiltration. With reference to Figure
23 (dotted line) this threshold appears to be about 10
mm of total rainfall. Beyond the threshold rainfall,
the percentage of rainfall contributing to infiltration
decreases with increasing total rainfalls. The infiltra-
tion amount could decrease to about 40% of the rain-
fall (Figure 23) for rainfall events producing a
higher total amount. This however, does not mean
that the total infiltration amount is less with rainfall
events producing high total rainfall than rainfall
events producing small total rainfalls. For example,
with 40% of the rainfall contributing to infiltration a
100 mm rainfall (higher total amount) would result
in 40 mm total infiltration. While with 100% of the
rainfall contributing to infiltration a 10 mm rainfall
would produce only 10 mm of total infiltration.



The data suggest that in residual soil slopes total
infiltration could range between 40% to about 100%
of the total rainfall depending on the rainfall amount.
The relationship (Figure 23) derived from the rain-
fall records in the residual soil slope has practical
significance. If the rainfall amount is known, Figure
23 could indicate the fraction of the rainfall that
could become infiltration. This information can then
be used in seepage analyses as flux boundary condi-
tions.


Figure 19. Field plot for studying rainfall, runoff and infiltra-
tion processes (from Lee et al., 1999)

Figure 20. Field installation of a flume with water depth probe
for runoff measurement (from Lee et al., 1999)

Figure 21. Runoff measurements from a simulated rainfall
event (from Lee et al., 1999)

Figure 22. Runoff measurement from a natural rainfall event
(from Lee et al., 1999)



24

4.3 Evaporation
Evaporation from a slope also plays an important
role in matric suction recovery. Potential evapora-
tion from a slope can be measured using a Penman
pan shown in Figure 24. However, the actual evapo-
ration depends on the negative pore-water pressure
near the soil surface (Figure 25).



5 INSTRUMENTATION
Instrumentation is an important aspect of stability
assessment against rainfall-induced slope failures.
The instruments involved are a rain gauge, piezome-
ters and tensiometers for measuring positive and
negative pore-water pressures, respectively. A
schematic diagram of relative position and arrange-
ment of field instruments in one of several instru-
mented slopes in Singapore is shown in Figure 8.
The mechanism of water flow within a slope dur-
ing evaporation and infiltration can be observed
through continuous real-time monitoring of the pore-
water pressure and rainfall events as illustrated in
Figure 26a. The groundwater table plays an impor-
tant role in governing the suction profile in the
deeper soil layers. Typical results from piezometer
readings in conjunction with rainfall at the
NTU-CSE slope are shown in Figure 26b. The pore-
water pressure distribution can then be used to as-
sess the stability of the slope over a period of time
(Figure 27).




Figure 23. Percentage of infiltration as a function of rainfall
amount (from Lee et al., 1999)

Figure 24. Evaporation pan (from Rahardjo, 2001)

Figure 25. Ratio of actual evaporation (AE) and potential
evaporation (PE) versus total suction for different soils (from
Wilson et al., 1997)

(a) Pore-water pressure variation with rainfall

(b) Ground water table variation with rainfall
Figure 26. Time series of pore-water pressure, rainfall and
ground water table from NTU-CSE slope (from Hritzuk, 1997)



25


The effect of soil cover can also be studied from
the instrumented slopes. Figure 28 shows the re-
sponses of NTU-IHPT93 slope under three different
surface covers to the same rainfall events. The effect
of vegetation on negative pore-water pressures in a
slope can be observed from the cyclic pattern of the
pore-water pressure within the root zone of a tree
(01.4 m depth) as illustrated in Figure 29.



Spatial and temporal variability in data causes
difficulty in representing the soil profile with a de-
terministic or a precisely defined set of hydraulic
parameters. Time series of pore-water pressure data
from an instrumented slope can also be used to ex-
amine the spatial and temporal variability in
pore-water pressures across the slope profile as
shown in Figure 30. Using autocorrelation analyses
the 'range' (a measure of the length of influence, for
Figure 27. Variation in factor of safety with time and rainfall
(from Rahardjo et al., 2000)

Figure 28. Changes of in situ total head profiles in response to
rainfall under different conditions of slope cover (from Lim et
al., 1996)

Figure 29. Effect of a Rain Tree on pore-water pressures (from
Gasmo et al., 1999)


(a) Distribution of range along slope length (c) Distribution of
sill along slope length



(b) Distribution of range along slope depth (d) Distribution of
sill along slope depth
Figure 30. Distribution of range and sill parameters along the
slope length and depth in NTU-CSE slope (from Rezaur et al.,
2002)



26
time series data) and 'sill' (a measure of temporal
variability, for time series data) were calculated for
the time series of pore-water pressures in NTU-CSE
slope. The spatial distribution of range and sill
across slope length and depth are shown in Fig-
ure 30.
6 ASSESSMENT OF SLOPE STABILITY
A parametric study was performed to examine the
effect of infiltration on the stability of unsaturated
residual soil slopes by Rahardjo et al. (1999, 2000).
A finite element seepage program, SEEP/W
(Geo-Slope, 1998a), was used to simulate rainwater
infiltration into a slope and generate transient
pore-water pressure distributions for different com-
binations of slope angle and height. The limit equi-
librium slope stability model, SLOPE/W
(Geo-Slope, 1998b), was then used to determine the
critical slip surface and the factor of safety for the
transient pore-water pressure distributions calculated
by SEEP/W. This provided information about the
effect of infiltration on slope stability. The results
from the parametric study were used to develop
charts to provide preliminary assessments of slope
stability against rainfall-induced slope failure for
typical residual soil slopes in Singapore.
In the parametric study a combination of four dif-
ferent slope angles of 18, 27, 45 63 (3H:1V,
2H:1V, 1H:1V, 0.5H:1V) and three different slope
heights of 10, 20, and 40 m were analyzed. A gener-
alized geometry of the slope model used in the pa-
rametric study is shown in Figure 31. The slope ma-
terial was assumed to be homogeneous, single lay-
ered with a saturated permeability of 1.010
4
m s
1
.
The shear strength properties of the slope material
are given in Table II. The SWCC and the permeabil-
ity function are shown in Figures 32 and 33, respec-
tively.





The initial pore-water pressure distribution for
each slope configuration was defined as a hydro-
static pore-water pressure profile with a limiting
negative pore-water pressure of 75 kPa. This limit
was selected based on field measurements of nega-
tive pore-water pressures. The top boundary was
specified as a flux boundary. A rainfall intensity of
80 mm hr
1
(2.210
5
m s
1
, a typical rainfall rate
during heavy rainfalls in Singapore) was applied
from time equal to 0 hour until time equal to 4
hours. A rainfall intensity of about zero (1.010
12

m s
1
) was then applied for a further period of 10
days to allow the water that infiltrated into the slope
to redistribute. The top boundary condition and all
other parameters (except slope angle and height)
were the same for all analyses.
Figure 31. Generalized geometry of slope model for parametric
study (from Rahardjo et al., 2000)
Table II. Shear strength parameters for the parametric study
Combination c
(kPa)

(deg)

b
(deg)

kN/m
3
1 10 26 26 20


Figure 32. Soil-water characteristic curve of residual soil from
Jurong sedimentary formation (from Rahardjo et al., 2000)

Figure 33. Permeability function computed from the drying
soil-water characteristic curve of Figure 33 (from Rahardjo
et al., 2000)



27
The results from the seepage analyses of the pa-
rametric study show that the slope height, slope an-
gle and the transient pore-water pressure distribution
influence the stability of a slope. Infiltration charac-
teristics were similar for each of the slopes. Exam-
ples of pore-water pressure and hydraulic head dis-
tribution in a slope (angle 27 and height 10 m) after
four hours of rainfall are shown in Figures 34a and
34b, respectively.



Figures 35a and 35b show the critical slip surface
for the 10 m, 27 slope before the rainfall event (i.e.,
time equal to 0 hours) and after four hours of rain-
fall. The critical slip surface was located at a depth
of about 9.4 m at time equal to zero hour (Figure
35a) and the factor of safety was about 2.9 (Figure
35b).
The factor of safety and the respective depth of
critical slip surface over time for 27 slopes with dif-
ferent heights are shown in Figures 36a and 36b, re-
spectively. A few trends are apparent from these
plots. Prior to the commencement of rainfall, slopes
with lower heights had a higher factor of safety
(Figure 36a) and the depths of critical slip circles are
deeper for slopes with greater heights (Figure 36b).
As rainfall continued, positive pore-water pressures
developed (between 1 and 4 hours) and these trends
changed.



Both the stability and the depth of critical slip
surface decreased over time during rainfall due to
infiltration and reached minimum values at the end
of four hours, and increased over time during
pore-water pressure redistribution (between 4 and 28
hours, see Figures 36a and 36b). It is also evident
that after the development of positive pore-water
pressures (time = 4 hours) due to infiltration, all 27
o

slopes regardless of their heights had about the same
factor of safety and the same depth of critical slip
(a). Pressure contours

(b). Total head contours
Figure 34. Pore-water pressure (kPa) and total hydraulic
head (m) distributions in the parametric slope model after four
hours of rainfall with an intensity of 80 mm hr
1
(from
Rahardjo et al., 2000)

(a). Time = 0 hour

(b). Time = 4 hours
Figure 35. Location of critical slip circle for the parametric
slope model at different times subjected to a rainfall intensity
of 80 mm hr
1
(from Rahardjo et al., 2000)



28
surface. This implies that once positive pore-water
pressures have developed, slope height has little ef-
fect on the factor of safety and the depth of critical
slip surface.



These examples show that the transient
pore-water pressure distribution appears to be a key
triggering mechanism behind rainfall-induced slope
failures. Infiltration may lead to the development of
positive pore-water pressure zones in a slope. After a
period of time when the positive pore-water pressure
zone has developed, the critical slip surface of the
slope becomes shallow passing through the positive
pore-water pressure zone.
The results of the parametric study were used to
develop preliminary assessment charts for slope sta-
bility against rainfall. As an example for a typical
residual soil slope in Singapore of 10 m height and
slope angle of 27, the assessment charts are shown
in Figure 37. Figure 37a shows the transient
pore-water pressure distributions in the slope sub-
jected to a rainfall of 80 mm hr
1
for different dura-
tion. Figures 37b and 37c show the respective fac-
tors of safety for different slope angles and heights,
respectively.



(a). Factor of safety

(b). Depth of critical slip circle
Figure 36. Time series of factor of safety and depth of critical
slip circle during rainfall and redistribution for a 27
o
slope
(from Rahardjo et al., 2000)


(a) Transient pore-water pressure distributions in a 10m slope
standing at 27 subjected to rainfall of 80 mm hr
1
and various
duration.

(b) Respective factor of safety versus slope angle calculated
using (a).

(c) Respective factor of safety versus slope height calculated
using (a)
Figure 37. Charts for preliminary assessment of slope stability
(from Rahardjo et al., 2000)



29
7 CONCLUSIONS
Slope responses to rainfall, infiltration and evapo-
transpiration processes are variable in space and
time and are largely influenced by the dynamic
climatic conditions, soil properties and vegetation.
Instrumented slopes can provide an understanding of
slope responses to such processes over a long-time
which is essential for advancing rainfall-induced
slope failure studies.
Slope stability against rainfall should be analyzed
as a dynamic problem involving environmental
changes instead of a static problem as convention-
ally considered. The unsaturated soil zone is the cru-
cial interface between environment and the slope
and unsaturated soil mechanics principles can be
used for realistic assessment of slope stability
against rainfall. The slope itself should be analyzed
as an integral system of unsaturated-saturated soil
where saturated soil is a special case of unsaturated
soil.
ACKNOWLEDGEMENTS
The work was funded by research grants from Na-
tional Science and Technology Board, Singapore
(Grant NSTB 17/6/16: Rainall-Induced Slope Fail-
ures) and the Nanyang Technological University
(NTU), Singapore (Grant RGM 11/96: Slope Stabil-
ity in Tropical Soils). The field assistance from the
Geotechnics Laboratory staff, School of Civil & En-
vironmental Engineering, NTU, in carrying out this
research is gratefully acknowledged.
REFERENCES
Agus, S.S., Leong, E.C., Rahardjo, H. (1999). Field measure-
ments of permeability for residual soils. Proceedings of the
Fifth International Symposium on Field Measurements in
Geomechanics, Singapore, 13 December. pp. 537542.
Aung, K.K., (2001). Engineering Characteristics of Residual
Soil with Respect to Degree of Weathering. M.Eng. Thesis.
School of Civil & Environmental Engineering. Nanyang
Technological University, Singapore. pp. 162.
Childs, E.C., Collis-George, N. (1950). The permeability of
porous materials. Proc. Royal Soc. 201A: pp. 392405.
Fredlund, D.G., Morgenstern, N.R., Widger, R.A. (1978). The
shear strength of unsaturated soils, Canadian Geotechnical
Journal 15(3). 313321.
Fredlund, D.G., Rahardjo, H. (1993a). An overview of unsatu-
rated soil behaviour. Proceedings of the 1993 ASCE Con-
vention on Unsaturated Soils, Dallas, Texas, 2124 Octo-
ber. pp 131
Fredlund, D.G., Rahardjo, H. (1993b). Soil Mechanics for Un-
saturated Soils. John Wiley & Sons, Inc., New York,
pp. 517.
Gasmo, J., Hritzuk, K.J., Rahardjo, H., Leong, E.C. (1999).
Instrumentation of an unsaturated residual soil slope. Geo-
technical Testing Journal 22(2), 128137.
Geo-Slope (1998a). Seep/w Version 4.21 Users Manual, GEO-
SLOPE International Ltd., Calgary, Alberta, Canada.
Geo-Slope (1998b). Slope/w Version 4.21 Users Manual,
GEO-SLOPE International Ltd., Calgary, Alberta, Canada.
Hritzuk, K.J. (1997). The effect of drainage systems in main-
taining soil suction. MEng. Thesis. School of Civil &
Structural Engineering. Nanyang Technological University,
Singapore.
Kumar, K.S., Ganesh, S. (1996). Infiltration of residual soil
ground with different surface cover. Final Year Project Re-
port. School of Civil & Structural Engineering. Nanyang
Technological University, Singapore.
Lee, T.T., Rahardjo, H. Leong, E.C. (1999). Simulation and
natural runoff measurement on residual hillslopes. Proceed-
ings of the Fifth International Symposium on Field Meas-
urements in Geomechanics, Singapore, 13 December. pp.
387391.
Teo, .K.L., Yap, .A.M. (1993). A study on the infiltra-
tion-runoff relationship in a sloping field plot. Final Year
Project Report. School of Civil & Structural Engineering.
Nanyang Technological University, Singapore.
Lim, T.T., Rahardjo, H., Chang, M.F. and D.G. Fredlund
(1996). Effect of rainfall on matric suction in a residual soil
slope. Canadian Geotechnical Journal 33: pp 618628.
Little, A.L. (1969). The engineering clssification of residual
tropical soils. Seventh International Conference on Soil
Mechanics and Foundation Engineering. Mexico, Vol. 1.
pp. 110.
Rahardjo H, Aung, K.K, Leong, E.C, Rezaur, R.B. (2002). Ef-
fects of pore-size distribution on engineering properties of
residual soils. The Second World Engineering Congress.
2225 July. Kuching, Sarawak, Malaysia. (Accepted).
Rahardjo, H. (2000). Rainfall-Induced Slope Failures. NSTB
17/6/16 Main Report. Nanyang Technological University,
Singapore.
Rahardjo, H., Leong, E.C. Deutscher, M.S. Gasmo, J.M. Tang,
S.K. (2000). Rainfall-Induced Slope Failures. Geotechnical
Engineering Monograph 3. NTU-PWD Geotechnical Re-
search Centre, Nanyang Technological University, Singa-
pore. pp. 86.
Rahardjo, H., Leong, E.C., Rezaur, R.B. (2001). Assessment
and instrumentation of slopes for stability against rainfall
Fourteenth Southeast Asian Geotechnical Converence.
Hong Kong, 1014 December. Vol. 1. 475684.
Rezaur, R.B., Rahardjo, H., Leong, E.C. (2002). Spatial and
temporal variability of pore-water pressures in residual soil
slopes in a tropical climate. Earth Surface Processes and
Landforms, 27(3): 317338.
Tuncer, E.R., Lohnes, R.A. (1977). An engineering classifica-
tion for Basalt-derived Lateritic soils. Engineering Geol-
ogy, 4. 319339.
Vargas, M. (1985). The concept of tropical soils. Proceedings
of the First International Conference on Geomechanics in
Tropical Lateritic and Saprolitic Soils. Vol. 3. pp. 101134.
Wilson, G.W., Fredlund, D.G., Barbour, S.L. (1997). The ef-
fect of soil suction on evaporative fluxes from soil surfaces.
Canadian Geotechnical Journal 34(4). 145155.