Journal of Environmental Management 92 (2011) 266274

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

In situ measurement of soil moisture and pore-water pressures in an incipient

landslide: Lake Tutira, New Zealand
Richard Hawke a, *, Jack McConchie b
a
b

Ministry of Economic Development, Wellington, New Zealand

Opus International, Wellington, New Zealand

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 21 October 2008
Received in revised form
5 May 2009
Accepted 12 May 2009
Available online 18 November 2009

The immediate cost of shallow regolith landslides in New Zealand has been estimated to exceed US$33M
annually. Since the majority of these landslides occur during prolonged wet conditions, or intense
rainstorms, moisture conditions are a critical control. The nature, dynamics, and character of soil
moisture conditions, and the piezometric response to rainfall, have been recorded within an incipient
landslide for more than 5 years. The study site, on pastoral hill country within the Lake Tutira catchment
in northern Hawkes Bay, is typical of large areas of New Zealand episodically affected by extensive
landsliding. Detailed continuous measurements show that both the soil moisture and piezometric
response within the regolith are highly storm- and site-specic. The development of positive pore
pressures is infrequent; they form only during intense rainstorms, and persist for a short time. The
hydraulic response of the soil is primarily a function of storm characteristics, but this response can be
modied by antecedent moisture conditions, topographic position, and heterogeneity of soil properties.
Stability analysis shows that most slopes in the study area are signicantly steeper than can be
explained by the frictional strength of the regolith. Measured hydraulic conditions also show that
positive pore-water pressures alone do not trigger slope instability. A recent slope failure followed
a period of extremely high antecedent moisture conditions, and occurred when maximum soil moisture
conditions, though not pore-water pressures, were recorded. Increased moisture content of the regolith
reduces matric tension, and therefore effective cohesion of the soil. This cohesion is critical to maintaining stability of the regolith on these slopes.
2009 Elsevier Ltd. All rights reserved.

Keywords:
Landslips
Triggering
Pore-water pressures
Matric tension
Stability analysis

1. Introduction
Shallow regolith slope failures are a widespread and recurrent
problem in New Zealand, the direct cost of which has been
estimated to exceed US$33M annually (Glade, 1998). While most
slope failures are triggered by the interaction of water with the
slope material, many failure sites are predisposed to instability as
a result of New Zealands dynamic tectonic and climatic history. The
length of time it takes the landscape to adjust to changes in
conditions has led to many slopes being out of equilibrium with the
present topographic and climatic setting. Many slopes are thus
preconditioned for landsliding and require only a small change in
one factor to trigger failure. Failure can be caused by either an
increase in the shearing force (the force tending to make the
material slide down slope) or a decrease in the materials shear

* Corresponding author.
E-mail address: richard.hawke@med.govt.nz (R. Hawke).
0301-4797/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2009.05.035

strength (the properties resisting movement). Water content,

because it varies rapidly, is the most common trigger of these slope
failures (McConchie, 1992).
Considerable research has now quantied the general relationship between rainfall and the triggering of slope failures in New
Zealand (Crozier, 1986; McConchie, 1992; Glade, 1998; Crozier and
Glade, 1999). Similarly, empirical relationships between rainfall
intensity, rainfall duration, and slope instability have been documented from around the world (Iverson and Major, 1987; Haneburg, 1991a, b; Iverson, 2000). From observational data, Haneburg
(1991b) developed an approach to incorporating rainfall inputs, soil
hydraulic properties, and the presence of an impervious surface to
explain pore-pressure uctuations. Such pore-pressure diffusion
models have been used to provide a physically-based method for
linking rainfall with the pore-pressure response in hillslopes
(Haneburg, 1991a; Reid, 1994).
However, debate still exists as to the exact manner in which
rainfall triggers instability. Quantifying temporal changes in the soil
water status, and the spatial distribution of moisture, are therefore

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

essential to understanding the role of landslide-triggering rainfall

thresholds.
The history of rainfall, land-use change, and landsliding in the
Lake Tutira catchment in northern Hawkes Bay (Fig. 1) has been
well documented (Guthrie-Smith, 1969). The last 600 years has
seen the vegetation cover change from indigenous forest to fern/
scrub and then to pasture. Sedimentation rates in Lake Tutira
under present conditions, primarily the result of landsliding, are
seven times those under indigenous forest cover (Eden and Page,
1998). Conversion to pasture has also reduced the return period of
the critical landslide-triggering rainfall. However, there is some
evidence that the rapid removal of soil from the catchment has
led to increasing resistance, such that larger magnitude storms
are now required to trigger slope failure (Crozier and Preston,
1998).
The Lake Tutira catchment has extreme and persistent susceptibility to landsliding. This is a result of the rapid tectonic uplift
rates and soft rock characteristic of the East Coast deformed belt in
which the catchment occurs. Active stream incision has caused
over-steepening of the valley side slopes to between 10 and 55 ,
and dramatic local relief exceeding 250 m. These conditions
predispose the slopes to instability, but rainfall usually acts as the
trigger for failure. While low frequencyhigh magnitude rainstorms led to instability under forest cover, the conversion to
bracken and fern as a result of Maori land-use practices, and, since
the 1900s, conversion to pasture, have demonstrably reduced the
magnitude of slip-triggering storms. This has increased the incidence of landsliding, particularly since European settlement (Eden
and Page, 1998).
The soils prone to failure are derived from weathering of the
Tertiary marine sediments forming the bedrock, tephra from the
Central Plateau, and tephric loess of local origin. In undisturbed
situations the soils are moderately to strongly leached, intergrading between yellowbrown pumice soils and yellowbrown
loams (Pohlen, 1971). They have an A horizon up to 30 cm thick,
a B horizon that may contain pumice from the underlying layer of
tephra, and a C horizon of highly weathered mudstone or
sandstone. The soils are thin, with a median depth of only 75 cm.
The soils are also permeable, variable in physical properties, and
overlie low permeability bedrock. The bedrock interface forms an
abrupt hydraulic discontinuity and acts as the shear plane for
landsliding. As over 80% of the regolith in the catchment has
been disturbed by landsliding (Fig. 1) soil proles are highly
variable.
Daily rainfall has been recorded at Lake Tutira since 1895.
Mean annual rainfall is 1400 mm, with a wet winter (JulyAugust

Fig. 1. Lake Tutira, looking northeast, showing the study site on one of the few
remaining undisturbed regolith surfaces.

267

accounts for 32% of annual rainfall) and a dry summer

(SeptemberNovember accounts for 19% of annual rainfall).
However, rainfall is highly variable. There can be long summer
droughts, and ex-tropical cyclones can cause intense rainstorms,
usually in the period from March to May. For example, in March
1988 Cyclone Bola produced 753 mm of rain (more than half the
average annual rainfall) over a 4-day period. This is the largest
rainstorm event in the 93-year rainfall record. The rainfall event
generated an average sediment yield of 4203 m3/ha, equivalent to
a denudation rate of 42 mm, with 89% of the sediment coming
from landslides (Page et al., 1994). Studies of the climatic record
have shown that 24-h rainfalls exceeding 300 mm trigger landslides (Glade, 1998). Such events have a recurrence interval of
approximately 21 years.
While good conceptual models exist of how, and why, slope
instability occurs; relating these models to the eld situation is
not easy (Beven, 1991). The prediction of landslide activity, for
example, requires the determination of the temporal changes in
the soil water status, and the spatial distribution of the effects of
these changes on material resistance and stresses acting within
a slope. Also, conrming these conceptual models requires
empirical eld data from the incipient landslip immediately prior
to, and preferably during, failure. This paper uses detailed
monitoring records from an incipient landslip to conrm the
triggering mechanism that leads to slope failures in the study
area.
2. Conceptual background
While there is general consensus as to the importance of
water in triggering slope instability there is no unanimity on the
actual process, or causes. Two theories are discussed in the
literature; these being the saturated and the unsaturated models;
both of which are intimately linked to the theoretical understanding of the controls on shear strength. For saturated soils, the
shear strength is computed using the MohrCoulomb failure
criterion and the effective stress concept (Fredlund and Rahardjo,
1993):

sff c0 sf  uw tanf0
f

where
s

shear strength
shear stress on the failure plane at failure
effective cohesion, which is the shear strength
c0
when the effective normal stress is equal to zero
(sf  uw)f effective normal stress on the failure plane at failure
sff
total normal stress on the failure plane at failure
pore-water pressure at failure
uwf
f0
effective angle of internal friction

sff

The saturated model for slope instability argues that while

both cohesion and friction combine to produce resistance, it is the
reduction of the frictional strength that induces failure. As
a saturated zone develops upwards from the shear plane (usually
the bedrock interface) an increasing amount of the normal stress
is carried on the pore uid rather than the particleparticle
contacts. In this manner the effective normal stress, and therefore
resistance to slope failure, is reduced by upthrust from positive
pore-water pressures. Under critical conditions the shear strength
is reduced below the shear stress and failure occurs (McConchie,
1992).
The second model for slope instability includes the additional
strength component provided by the fact that moisture within

268

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

c0

effective cohesion, which is the shear strength

when the net normal stress and the matric suction
at failure are equal to zero
(sf  ua)f net normal stress on the failure plane at failure
pore-air pressure on the failure plane at failure
uaf
f0
angle of internal friction associated with the net
normal stress state variable (sf  ua)f
(ua  uw)f matric suction on the failure plane at failure
fb
angle indicating the rate of increase in shear
strength relative to the matric suction (ua  uw)f
Fig. 2. The instrumented slope showing the location of the sensors and their position
relative to the morphology of the slope and the soil slip that occurred during
monitoring.

the regolith is held under tension (Anderson, 1984; Gan and

Fredlund, 1992; Fredlund and Rahardjo, 1993). As the regolith
absorbs moisture, this cohesion is progressively lost, leading to
a reduction in strength. The shear strength equation for unsaturated soils is more complicated, if only an extension of, the shear
strength equation for saturated soil (Fredlund and Rahardjo,
1993):

sff c0 sf  ua
where


f

tanf0 ua  uw f tanfb

As the soil approaches saturation, the pore-water pressure, uw,

approaches the pore-air pressure, ua, and the matric suction,
(ua  uw), tends toward zero. Hence, the matric suction component
of the shear strength vanishes, and the equation for unsaturated
conditions becomes the same as for saturated conditions.
The effect of this capillary water on cohesion is controlled by
the pore size distribution; smaller pores resulting in greater
additional cohesive strength. As the regolith absorbs moisture,
the smallest pores are saturated rst leading to a very rapid
reduction in cohesion. Given that the variation in tension (cohesion) with moisture content is greatest at the surface this
mechanism might be expected to result in failure towards the top
of the prole, rather than at the bedrock interface as is usually
seen in the eld. There are two reasons for this apparent

Fig. 3. Variation in soil moisture content throughout the prole in response to rainfall over a year.

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

269

Fig. 4. Variation in soil moisture content at 250 mm depth at different sites in response to rainfall over a year.

anomaly. First, the shear stresses are lowest at the surface and
may be insufcient to overcome even a very low strength regolith. Second, although failure conditions within the soil prole
may already be critical, the material may be being held in place
by the more resistant, root-reinforced, surface regolith. The

removal of cohesion from this resistant layer therefore results in

the failure of the entire depth of regolith (McConchie, 2004;
Preston and Crozier, 1999).
This study therefore aims to determine which model is most
appropriate for the conditions found in the Lake Tutira catchment.

Fig. 5. Development of positive pore-water pressures in response to rainfall over 2 years.

270

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

The results also have more general applicability because they are
derived from an extensive dataset of eld conditions that includes
data from an actual slope failure.
3. Instrumentation
Real-time monitoring of hydraulic conditions, and their control
on slope instability, requires continuous measurement of a number
of variables: the inputs and outputs of moisture; volumetric
moisture content; the tension at which this moisture is held; and
the development of any positive pore-water pressures above the
bedrock interface.
The instrumented site is typical of areas affected by shallow
regolith failures in New Zealand. This site has remained stable
despite widespread failures within the catchment over the past
100 years. Stability is indicated by the relatively deep soil (1.2 m)
and a thick A horizon characteristic of previous forest cover.
Despite this, the site was regarded as being susceptible to failure at
some stage in the future because of its morphology and topographic position.
Five locations were instrumented; three down the axis of the
slope (slope length 20 m, relief 7 m) and one to each side (slope
width 16 m) (Fig. 2). At each site, TDR probes, tensiometers, and
thermistors were installed at depths of 250, 500, and 1000 mm.
A piezometer was installed at the bedrock interface, the
potential shear plane, approximately 1200 mm beneath the

ground surface. The sensors were connected, via a multiplexer,

to a Campbell Scientic CR10 datalogger programmed to read
the value of the piezometers every 5 min and the other sensors
every 15 min. A climate station, measuring all the variables
necessary to record the input and output of moisture from the
system, was installed a short distance (50 m) from the instrumented slope (Hawke and McConchie, 2003). Continuous
monitoring of these data occurred over 5 years, including
a period when two new shallow regolith slope failures occurred.
4. Results
4.1. General pattern of hydraulic response
Despite the small spatial extent of the study area, approximately
20 m by 16 m, the physical properties of the regolith show a high
degree of variability. This variability is apparent both within
a single prole, and between adjacent proles (Hawke and
McConchie, 2003). These physical differences inuence both
antecedent moisture conditions and the nature of any hydraulic
response to precipitation at specic sites.
The total rainfall during 2002 (1298 mm) was only 93% of the
long-term average (Fig. 3). However, the rainfall was typically
distributed with rain days relatively evenly spread throughout the
year. There was the usual winter peak (June to August) in rainfall,
but only 3 days had rainfall above 50 mm.

Fig. 6. Development of positive pore-water pressures at three sites in response to two rainfall events with different characteristics.

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

The soil moisture and piezometric response of the regolith to

rainfall at the study site has been discussed in detail (Hawke and
McConchie, 2003). However, it is worth summarizing the key
ndings to place the rainfall-triggering event in context. The soil
moisture content at 250 mm depth responds rapidly to any rainfall
over approximately 10 mm/day. However, the sensitivity of the
response is greater during the summer (DecemberFebruary)
because higher evapotranspiration produces greater available
storage. The soil also wets up far more rapidly than it dries out. The
open structure and physical properties of the regolith allow any
rainfall to inltrate quickly and ll the available pore space.
Although available pore space is considerable, the small pore size
means that the water is held tightly by capillary tension; thus,
drainage is slow. There is an increase in the average level of soil
moisture during the winter, indicated by the doming of the graph
between July and October, as inputs of moisture exceed losses to
evapotranspiration and seepage (Fig. 3).
The soil moisture content response at a depth of 500 mm is
similar to that at 250 mm depth; however, it is much more
subdued. This reects the ltering effect of the upper 500 mm of
the soil prole. For example, at this depth approximately 40 mm/
day of rainfall is required before any response is detected. However,
even this amount of rain causes very little response at 1000 mm
depth. In fact, there is almost no response in the soil moisture
content throughout the year at approximately 1 m depth. The
variation in response through the prole is caused by two
phenomena. First, the higher porosities towards the top of the
prole allow the soil to capture most of the available moisture.
Little moisture remains to percolate and ll empty pore spaces
lower down. Second, much of the reduced pore space in the lower
prole already contains moisture, reducing the potential for additional storage (Fig. 3).
The highly site-specic nature of the soil moisture response to
rainfall is shown in Fig. 4. Data from the left ank (ML) is signicantly more subdued than either that collected in the centre (MC)
or on the right ank (MR). The response of ML is representative of
data usually collected at a greater depth where any changes are
smoothed by the soil prole above. In contrast, MC and MR show
highly responsive proles with signicant changes in moisture
content following each rainfall event. The site located on the right
ank (MR) is the most responsive and variable. While there is
a general pattern of soil moisture response to rainfall events, the
detail of the response is highly site-specic. These three sites are
within approximately 16 m horizontal distance.
The piezometric response of the middle of the slope (MC) shows
a number of distinct characteristics (Fig. 5). First, positive pore
pressures can form at any time of the year. Second, only the large
input of precipitation, from either high-intensity or long-duration
events, results in the formation of a saturated zone above the
bedrock interface, and hence the development of positive porewater pressures. Even though the centre of the slope maintains
positive pore-water pressures for longer than any other location,
a saturated zone was recorded only 8% of the time. Third, any
piezometric response occurs soon after rainfall begins and dissipates quickly once the rain stops. However, the exact nature of the
piezometric response is highly site- and storm-specic (Fig. 6).
Rainfall intensity generally controls the magnitude of any piezometric response but storm duration controls its persistence. Some
long durationlow intensity storms, however, can lead to the
development of pore-water pressures. The interaction of rainfall
intensity and duration is consistent with that observed elsewhere
(Haneburg, 1991a, b; Reid, 1994). Elevated sites on the anks
respond more quickly to rainfall events which lead to positive porewater pressures. However, the persistence of any response is
enhanced by topography with sites having larger catchment areas

271

Fig. 7. During monitoring a large rainstorm event triggered two small shallow soil
slips.

sustaining positive pore-water pressures for longer than sites on

the anks e.g., MC sustains positive pore-water pressures for longer
than either MR or ML (Fig. 6). While it is possible to relate the size of
the piezometric response at a specic site to the amount of rainfall
(Fig. 6; Pierson, 1980), slope morphology creates considerable intersite variability, even at the scale of a few metres.
4.2. Conditions during slope failure
Between eld visits in August and September 2003 a major
storm event occurred within the catchment. This triggered two
small landslides, one directly below the instrumented study site
and only 17.1 m from the nearest sensors (Fig. 7). The characteristics
of this landslip are summarized in Table 1. This event, and the
instability it triggered, provided the opportunity to test the validity
of the two general regolith stability models discussed above.

Table 1
Characteristics of the slope failure immediately below the instrumented site.
Characteristic

Dimension

Length (m)
Width (m)
Depth, z (m) median of 143 points
Volume (m3)
Original slope angle ( )
Shear plane angle, b,( )
Saturated bulk density, g, (kg/m3)
Proportion regolith saturated, m
Cohesion, c0 , (kg/m2)
Friction coefcient, 40

5.7
3.8
0.465
6.65
28
33-37
1738
0.92
713.5
0.415

272

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

Fig. 8. Piezometric response to rainfall throughout the length of record from the site (L) nearest the landslide.

During August 2003 533 mm of rainfall were recorded. This

compares to the average over the previous 5 years of only 173 mm.
Although August is usually one of the wettest months; to receive
38% of the average annual rainfall is extreme. Rainfalls on the 2123
August inclusive were 44, 114, and 193 mm respectively. The wetter
than average August led to a slightly higher than average annual
rainfall of 1780 mm (127% of normal).
Figs. 8 and 9 show the variation in rainfall, piezometric response,
and soil moisture content at three depths from the instrumented site
nearest the new landslip. This rainstorm caused signicant changes
in the soil moisture content, and the development of positive porewater pressures at the bedrock interface. Soil moisture contents at
all three depths reached their maximum level recorded during the
5 years of study. Positive pore-water pressures during the storm,
however, were only 60% of those which had been recorded previously at this site with no slope failures (Table 2). Soil moisture
content therefore is likely to be more important in triggering slope
instability in the study area than positive pore-water pressures.
To test this hypothesis a limit equilibrium analysis, assuming an
innite planar slide-type failure, was undertaken. Such a failure,
i.e., a shallow regolith failure with a failure plane at the bedrock

surface (see Fig. 2 for the failure location) is typical of the shallow
landslips that occur throughout the entire east coast of the North
Island of New Zealand, even though this particular failure is slightly
atypical in shape. Using the data in Table 1, the regolith properties
from Preston (1996), and the hydraulic data in Table 2 the factor of
safety can be calculated for this landslip during the storm event. For
his investigation of regolith resistance in the Tutira catchment
Preston (1996) analysed data from 54 soil proles, undertook 101
measurements of shear strength, and carried out back analyses of
15 soil slips. Given the proximity and spatial coverage of his data, it
is considered appropriate to use his results in the factor of safety
analysis.
While not mechanically correct it is usual to express shear
strength and stress in the factor of safety equation, when using the
innite slope method, as (McConchie, 1992; United States Army
Corps of Engineers, 2003):

FS

c0 g  m,gw ,z,cos2 b,tanf

g,z,sinb,cosb

where

Fig. 9. Soil moisture response to rainfall throughout the length of record from the site (L) nearest the landslide.

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

Table 2
Hydraulic conditions within the slope over the length of record, and at the time of
slope failure.
Middle left site

Full
record

Aug/Sept
2003

Maximum thickness of saturated zone

piezometer @ 1.2 m depth on the bedrock
interface
Maximum soil moisture content:
250 mm
500 mm
1000 mm

0.72 m
(m w 0.6)

0.43 m
(m w 0.36)

0.477
0.538
0.503

0.477
0.538
0.503

Porosity

0.49
0.50
0.58

shear strength
shear stress
effective cohesion
c0
g
saturated bulk density
m
saturated depth ratio
gw
density of water
z
regolith depth
b
shear plane angle
tan4 friction coefcient

From Table 2, the failure occurred in shallow regolith only

0.465 m thick (which is equal to the depth to the bedrock surface)
when the phreatic surface was 0.43 m above the bedrock surface;
hence, m 0.92. Therefore,

FS

819
2:16
380

Given the soil is thin and cohesive, it is not surprising that failure
occurred along the bedrock surface. However, the analysis clearly
indicates that under the recorded conditions this slope should not
have failed, i.e., the factor of safety remained signicantly greater
than 1. Therefore, what could have reduced the effective cohesion
such that the shear stress exceeded the shear strength at the time of
failure?
5. Discussion
Soil moisture content towards the top of the prole is highly
variable in time and space (Famiglietti et al., 1998). Both topography
and soil properties contribute to this variability, but, because the
soil properties are inuenced by topography, separating the specic
effect of each is difcult. Even using the extremely detailed data set
collected for this study, it is impossible to decipher the level of
control related specically to porosity, hydraulic conductivity, or
topography. The heterogeneous soil characteristics in the current

273

study, i.e., soil texture variability is at the centimetre scale, mean

that it is unlikely that either the time distribution, or the rate of
surface-water input, will explain the distribution of water within
the soil prole (Price and Bauer, 1984).
Slope morphology and instrument location with regard to
morphology both clearly inuence the nature of the observed
piezometric response. On the anks of the slope large rainfall
events are required to form a saturated zone. These develop and
dissipate quickly because of the topographic gradient. Sites towards
the centre and base of the slope saturate more often and the
positive pore-water pressures persist longer.
The spatial and temporal distributions of soil moisture content
and pore-water pressures are therefore not a simple function of
precipitation inputs. Changes in both variables are inuenced
strongly by topography (Burt, 1996). Thus, trying to use rainfall to
explain variations in soil moisture content at depth, and the
development of positive pore-water pressures, is complicated by
the ltering effect of the soil within the contributing area. The
attenuation of rainfall-induced pore-pressure uctuations, the
interaction of storm intensity and duration, and the need to include
pre-storm soil moisture content to understand the interaction
between rainfall and slope stability have been discussed previously
(Haneburg, 1991a.). The storage of moisture within the soil prole
has also been used previously to help explain the time lag between
rainfall inputs and landslide motion (Iverson and Major, 1987).
Given the largely static nature of the soil water content below
depths of approximately a metre and the extremely variable nature
of soil water content in the upper soil horizons, piezometric
response cannot be explained as a simple function of antecedent
soil moisture. While this result differs from that of Pierson (1980),
both studies recorded considerable variation between sites as
a result of the heterogeneity in soil physical properties. Thus,
apparently similar slopes can have quite different susceptibilities to
slope failure (Hawke and McConchie, 2003).
The above analysis indicates that the development of positive
pore-water pressures during extreme storm events alone does not
always explain the episodic occurrence of landslips in the region.
Furthermore, the majority of slopes remain stable for considerable
periods of time at angles signicantly above that which can be
explained by the frictional strength of the regolith. Therefore, some
other factor must be imparting additional strength to the regolith. It
is argued that the cohesive strength of the regolith is critical in
maintaining stability of these slopes.
There are two major sources of cohesive strength other than
simple chemical bonding between the mineral grains. These are the
tensile strength of the turf mat, and the matric tension within
unsaturated regolith.

Fig. 10. Variation in matric tension with changes in soil moisture content. (A) The difference in response between the wetting and drying curves of a single sample. (B) The
behaviour of another sample, illustrating the consistent pattern of response for soils in the study area (after Hennrich, 2001).

274

R. Hawke, J. McConchie / Journal of Environmental Management 92 (2011) 266274

Fig. 10 shows the variation in matric tension with changes in

water content over a range typical of that recorded in the eld.
Throughout the year matric tension varies by more than 1 m. The
typical hysterisis between the wetting and drying cycles is also
apparent, with the matric tension changing much more rapidly as
the sample wets compared to when it dries. Signicant changes in
matric tension (which affects cohesion) therefore occur with
relatively small changes in moisture content. These changes are
more signicant, and rapid, on the wetting curve than the drying
curve. This pattern of response in matric tension to variation in
moisture content is consistent across all samples (Fig. 10). It is only
the degree of variation that differs, not the nature of the
relationship.
6. Conclusions
Detailed monitoring has shown that hydraulic conditions within
the regolith are highly variable in both time and space. The
response in moisture status to rainfall events is both storm- and
site- specic. While primarily a function of storm characteristics,
any response can be modied by antecedent moisture conditions,
topographic position, and soil heterogeneity.
The physical properties of the regolith result in highly responsive,
and highly variable, soil moisture changes. The response, although
site-specic, reects both the total rainfall and antecedent conditions. Any piezometric response is very rapid and usually of short
duration. The magnitude of any response is a function of rainfall
intensity, while its persistence is largely controlled by storm duration.
Positive pore-water pressures do not explain the observed slope
failure; at least not alone. During the 5 year record from this site
a saturated zone at least twice as thick has been recorded in the
past which did not cause any slope failures. However, the peak
piezometric response was synchronous with the peak in rainfall
and this had not been recorded previously.
Failure at this site followed a period of extremely high antecedent moisture conditions. Peak conditions recorded during the
storm were near the theoretical maximum based on estimates of
total soil porosity, particularly in the upper half of the soil prole.
This near-saturation of the regolith reduced the matric tension, and
as a result, the effective cohesion of the soil. This cohesion is critical
to the maintenance of stability of the regolith on these slopes. The
reduction in matric tension therefore resulted in a loss of cohesion,
a loss of strength, and consequently slope failure.
The reduction of the matric tension was more important to
triggering slope failures than a rise in positive pore-water pressures.
Pore-water pressures alone were not great enough to trigger failure.
Acknowledgements
RMH acknowledges the guidance and insight provided by Prof.
Rorke Bryan while a PhD student. The authors gratefully acknowledge support from the Hawkes Bay Regional Council, and the
University and Science Faculty Research Funds of Victoria University of Wellington. They are also most grateful to the reviewers of
this paper who made many valuable suggestions.

References
Anderson, M.G., 1984. Prediction of soil suction for slopes in Hong Kong. GCO
publication 1/84, Geotechnical Control Ofce. Engineering Development
Department, Hong Kong.
Beven, K., 1991. Spatially distributed modelling: conceptual approach to runoff
prediction. In: Bowles, D.S., OConnell, P.E. (Eds.), Recent Advances in the
Modelling of Hydrologic Systems. Kluwer Academic Publishers, pp. 373387.
Burt, T., 1996. The hydrology of headwater catchments. In: Petts, G., Calow, P. (Eds.),
River Flows and Channel Forms. Blackwell Scientic, Oxford, pp. 631.
Crozier, M.J., 1986. Landslides: causes, consequences, and environment. Croom
Helm Ltd, London, 252 p.
Crozier, M., Glade, T., 1999. The frequency and magnitude of landsliding: fundamental research issues. Zietschrift fur Geomorphologie Supplementband 115,
141155.
Crozier, M., Preston, N., 1998. Modelling changes in terrain resistance as a component of landform evolution in unstable hillcountry. In: Hergaten, S.,
Neugebaeur, H. (Eds.), Process Modelling and Landform Evolution. Lecture
Notes in Earth Sciences. Springer-Verlag, Berlin, pp. 267284. vol. 78.
Eden, D., Page, M., 1998. Paleoclimatic implications of a storm erosion record from
Late Holocene lake sediments, North Island, New Zealand. Palaeogeography,
Palaeoclimatology, Palaeoecology 139, 3758.
Famiglietti, J., Rudnicki, J., Rodell, M., 1998. Variability in surface soil moisture
content along a hillslope transect: Rattlesnake Hill, Texas. Journal of Hydrology
210, 259281.
Fredlund, D., Rahardjo, H., 1993. Soil Mechanics for Unsaturated Soils. John Wiley &
Sons, New York, 517 p.
Gan, J.K., Fredlund, D.G., 1992. Direct shear testing of a Hong Kong soil under various
applied matric suctions. GEO report 11, Geotechnical Engineering Ofce, Civil
Engineering Department, Hong Kong.
Glade, T., 1998. Establishing the frequency of landslide-triggering rainstorms.
Environmental Geology 35 (2-3), 160174.
Guthrie-Smith, H., 1969. the story of a New Zealand sheep station, fourth ed. A.H. &
A.W. Reed, Wellington, Tutira, 464 p.
Haneburg, W., 1991a. Pore pressure diffusion and the hydrologic response of nearly
saturated, thin landslide deposits to rainfall. Journal of Geology 99, 886892.
Haneburg, W., 1991b. Observation and analysis of pore pressure uctuations in
a thin colluvium landslide complex near Cincinnati, Ohio. Engineering Geology
31, 159184.
Hawke, R.M., McConchie, J.A., 2003. Variability of in situ moisture measurements
and implications for modeling hillslope processes. Environmental and Engineering Geoscience IX 3, 621631.
Hennrich, K.P., 2001. Probability of occurrence and extent of rainfall-induced
landslides, Hawkes Bay, New Zealand. Unpublished PhD thesis, Victoria
University of Wellington, New Zealand.
Iverson, R., 2000. Landslide triggering by rain inltration. Water Resources Research
36, 18971910.
Iverson, R., Major, J., 1987. Rainfall, ground-water ow, and seasonal movement at
Minor Creek landslide, Northwestern California: physical interpretation of
empirical relations. Geological Society of America Bulletin 99, 579594.
McConchie, J., 1992. Water and slope stability. In: Mosley, M. (Ed.), Waters of New
Zealand. New Zealand Hydrological Society, pp. 381408.
McConchie, J., 2004. The inuence of earthow morphology on moisture conditions
and slope instability. Journal of Hydrology (NZ) 43 (1), 317.
Page, M., Trustrum, N., Dymond, J., 1994. Sediment budget to assess the geomorphic
effect of a cyclonic storm, New Zealand. Geomorphology 9, 169188.
Pierson, T., 1980. Piezometric response to rainstorms in forested hillslope drainage
depressions. Journal of Hydrology (NZ) 19, 110.
Pohlen, I., 1971. Soils of Hawkes Bay region. New Zealand Soil Bureau Publication
495, Department of Industrial and Scientic Research, Wellington.
Preston, N.J., 1996. Spatial and temporal changes in terrain resistance to shallow
translational regolith failures. Unpublished MSc thesis, Victoria University of
Wellington, New Zealand., 110p.
Preston, N.J., Crozier, M.J., 1999. Resistance to shallow landslide failure through
root-derived cohesion in East Coast hill country soils, North Island, New Zealand. Earth Surface Processes and Landforms 24, 665675.
Price, A., Bauer, B., 1984. Small scale heterogeneity and soil-moisture variability in
the unsaturated zone. Journal of Hydrology 70, 277293.
Reid, M., 1994. A pore pressure diffusion model for estimating landslide-inducing
rainfall. Journal of Geology 102, 709717.
United States Army Corps of Engineers, 2003. Engineering and Design Slope
Stability, EM 1110-2-1902, 255 p.