You are on page 1of 14

Landslides on Gault: geomorphological identification and

qualitative risk assessment


L.A. Ellis1,*, E. Harrison2 & A.J. Bowden3
1
Coffey Geotechnics, 8/12 Mars Road, Lane Cove West, Sydney, NSW 2066, Australia
2
Atkins Ltd., Woodcote Grove, Ashley Road, Epsom KT18 5BW, UK
3
Mouchel Parkman Services Ltd., Bridge House, Oyster Lane, Byfleet KT14 7JQ, UK
*Corresponding author (e-mail: lucy_ellis@coffey.com)

Abstract: A geomorphological assessment was undertaken to determine the risk of landslide damage
to 38 km of the M25–M26 motorway underlain by the Gault Formation (a stiff clay to weak mudstone).
Conventional methods for geomorphological mapping using stereoscopic aerial photography were
found to be unhelpful, owing to the low topographic expression of landslips in the study area. In
contrast, Light Detection and Ranging (LiDAR)-generated topographic contour maps proved invaluable,
giving the landscape a pseudo-relief and allowing identification of the subtle landslide features. Relict
(inactive) landslides were found to lack a definable backscarp and were mainly discernible by their
indistinct toe lobes. Two types of active landslides were identified within the study area: slow creep
failures barely distinguishable from relict slides, and rapid slides. A specific list of descriptors for
landslides on Gault is presented based on the results of the geomorphological investigation. Four
factors were found to be of particular importance in governing the potential for landsliding: the presence
of relict slips; slopes of 7° or over; an abundant source of water; and close proximity to the top of the
Gault where extremely high-plasticity Upper Gault crops out. The results of this investigation indicated
that the section of motorway most affected by historical landsliding was also most at risk from future
landslides. The similarity in morphology between active creep-type landslides and relict slides has
implications for landslide management on Gault.

This paper presents the relevant findings of a geo- Highways Agency in their Geotechnical Asset Database
morphological study commissioned by the Highways (GAD), i.e. damage to the infrastructure, but not involv-
Agency and undertaken by Mouchel Parkman, assessing ing motorway users. In this area, continuing landslide
the landslide risk for the 38 km stretch of the M25 and movements and the reactivation of historical failures are
M26 motorways mainly located on side-long ground a hazard today. The aim of this study was to determine
(ground that slopes across the line of the motorway) and the factors that govern slope stability using a geomor-
underlain by the Gault Formation in Kent and Surrey, phological approach and to identify the stretches of the
UK (see Fig. 1). The study area included c. 500 m either M25–M26 underlain by the Gault Formation that are
side of the motorway; that is, a 1 km wide corridor most at risk from future landsliding.
was mapped. This was extended to include pertinent The Gault Formation (henceforth referred to as
features, where appropriate. ‘Gault’) is typically of high to extremely high undrained
Since the construction of the M25–M26 in the 1970s shear strength in its fresh state, and can weather to a low
and 1980s, a number of shallow landslides within the to medium shear strength clay. The Gault within the
Gault Formation have resulted in damage to Highways study area forms a sloping vale between more resistant
Agency assets. These include the highway pavement, strata: the Chalk Group of the North Downs to the
services, earthworks and structures, as recorded by the north and the Hythe, Folkestone and Sandgate beds of

N NORTH DOWNS
Junction 3

25 M26
M
NORTH DOWNS Junction 2a
Titsey
Park JJunction 5
M1
Sevenoaks
M1
1

Flint Hall
Farm SOUTH-EAST M40 M25
M23 M2
M25 ENGLAND
Junction 7
Jun
Junc Westerham M4
Junction 6 Limpsf
Limpsfield
Merstham M3
M2 M2
0
M26
Godstone M27
Oxted
Ox 0 5km
Redhill

Fig. 1. Location map of study area.


Quarterly Journal of Engineering Geology and Hydrogeology, 44, 35–48 1470-9236/11 $15.00  2011 Geological Society of London
DOI 10.1144/1470-9236/08-096
36 L. ELLIS ET AL.

N
Cl
Ck
NORTH DOWNS NORTH DOWNS
Cl 25 M26 J3
M
3

Cl UGS
M2

G J2a
J5 LG Cl
Cl Ck Titsey
Park Ck M1
A' Sevenoaks

M1
1
Flint
Cl Hall G
5 SOUTH-EAST M40
Farm M2 Westerham LG
ENGLAND
M25
UGS GREATER
J7 Cl M4
Limpsfield LONDON
Merstham Cl M3 M25
M2 M2
J6

M23
Cl 0
Cl M26
Godstone M27
LG Oxted
Redhill 0 5km
WC
A

A A'
Cl

Ck

WC LG G UGS

Key
Cl Disturbed Blackheath Beds, Clay-with-Flints or Head G Gault Formation
Ck Chalk LG Lower Greensand
UGS Upper Greensand WC Weald Clay

0 Metres 500 1000


Note: Vertical exaggeration of geological cross-section = 3x
Fig. 2. Simplified geological map of the study area, with representative geological cross-section (based on BGS 1:50,000 Sheets 286
and 287, with permission of the British Geological Survey).

the Lower Greensand Formation, which form a low the study area, historical recorded landslides were pre-
ridge to the south (see Fig. 2). A narrow outcrop of dominantly shallow, typically between 3 and 5 m deep
Upper Greensand separates the Gault from the Chalk (Mouchel & Partners Ltd. 1991). Landslide surface
between Reigate and Westerham. However, in eastern extents ranged widely from a few square metres to
Kent, Chalk directly overlies the Gault. Typically a several hundred square metres.
spring-line is present at the interface between the higher The Gault Formation can be subdivided into Upper
permeability Chalk and Upper Greensand and the lower and Lower Gault on the basis of mineralogy, which has
permeability Gault. The Gault outcrop is typically an important bearing on geotechnical behaviour. The
1–2 km wide (Forster et al. 1995). Upper Gault is dominated by smectite-rich clays,
The geotechnical properties of Gault are such that it whereas in the Lower Gault this mineral is present only
is prone to slope instability, even on slopes as shallow in minor amounts. As a result, the Upper Gault tends to
as 7( (Forster et al. 1995). This is, in part, due to its have an extremely high plasticity index, exhibits signifi-
extremely high plasticity and the presence of a spring- cant volume change (swell and shrinkage) with moisture
line above (Garrett & Wale 1985). In addition, the upper content variation, and has low angles of internal friction
layers of the Gault are extensively disturbed by solifluc- and low residual shear strength. The Lower Gault, with
tion and cryoturbation, as a result of periglacial climatic only minor smectite content, has a lower plasticity index
conditions and processes both during and following the and is consequently far less susceptible to changes in
Devensian and Pleistocene glaciations. These processes, moisture content, exhibiting relatively higher angles of
especially solifluction, created shear surfaces throughout internal friction and residual shear strength (Forster
the upper 6 m or so of the Gault (Garret & Barnes 1984; et al. 1995; Davies et al. 2003). Figure 3 indicates the
Garret & Wale 1985; Newman 1985; Forster et al. 1995; relationship between the mineralogy and plasticity index.
Hutchinson 2001). Deep-seated shear surfaces have also Typically, peak shear strength values for Gault range
been found in the Gault, where erosion of the overlying from an effective cohesion (c#) of up to about 10 kPa in
Chalk has caused denudational unloading and pressure- weathered Gault to over 10 kPa in unweathered Gault.
release shearing of the clay (Griffiths et al. 1995). Within The effective angle of friction (φ#) ranges between about
IDENTIFICATION OF LANDSLIDES ON GAULT 37

Harwell BH 3+4
Mineral Composition

-80

Upper Greensand

PL LL Very High to
Extremely High
Plasticity
-100
Sm > I minor Ka Opal-CT

Zeolite
Depth (m.b.g.l.)

Sm > I minor Ka Pyrite

-120 Feldspar

Upper Gault

I + Ka minor Sm
-140
Low to High
Plasticity

Key
Lower Gault
Sm Smectite
I Illite
Ka Kaolinite
Lower Greensand PL Plastic Limit
-160 LL Liquid Limit

0 50 100 150 200


Moisture Content (%)
Fig. 3. Variability in plasticity index within the Gault Formation (after Forster et al. 1995).

20–24( for the higher plasticity Upper Gault and 24–29( Geomorphological analysis
for the lower plasticity Lower Gault. Residual strengths techniques
reduce to a c#r value of close to zero and φ#r of about 7(
for the Upper Gault and 12( for the Lower Gault Mapping of surface expressions of past and present
(Forster et al. 1995; Davies et al. 2003). slope instability is a fundamental element of any land-
Within the study area, previous investigations have slide study (Hutchinson 2001; Phipps 2003; Lee & Jones
indicated that the highest values of the plasticity index 2004). Geomorphological mapping of the site was,
occur close to the boundary with the Upper Greensand therefore, a key component of this assessment. Standard
(DoE 1976). Peak moisture contents were also observed geomorphological mapping techniques (as discussed by,
in the same region (DoE 1976; Mouchel & Partners Ltd. for example, Cooke & Doornkamp (1990), Griffiths
1991). This is to be expected owing to the proximity of (2001), Hutchinson (2001) and Brunsden (2002)) were
the spring-line. The motorway is particularly close to the used to record the local geology, morphology and
escarpment and the Upper Greensand between Junction geomorphological processes, as well as to locate histori-
7 and Titsey Park (as Fig. 2 indicates). This section is cal (relict) and contemporary (active) landslides. When
underlain by the higher plasticity Upper Gault and is combined with the other findings, the maps were then
located close to the spring-line. Within this section the used to examine the preparatory and trigger factors
majority of historical landslides have been recorded. involved in these past landslide events. Using the
Further to the east, the motorway diverges away from premise that understanding past events and processes is
the escarpment onto the lower plasticity Lower Gault. the key to prediction of future geomorphological change
38 L. ELLIS ET AL.

(Cooke & Doornkamp 1990), these findings were then identifying subtle ground features (Rowlands et al. 2003;
used to predict future slope behaviour. McKean & Roering 2004).
It was initially intended that the identification and Maps were generated with contours at 5, 1, 0.50 and
mapping of existing active and relict landslides would be 0.25 m intervals. A viewing program, UKP GeoViewer
carried out primarily by a conventional examination and 1.0, was used to examine these topographic maps,
analysis of aerial photographs. Given the large study overlain onto the corresponding orthorectified aerial
area (c. 38 km2), detailed field mapping was not consid- photographs. Other geospatial data processing software
ered cost-effective. The subtle nature and shallow relief is also commercially available. Some of this software can
of the landslide topography on the Gault made it be used to generate digital elevation models from the
virtually impossible to locate known slips, even when LiDAR data, which can then be shaded to enhance the
using good quality stereoscopic pairs, at a scale of topography. This is a very effective way of viewing
between 1:7500 and 1:10 000 (whichever was available). low-relief features. However, this software was not avail-
As Forster et al. (1995, p. 41) observed: ‘Landslides on able to us during this study. The 1 and 0.25 m contours
Gault clay are often difficult to classify in detail since proved to be the most useful: the 1 m contours were used
the characteristic topographic features which enable the to create panels small enough in size for GeoViewer to
type of movement to be identified frequently have manage without significant loss of processing speed (Fig.
degraded to an indistinct hummocky topography of 4b) and the 0.25 m contours were then brought into each
coalesced slumps and flows with varying degrees of panel in turn to observe the detailed morphology (Fig.
weathering’. 4c). The 0.25 m contours were sufficiently close that they
This difficulty in landslide identification was com- produced a shading effect, giving the landscape of the
pounded by the fact that crop growing or gathering and study area a pseudo-relief. This simple method permit-
ploughing had, in places, caused the partial destruction ted recognition of subtle landslide features, in particular
of the surface expression of slope instability (Forster the presence of hummocky terrain and distinct toe lobes
et al. 1995). These problems were clearly evident at sites (illustrated in Fig. 4). Areas of probable and possible
of known relict landslides (shown in Fig. 4a). However, slope instability and landsliding were marked on the
LiDAR-generated topographic contours, when used in geomorphological maps (shown in Fig. 4d). Figure 5
conjunction with orthorectified aerial photographs, shows an area of hummocky morphology with a prelimi-
proved to be invaluable for this study. The LiDAR data nary interpretation of the toe lobes of small relict slides
revealed subtle landforms most clearly, in particular the clustering around the foot of the steep Chalk escarp-
presence of hummocky terrain (illustrated in Fig. 4b and ment. Figure 5 demonstrates the shading effect of 0.25 m
c), which is the main surface expression of landslides on contours.
Gault (Forster et al. 1995), and the orthorectified aerial The LiDAR data had been processed to remove
photographs formed a base plan showing the location, vegetation and other artefacts. However, this was only
surface features and colour. Many of the identified partially successful. In some cases local distortion
features were very difficult to identify during the ground- occurred in areas where fields had been ploughed or
truthing phase of the project, even with the LiDAR- there was impenetrable vegetation, such as dense bushes
based mapping, emphasizing the subtlety of landslide or trees. These circumstances created an unrepresenta-
features on Gault. tive image of the landscape surface (shown in Fig. 6).
LiDAR uses pulsed light emitted in a swathe, gener- However, these areas were easy to distinguish and
ally from an aerial laser system mounted beneath an measures could thus be taken to interpret them accord-
aircraft (Duffell et al. 2003), or mounted on a tripod ingly. It was also noted that waterlogging of land
(Rowlands et al. 2003; Jones 2006). The laser pulses are occasionally gave the appearance of lobate features on
reflected by the ground surface back to a receiver, the aerial photographs. The LiDAR-generated contours
allowing topographic data to be calculated by measuring could easily be used to determine if there was any
the time taken for the pulses to return. For the purposes corresponding topographic expression in these areas.
of this study, helicopter LiDAR was used to map the Processing algorithms have now improved and this
carriageway between fence lines, using a density of 10 factor does not tend to pose such a problem when using
points per m2. Over the remainder of the study area, less more recently developed software.
detailed LiDAR from a fixed-wing aircraft was used (1
point per m2). The density of data points is dependent
on the number of pulses emitted per second by the laser, Landslide identification on Gault
the flying height and speed, the swath width, and the
scanning pattern. The accuracy of the LiDAR data used Relevant recent publications regarding geomorphologi-
for this study has been assessed to be 150 mm r.m.s.e. cal mapping and identification of landslides were
(root mean square error), considerably less than the reviewed for this study, most notably those by Cooke &
contour spacing of 0.25 m. The high resolution of Doornkamp (1990), Jones & Lee (1994), Forster et al.
LiDAR data makes LiDAR invaluable as a tool for (1995), Cruden & Varnes (1996), Dikau et al. (1996),
IDENTIFICATION OF LANDSLIDES ON GAULT 39

N N

(a) (b)

M25

Key

N Earthworks Geology N
Buildings Gault Formation
Minor Roads Upper Greensand

(c) (d) 0 Metres 100 200 300

Fig. 4. Example of LiDAR-generated imagery, relict landslide SE of Flower Lane. (a) Orthorectified aerial photograph, overlain by:
(b) contour spacing of 1 m; (c) contour spacing of 0.25 m; (d) geomorphological map of relict landslide, S04-R, also showing
hummocky ground, toe lobes and position as recorded by earlier studies (H4, H9 and H14).
40 L. ELLIS ET AL.

.H\ 1
3UHOLPLQDU\LQWHUSUHWDWLRQRI
LQGLVWLQFWUHOLFWVOLSWRHOREHV
,QGLFDWLYHVORSHGLUHFWLRQ
 0HWUHV  

)OLQW+DOO
)DUP

0RWRUZD\

0 &XWWLQJ

Fig. 5. LiDAR imagery showing hummocky terrain of relict slip, west of Flint Hall Farm.

Birch (2001), Lee & Jones (2004) and Griffiths (2005). landslide features summarized in Table 1 (left-hand
These papers provided the basis for a generic list of column) were not entirely consistent with the landslide
landslide descriptors, as summarized in Table 1. The features seen within the study area during this investi-
importance of distinguishing active from inactive land- gation. Thus, to provide a more representative sum-
slides is emphasized, especially for future management mary, the generic landslide descriptors were amended to
purposes (Cooke & Doornkamp 1990; Cruden & Varnes provide a specific description of characteristic landslide
1996; Griffiths 2005). Active landslides are those that morphology on Gault (given in Table 1, right-hand
have shown recent movement and that are also expected column).
to show further movement in the near future. In general, Within the study area, inactive or relict landslides
the sharp, distinct morphological features typical of typically lacked a definable backscarp and could be
active landslides, such as a well-defined backscarp and identified only by their indistinct toe lobes: both major,
toe lobe, are modified over time by weathering and at the base of the slide; and minor, hummocky topogra-
erosion. As a result, inactive (relict) landslides have phy within the body of the slide. All relict landslides
more subdued, rounded morphological features. observed in the study area were assumed to be relict
A total of 13 landslides were identified by the desk- extremely slow creep failures, as rapid slides had been
based geomorphological mapping, including one active remediated previously.
and 12 relict. All features were confirmed during walk- Only one active landslide was observed during the
over surveys. The vast majority of landslides identified study, the features of which were barely distinguishable
(both active and relict) were clustered along a 6 km from the observed relict slides. This active landslide is of
section of the motorway between Junction 6 of the M25 a creep-movement type, moving extremely slowly (using
and Titsey Park (see Fig. 1 for location). The generic definitions from Cooke & Doornkamp (1990), Cruden &
IDENTIFICATION OF LANDSLIDES ON GAULT 41

(a)

N
0 Metres 50 100

(b)
Fig. 6. Examples of problems associated with LiDAR-generated imagery: (a) contours distorted by agricultural activity, west of
Wrotham; (b) contours distorted by dense vegetation on earthworks, Clacket Lane Services.

Varnes (1996), Dikau et al. (1996) and Griffiths (2005)), with rotational elements at the head (Davies et al. 2003;
with movements of the order of 10 mm over 6 months. Phipps 2003). It is possible that this type of failure
Creep landslides are defined as flows, as the slide moves represents an acceleration of the naturally occurring
as a plastic deformable mass. This mechanism results in creep-type landslides, with periods of little or no activity
a very indistinct backscarp and minor scarps (frequently interrupted by rapid movement. Artificial destabiliza-
only shallow surface depressions) and indistinct toe tion of slopes, such as through the construction of
lobes (forming hummocky ground). At the time of cuttings at the slope toe, has historically contributed to
inspection, the only visible morphological expression of this type of failure.
the active landslide was the indistinct toe lobes. There is An example of an active rapid slide within the study
a possibility that other features classified as relict in this area is shown in Figure 7 (exact location not recorded,
study are moving imperceptibly. This landslide could be but probably the December 2000 landslide, defined as
classified as active with any degree of confidence only H1 during this study). This example displays the typical
because inclinometer monitoring is indicating continu- features of an active landslide, including a distinct
ing subsurface movement. This highlights the difficulty backscarp and well-defined toe lobe, with a hummocky
of classifying landslide activity in the Gault. main body. However, this description is based only on
However, rapid, slide-type failures can also occur the historical photograph shown and this type of land-
within the Gault of the study area (again using defini- slide was not directly observed during this investigation.
tions from Cooke & Doornkamp (1990), Cruden & The morphological expression of rapid slides within the
Varnes (1996), Dikau et al. (1996) and Griffiths (2005)). study area is not generally long-lasting, as rapid re-
A failure of this type, close to Flint Hall Farm, was mediation is necessary to avoid further damage to assets.
triggered by prolonged heavy rain in December 2000 Hence, there were no active or relict landslides of this
(Davies et al. 2003; Phipps 2003). This landslide caused type recorded during this study.
heave of the hard shoulder of the anti-clockwise car- The difficulty in distinguishing active creep-type land-
riageway of the M25. The slide was largely translational, slides from relict landslides on Gault has significant
42 L. ELLIS ET AL.

Table 1. Comparison between generic landslide descriptors and specific descriptors for landslides on Gault

Generic landslide descriptors Descriptors for landslides on Gault


(after Cooke & Doornkamp 1990; Jones & Lee 1994; Forster (from results of this investigation)
et al. 1995; Cruden & Varnes 1996; Birch 2001; Griffiths 2005)

Active landslides
Arcuate main backscarp, with fresh shear surface, sharp edges Rapid slide landslides (generally artificially destabilized) have:
and secondary movement
Tension cracks with little or no secondary infilling Distinct major backscarp, with exposed bare rock or soil, often
with multiple minor scarplets within the body of the slip
Water ponded in depressions on slip surface owing to Tension cracks associated with the backscarps
disrupted drainage
Hummocky ground surface, with ridges, benches, minor Distinct steeper toe lobe, often with multiple minor lobes
scarplets and hollows arranged roughly as a staircase within the body of the slip
Tilted trees with no new vertical growth Extremely slow creep landslides (generally naturally occurring)
have:
Disturbed field boundaries Less obvious, generally non-existent, major and minor
backscarps, without tension cracks
Vegetation differences on and off slip surface Hummocky slip topography, often owing to minor slumps
within the body of the slip
Distinct toe lobe of accumulated failure material Soil creep causing small-scale hummocking and minor
terracing
All features are vegetated, with no bare rock or soil
Features are virtually indistinguishable from relict landslide
features; monitoring is required to assess if movement is
continuing
Both types result in:
Ground surface disturbances, such as leaning trees and
disturbed fencelines, the amount depending on the rate of
movement
Relict landslides
Rounding of backscarp edge and tension crack edges owing to Indistinct toe lobes, both major, for the entire slip, and minor,
weathering within the body of the slip. An arcuate toe lobe was often
the main distinguishing feature of the identified relict slips
Secondary infilling of tension cracks Less obvious, generally non-existent, major and minor
backscarps
Tilted trees with vertical growth above inclined trunk Hummocky slip topography, owing to minor toe lobes within
the slip body
Indistinct hummocky topography of coalesced slumps and All features are vegetated, with no bare rock or soil
flows
Homogeneous and slow-growing vegetation

Arcuate
Backscarp

Hummocky
Ground Surface

Distinct
M25 Fenceline Distorted by
Toe Lobe

Ground Movement
at Landslide Toe
Note: Date of photograph unknown but scanned from 1991 report. 0 Metres 25 50
Although poor quality, this is the only known rapid slide-type
landslide example within the study area.

Fig. 7. Historical aerial photograph of active landslide near Flint Hall Farm (after Mouchel & Partners Ltd. 1991).
IDENTIFICATION OF LANDSLIDES ON GAULT 43

implications for determining appropriate management Qualitative risk assessment


strategies. Movement within active slides can be inter-
mittent. Destabilizing factors, such as heavy rainfall or A qualitative risk assessment was undertaken to identify
toe excavation (e.g. cutting construction) may trigger sections of the motorways most likely to be damaged by
movement of previously imperceptibly creeping land- landsliding. After reviewing standard approaches to
slides, and possibly even relict landslides. This move- landslide risk assessment, as discussed in detail by Lee
ment can range from extremely slow creep to rapid, & Jones (2004) and Australian Geomechanics Society
visible failure, depending on the degree of destabiliza- (2000, 2007), it was decided that a quantitative risk
tion. However, to conclusively determine whether a assessment would prove too time consuming and costly
creep-type landslide in Gault is active, monitoring may, within the scope of the project. The large study area
in many cases, be necessary. Where landslides, whether (c. 38 km2) required a simplified approach and therefore
active or relict, are located in close proximity to physical the project adopted a qualitative risk assessment
assets, continuous inclinometer monitoring is recom- approach. The method used is not described in detail
mended as part of any risk management strategy. here, although the general approach is outlined below
and summarized in Figure 9 and Tables 3 and 4.
The hazard factors were used to determine the worst
possible scenario, or potential maximum consequence,
Factors governing slope stability on expressed as a maximum consequence index, ranging
Gault from one (negligible damage to Highways Agency infra-
structure) to four (severe damage to Highways Agency
Information on a range of factors either thought to infrastructure). The potential maximum consequence
influence landsliding (such as slope gradients and drain- was assessed to be proportional to the likely severity of
age conditions) or that provide an indication of slope future landslides: the greater the potential magnitude of
stability (such as slope monitoring data and motorway the slip, the greater the potential for damage to the
earthworks inspection records) was gathered for the motorway. This approach assumes that the location of
38 km section of motorway within the study area. These the landslide along the length of motorway does not
landslide hazard factors (F1–F15) and the rationale influence the potential consequences, a reasonable
behind their choice are summarized in Table 2. The most assumption in this case, given that the construction and
suitable method of assessing the relative contribution of conditions adjacent to the motorways were fairly homo-
the various hazard factors was considered to be a bar geneous within the study area. Only hazard factors that
chart summary. Highways are linear structures and are likely to affect the magnitude of a landslide and their
therefore lend themselves to this form of 1D simplifica- proximity to the motorway were used to determine the
tion. This style of presentation is also familiar to high- maximum consequence index value.
ways engineers. The presence of each hazard factor at The probability of occurrence of new landslides was
any given location on the geomorphological map was determined by assigning a probability index, ranging
shown within the corresponding bar beneath the site from A (unlikely) to D (almost certain), based on the
map, with a descriptor if applicable (shown in sample frequency of factors that indicated areas that were
map extract, Fig. 8). potentially more unstable (see Fig. 9). This assessment
Analysis of the bar chart in conjunction with the could only be subjective, as a statistical analysis of the
results of the geomorphological mapping indicated that precursors to active landsliding was not feasible owing
four hazard factors had the most significant influence on to the lack of appropriate historical data and active
the occurrence and magnitude of landslides in Gault. landslides in the area.
Landslides were found to occur predominantly where The maximum consequence and probability indices
there was a combination of some, or all, of the follow- were then combined to give a risk level (see Tables 3 and
ing: (1) where landslides had previously been recorded 4). These values were plotted on the bar chart beneath
(Factors F2–F4); (2) where natural slopes in Gault are the site map, to show the risk of landsliding at any given
7( or more (Factor F5); (3) close to spring-lines or location along the length of the motorway (see Fig. 8).
streams (Factor F6); (4) close to the boundary between This simple approach allowed subsequent, more detailed
the top of the Gault and the Upper Greensand or Chalk, investigations (possibly including quantitative risk
where Upper Gault with an extremely high plasticity assessments) to focus on the high-risk areas identified in
index is present (Factor F8). this investigation.
The most significant landsliding was recorded where The findings of the qualitative risk assessment indi-
all four factors were in operation. These key factors were cated that the sections of the motorway that were found
prevalent along the 6 km section of the M25 between to have the highest risk of damage owing to future
Junction 6 and Titsey Park (see Fig. 1 for location), instability were predominantly coincident with areas of
where, as mentioned above, the majority of historical relict landsides in close proximity to the motorway. Risk
landslides were recorded. levels were found to be less severe where earthworks’
44 L. ELLIS ET AL.

Table 2. Landslide hazard factors on Gault

Hazard factor Ref. Description Assessment criteria1

Motorway earthworks F1 The construction of the motorway has a significant Presence of cuttings or
influence on the stability of the adjacent natural embankments,
slopes. Earthworks may cut through already unstable particularly if
areas or may cause toe unloading and over- over-steepened
steepening. Failure in earthworks is more likely if they
are over-steepened (Perry 1989)
Natural factors contributing to
slope instability (2005 recorded
condition)
Landslide crossing motorway F2 Where the motorway cuts through existing landslides, Present?
there is a possibility of reactivation, especially on the
upslope side
Landslide upslope (north) F3 Existing landslides close to the upslope (north) side of Present?
motorway have the potential to become reactivated,
creating obvious concerns if they are close enough to
the motorway to cause damage or encroach onto the
motorway
Landslide downslope (south) F4 Landslides downslope (south) of the motorway are of Present?
potential concern because, if reactivated, they may
initiate retrogressive instability further upslope, which
may impinge on the motorway
Slope angle F5 A limiting angle of stability of 7( was noted for Gault 7( or over?
(Forster et al. 1995). The identification of steeper
slopes is thus significant in determining likely failure
areas. This factor excludes earthworks
Natural drainage F6 Gault is susceptible to swelling and softening as the Present?
moisture content increases. Saturation levels were
found to be the triggering factor causing failure.
Levels vary significantly as a result of single
precipitation events (in perched water tables),
seasonally (in main water table) and through
long-term aquifer fluctuations (Garret & Wale 1985).
Identification of water sources or other drainage
features that augment groundwater conditions is thus
appropriate
Geology: lithology F7 The M25–M26 study area is predominantly underlain by Motorway located on
Gault. However, localized areas of Head, Alluvium Gault or Head
and River Terrace Gravels are encountered. Gault Deposits?
and Head derived from Gault are considered
susceptible to instability (Newman 1985; Forster et al.
1995). The Alluvium and River Terrace Gravels are
more granular in nature and are therefore less
susceptible to slope instability
Geology: surface distance to F8 Geotechnical investigations of Gault have shown that Within 300 m of top of
upper boundary of Gault plasticity and moisture content are at a maximum in Gault (surface, not
Upper Gault, close to boundary with Upper depth)?
Greensand (DoE 1976; Mouchel & Partners Ltd.
1991; Forster et al. 1995). These decrease with depth,
down the stratigraphic sequence away from the Upper
Greensand. Landsliding and slope instability are thus
more likely to occur close to the Upper Greensand.
Also, a spring-line exists at the interface between the
Gault and Upper Greensand, providing a potential
groundwater source
Other evidence of slope F9 Factors such as soil creep, hummocky ground and Present?
instability leaning trees are all indicators of slope instability
(Birch 2001). Thus, areas where these features are
observed have a higher probability of landsliding
Indirect evidence of slope
instability
GAD2 records of earthworks’ F10 Records of earthworks’ defects that indicate potential 1A–1D present? 1A more
instability slope instability, such as motorway cracking, soil significant
creep and leaning trees
IDENTIFICATION OF LANDSLIDES ON GAULT 45

Hazard factor Ref. Description Assessment criteria1

Pavement subsidence repairs F11 These records indicate sections of the motorway that Present?
have required patching owing to subsidence, indicat-
ing sites of potential instability
Culvert cracking F12 Cracked culverts may indicate sites of potential instabil- Significant cracking
ity recorded?
Records of previous landslides
Landslides recorded in previous F13 The position and character of landslides that have Present?
studies occurred in the past are important indicators of the
possibility of future slope instability. These historical
landslides records should be treated with some cau-
tion, as inconsistencies occur in the records
Large-scale remedial works F14 Past landslides have necessitated large-scale remedial Present, no evidence of
works in some areas to maintain a satisfactory degree further movement indi-
of future stability, reducing future landslide hazards. cating hazard mitigation
Without remediation, these sites would have a high and risk reduction
potential for landsliding. Smaller remedial works have
not been recorded with any degree of consistency and
could not be summarized here
Continuing landslide movement F15 Areas that are currently being monitored for potential See F14
monitoring instability are shown in order to highlight known
locations of continuing landslide movement or where
stabilization has been achieved
1
The assessment criteria column indicates the conditions that, if present, are a contributing factor to landslide hazard. 2GAD Records are Highways
Agency Geotechnical Asset Database records of earthworks’ instability and distress, in accordance with HD41/03 (Highways Agency 2003).

failure alone was likely. However, the presence of earth- generated topographic maps, with 0.25 m contours
works cutting into slopes in areas with relict slips superimposed onto orthorectified aerial photo-
increased the level of risk. graphs. The closely-spaced contours produced a
Historical landslides in Gault that have previously pseudo-relief effect, with distinct shading, which
affected the M25–M26 motorways have occurred only allowed identification of the subtle features charac-
between Junction 6 and Titsey Park, where Upper Gault teristic of landslides on Gault. More recent devel-
with a extremely high plasticity index crops out on opments in spatial data processing software have
relatively steep-angled slopes close to the motorway. The produced even more effective visualization tools
sections of motorway identified by this investigation as for low-relief landforms.
being at the highest risk from future landsliding are also (3) Two types of active landslide were identified on
within this problematic area. The only exception to this Gault: rapid and creep-type. Rapidly moving
is at the location of the 2000 Flint Hall Farm landslide, slides were generally triggered by artificial destabi-
where remediation has lowered the likelihood of future lization, such as through removal of material
failure (Davies et al. 2003). Although relict landslides during cutting construction. Very slow creep-type
were identified along other sections of the study area, the and relict failures were found to exhibit a specific
risk of reactivation was significantly lower, owing to the morphology, requiring modification of more
absence of other hazard factors, with minimal damage if generic landslide descriptors. Both relict and active
reactivation were to occur. creep-type landslides can lack a definable back-
scarp and were mainly discernible by their indis-
tinct toe lobes. The lack of distinction between
Conclusions active and relict landslide morphology has signifi-
The geomorphological assessment proved a valuable cant implications for those managing landslide risk
method of determining the risk of landsliding along the in Gault, possibly requiring monitoring to deter-
stretch of M25–M26 underlain by Gault. The key find- mine the status of the slide.
ings of this study are summarized as follows. (4) Four key hazard factors were found to be most
(1) Conventional analysis of stereoscopic aerial pho- important in governing the potential for landslid-
tography, at a scale of between 1:7500 and ing in Gault, in particular: (a) the presence of relict
1:10 000, proved limited for the purposes of geo- landslides; (b) natural slopes of 7( or greater; (c)
morphological mapping and identification of a source of water: either at a spring-line or streams
active and relict landslides on Gault. or rivers; (d) proximity to the boundary of the top
(2) A far more effective method of landslide identifi- of the Gault, where Upper Gault with an
cation proved to be the analysis of LiDAR- extremely high plasticity index crops out.
46 L. ELLIS ET AL.

Ck
Ck

B
UGS

Ck

UGS

UGS

F F

Ck B

UGS F

Base mapping: Crown copyright Ordnance Survey. All rights reserved. Geological mapping: Interpreted from BGS 1:50,000 Sheets 286 and 287, with permission of the British Geological Survey

Fig. 8. Geomorphological mapping extract showing bar chart hazard and risk assessment (reproduced here in black and white from
an original colour drawing).
IDENTIFICATION OF LANDSLIDES ON GAULT 47

Maximum Consequence Index Probability Index

NEGLIGIBLE

UNLIKELY
Natural factors Evidence of Landslide Ground

DAMAGE*
Earthworks + contributing to + previous Trigger + Movement
1= slope instability instability A=
F1 OR F5 OR F6 OR F8 + F9-F12
F1 + F2-F9 + F13
Landslide Ground
Natural factors Evidence of Trigger + Movement
+

DAMAGE*
( ) ( )
+ i.

LIMITED
Earthworks contributing to previous

CONCIEVABLE
OR slope instability instability
2= F1 + F5 OR F6 + F8 + F9-F12*
F1 + F2-F9 + F13 B
OR OR
*F10 NOT 1A or 1A(F)
OR
Remedial Landslide
Measures + Monitoring

MODERATE
Large, over- Natural factors

DAMAGE*
steepened OR contributing to (+) Landsliding ii.
3= earthworks slope instability F14 + F15* *F15=No movement recorded

F1 OR F2-F9 + F2-F4
(+) F13 OR Landslide Ground
Trigger + Movement

( ) ( )
i.
F1 + F5 OR F6 OR F8 + F9-F12*
4 = Deterioration of SEVERE DAMAGE*
*F10 NOT 1A or 1A(F)
3 Landslide Ground

LIKELY
Trigger Movement

( ) ( )
* Damage to Highways Agency infrastructure
C ii.
F1 + F5 OR F6 + F8 + F9-F12*
OR OR
*F10 NOT 1A or 1A(F)
Remedial Landslide
Risk Level Measures + Monitoring
Matrix iii.
KEY
(see Table 3) F14 + F15
F1 Factor Present
Landslide
+
Ground + Landslide
F2-F9 Only a Few Factors Recorded Trigger Movement OR Monitoring

( ) ( )
i.
RISK LEVEL + + +
F1 Factor Absent (see Table 4) F1 OR
F5 OR F6 OR
F8 + F9-F12 OR F15
D
(+) Generally (but not Necessarily)
Ground
Movement
ii. ALMOST
F10* *F10=1A or 1A(F) CERTAIN

Fig. 9. Flow chart summarizing qualitative risk assessment method (after Lee & Jones 2004; Australian Geomechanics Society
2007).

Table 3. Risk level matrix (after Lee & Jones 2004; Australian Table 4. Risk level descriptions and recommended actions (after
Geomechanics Society 2000, 2007) Lee & Jones 2004; Australian Geomechanics Society 2000,
2007)
Consequence index Probability index
Risk level Description Recommended action
A B C D
1 N N A H N Negligible No action required.
2 N A H H Re-inspections on a
3 N A H S medium- or long-term
4 A H S S (bi-decadal) basis
A Acceptable Remedial action not
required. Re-inspections
(5) The qualitative risk assessment indicated that the on a short-term (1–2 year)
basis
section of motorway most affected by historical H High Further investigation
landsliding was also most at risk from future required. Depending on
landslides. The section between Junction 6 and the findings of these
Titsey Park (especially close to Flint Hall Farm) studies, low-intensity
was at severe risk owing to the presence of key monitoring, or remedial or
preventative action may be
hazards close to the M25, except where stabiliza- required
tion measures had already been implemented. S Severe Detailed investigation and
This study was specifically concerned with landsliding research required.
in Gault. However, the relatively simple assessment Monitoring must be
techniques, especially the use of LiDAR-generated topo- undertaken. Preventative
or remedial action
graphic contour maps, are readily applicable to other required
environments, particularly in areas of low relief.
48 L. ELLIS ET AL.

Acknowledgements. The Highways Agency and Mouchel Park- G, C. & B, S.J. 1984. The design and performance
man Services Ltd are thanked for giving permission to publish of the Dunton Green retaining wall. Géotechnique, 34,
this paper. Ove Arup, which is researching the use of LiDAR 533–548.
in highway maintenance, is thanked for assistance with the G, C. & W, J.H. 1985. Performance of embankments
supply and manipulation of the LiDAR images. and cuttings in Gault Clay in Kent. In: Failures in
Earthworks, Proceedings of the Symposium of Failures
in Earthworks, Institution of Civil Engineers. Thomas
Telford, London.
References G, J.S. (ed.) 2001. Land Surface Evaluation for Engi-
neering Practice. Geological Society, London, Engineer-
A G S 2000. Landslide risk ing Geology Special Publications, 18.
management concepts and guidelines. Australian Geo- G, J.S. 2005. Landslides. In: F, P.G, L, E.M.
mechanics, 35, 49–92. & M, G. (eds) Geomorphology for Engineers.
A G S 2007. Practice note Whittles, Caithness, 173–217.
guidelines for landslide risk management 2007. Australian G, J.S, B, D., L, E.M. & J, D.K.C.
Geomechanics, 42, 63–114. 1995. Geomorphological investigations for the Channel
B, G.P. 2001. Mapping high pressure gas pipelines in Tunnel terminal and portal. Geographical Journal, 161,
South Wales. In: G, J.S. (ed.) Land Surface 275–284.
Evaluation for Engineering Practice. Geological Society, H A 2003. Maintenance of highway geotechni-
London, Engineering Geology Special Publications, 18, cal assets. In: Design Manual for Roads and Bridges, Vol.
73–82. 4, Section 1, Part 3, HD41/03. Highways Agency,
B, D. 2002. Geomorphological roulette for engineers London.
and planners: some insights into an old game. Quarterly H, J.N. 2001. Reading the ground: morphology and
Journal of Engineering Geology and Hydrogeology, 35, geology in site appraisal. Quarterly Journal of Engineering
101–142. Geology and Hydrogeology, 34, 7–50.
C, R.U. & D, J.C. 1990. Geomorphology in J, D.K.C. & L, E.M. 1994. Landsliding in Great Britain.
Environmental Management, 2nd edn. Oxford University HMSO, London.
Press, Oxford. J, L.D. 2006. Monitoring landslides in hazardous terrain
C, D.M. & V, D.J. 1996. Landslide types and using terrestrial LiDAR: an example from Montserrat.
processes. In: T, A.K. & S, R.L. (eds) Quarterly Journal of Engineering Geology and Hydrogeo-
Landslides Investigation and Mitigation. Transportation logy, 39, 371–373.
Research Board Special Report, 247, 36–75. L, E.M. & J, D.K.C. 2004. Landslide Risk Assessment.
D, J.P., L, F.A., P, J., P, D. & Thomas Telford, London.
C, D. 2003. Stabilization of a landslide on the M25 MK, J. & R, J. 2004. Objective landslide detection
highway, London’s main artery. World Wide Web and surface morphology mapping using high-resolution
Address: www.hatch.ca/Infrastructure/Stabilization_of_ airborne laser altimetry. Geomorphology, 57, 331–351.
a_Landslide.pdf. M & P L. 1991. M25 Improvements between
D, R., B, D., S, L. & I, M.L. (eds) Junction 2 and Junction 8, Geotechnical desk study. High-
1996. Landslide Recognition. Report No. 1 of the Euro- ways Agency Report, 27153G/001/C.
pean Commission Environment Programme Contract No. N, R.L. 1985. Geological and geotechnical considera-
EV5V-Ct94-0454. Wiley, Chichester. tions associated with some sections of the M25 construc-
DE 1976. M25 South Orbital Motorway, Godstone to West- tion. Quarterly Journal of Engineering Geology and
erham, Soil and Materials Engineer’s Report. Department Hydrogeology, 18, 149–160.
of the Environment, South East Road Construction Unit, P, J. 1989. A survey of slope construction on motorway
Soils and Materials Group Report, 7/76. earthworks in England and Wales. Transport and Road
D, C., R, D.M. & W, M.R. 2003. Earth- Research Laboratory, Research Report, 199.
works asset assessment using remote sensing. In: Railway P, P.J. 2003. Geomorphological assessments for transport
Engineering, 5th International Conference, London, July infrastructure projects. Proceedings of the Institution of
2003. Civil Engineers, Transport, 156, 131–143.
F, A.H, P.R.N.  . 1995. Engineering geology of R, K.A., J, L.D. & W, M. 2003.
British rocks and soils: Gault Clay. British Geological Landslide laser scanning: a new look at an old problem.
Survey, Engineering Geology and Geophysics Group, Quarterly Journal of Engineering Geology and Hydrogeo-
Technical Report, WN/94/3. logy, 36, 155–157.
Received 7 October 2008; accepted 13 March 2010.

You might also like