4
DEBRIS FLOW TYPES AND MECHANISMS
by I M Nettleton, S Martin, S Hencher and R Moore
4.1
FLOWS
4.1.1 Classification of Flows
The flow type (see Section 2.1) of landslide movement is classified according to whether the
materials involved are rock or engineering soils (Varnes, 1978).
In the context of the recent flow type landslides on the road network in Scotland it is the
engineering soil flows that are pertinent. The descriptions of these materials and the
corresponding classifications are shown in Table 4.1.
Table 4.1 – Engineering soils and associated flow types (after Varnes, 1978; Cruden and
Varnes, 1996; Hutchinson, 1988).
Engineering Material Description
Soil Types
Category
Debris
Debris Flow
Earth
Peat
A mixture of fine materials (clay, silt, sand) and coarse materials
(gravel, cobbles, boulders). Often coarse material predominates.
Material comprises a high proportion of fine materials (clay, silt,
sand)
Peat
Earth Flow
Peat Flow (Bog Burst)
Figure 4.1 shows the predominantly fine materials deposited by an Earth Flow at one of the
events on the A9 Trunk Road north of Dunkeld (August 2004). Coarser material was in
evidence at the other two main locations.
Figure 4.1 – Hillslope flow which has formed its own channel by erosion (A9 north of
Dunkeld, August 2004). (Courtesy of Alan MacKenzie, BEAR.)
45
TYPES AND MECHANISMS
Figure 4.2 shows the predominantly coarse debris deposited by Debris Flows on the A887
Trunk Road at Invermoriston (August 1997). This is by far the most common type of flow
encountered on the road network in Scotland.
Figure 4.2 – Debris flow material on the A887 Trunk Road at Invermoriston, August
1997. The debris consists predominantly of coarse material (gravel, cobbles and
boulders), with some finer material (clay, silt and sand) and tree trunk debris.
(Photograph courtesy of Northpix.)
Pierson and Costa (1987) classified flow type landslides on the basis of the flow velocity and
sediment concentration and their rheological classification of sediment water flows is shown
in Figure 2.2. This rheological approach to classification may be particularly appropriate for
the development of remedial measures such as channels, overshoots, culverts and so forth.
Based on Pierson and Costas’ (1987) classification it is considered that most recent Scottish
flow landslides would fall within the Debris Flow category - which description is adopted as
an all-encompassing term for this work. Peat flows are discussed further in Section 4.3.4.
4.2
DEBRIS FLOWS
4.2.1 Debris Flow Materials
Debris flows usually comprise a mixture of fine (clay, silt and sand) and coarse (gravel,
cobbles and boulders) materials with a variable quantity of water. The resulting mixtures
often behave like viscous “slurries” as they flow down slope. They are often of high density,
60% to 80% by weight solids (Varnes, 1978; Hutchinson, 1988), and may be described as
being analogous to “wet concrete” (Hutchinson, 1988).
Debris flows are potentially very destructive as they cause significant erosion of the
substrates over which they flow, thereby increasing their sediment charge and further
increasing their erosive capabilities. The density and rapid movement of debris flow materials
yield a mass with significant energy (Table 4.2). This has the ability to pick-up and transport
even large and/or well secured objects, thereby giving rise to the potential for significant
46
TYPES AND MECHANISMS
damage. Examples of such “accidental detritus” (Johnson and Rodine, 1984) picked-up by
debris flows include tree trunks, branches, large boulders, parts of structures and vehicles (see
Figures 4.1 and 4.2).
Table 4.2 – Landslide rates of movement (WP/WLI, 1995).
Movement Rate Velocity Class
Extremely rapid
6
Rapid
5
Moderate
4
Very slow
Extremely slow
Rate
(mm/sec)
Debris Flow
Range
7
Very rapid
Slow
Velocity
Limits
5m/sec
5 x 103
3m/min
50
1.8m/hour
0.5
13m/month
5 x 10-3
1.6m/year
50 x 10-6
16mm/year
0.5 x 10-6
3
2
1
The debris flows experienced in Scotland occur on hillsides with relatively thin (typically
<3m), predominantly unconsolidated and relatively coarse-grained superficial deposits
(Ballantyne, 1986). The Scottish superficial deposits that are most prone to debris flows are
shown in Table 4.3.
Table 4.3 – Superficial deposits prone to debris flows. Descriptions based on McMillan
and Powell (1999).
Superficial Deposit
Description
Colluvium/hillwash
Unconsolidated, heterogeneous soil mass deposited by water run-off or slow down
slope creep.
Accumulation of angular rock fragments at the base of a cliff or steep rock slope due
to weathering, spalling/ravelling and rockfalls.
Unconsolidated detrital material laid down by a stream, river or other body of water
Unconsolidated heterogeneous soil mass (clay, silt, sand, gravel, cobbles and
boulders) in Proglacial (e.g. glaciofluvial, glacio lacustrine and glaciomarine) and
Glacial (glaciofluvial, morainic and some till) deposits.
Mantle of unconsolidated rock fragments (gravel, cobble and boulder sized), sand,
silt and clay covering bedrock, and formed by the in situ, or nearly in situ,
weathering of bedrock.
Talus
Fluvial
Glacial
Regolith
Superficial deposits dominated by a fine-grained soil matrix, and exhibiting apparent
cohesion, are much less prone to debris flows. Due to their apparent cohesion and lower
permeability these materials tend to be less prone to erosion than coarser grained frictional
materials. The lower permeability will reduce infiltration into these soils (Ballantyne, 1986).
Examples of these “cohesive”, fine-grained matrix superficial deposits include some of the
glacial tills which are generally overconsolidated (e.g. lodgement till). Figure 4.3 shows a
hillslope debris flow where the failure has not cut down into a glacial till with a fine-grained
“cohesive” matrix.
47
TYPES AND MECHANISMS
Figure 4.3 – Hillslope debris flow on the North side of Maol Chean Dearg, 2004. Note
the failure surface lies on top of a band of superficial deposits which contain a higher
proportion of more silty/clayey material, that possesses an apparent “cohesion”.
4.2.2 Debris Flow Forms
Two forms of Debris Flows are distinguishable, based on the topographic and geological
characteristics of their locations.
Hillslope (Open-Slope) Debris Flows
These form their own path down valley slopes as tracks or sheets (Cruden and Varnes, 1996),
before depositing material on lower areas with lower slope gradients or where flow rates are
reduced: e.g. obstructions, changes in topography (Figures 4.4, see also Section 4.3.2 and
Figure 4.15). The deposition area may contain channels and levees.
Channelised Debris Flows
These follow existing channel type features: e.g. valleys, gullies, depressions, hollows and so
on (Figures 4.4, 4.5 and 4.6). The flows are often of high density, 80% solids by weight
(Cruden and Varnes, 1996), and have a consistency equivalent to that of wet concrete
(Hutchinson, 1988). Hence, they can transport boulders that are some metres in diameter, for
example a 9 tonne boulder was reported at the debris flow on the A85 at Glen Ogle (see
Section 5).
48
TYPES AND MECHANISMS
Figure 4.4 – Hillslope (a) and channelised (b) debris flow.
Figure 4.5 – Hillslope/channelised debris flow on the A890 Stromeferry Bypass, October
2001. The figure in orange is at the base of the source area. Note the drainage pipe, to
the side of the gully, installed to take water from an interceptor trench above the debris
flow scarp and convex slope break at the pine tree.
49
TYPES AND MECHANISMS
Figure 4.6 – Stilling basins filled with coarse debris flow material the base of
Frenchman’s Burn on the A890 Stromeferry Bypass. The basins have a combined
capacity of 100m3. The upper basin dam was formed using “armour” stone blocks from
the stream while the lower basin dam was formed using gabion baskets.
Coarser material may form natural levees or accumulate as debris dams (Figure 4.7) at
obstacles (e.g. trees and large boulders and so on.) or changes in channel gradient, thus
leaving finer material in suspension to continue down the channel. Suspended material in
channel flows will typically be deposited in lower gradient sections of channels, where
channels widen and upon emergence from the channel.
In practice many debris flows may start as the hillslope form, but during the course of flowing
down slope they may enter channel type features, form their own channel flow tracks in
superficial deposits or may cut through superficial deposits and then be channelled down preexisting channel features in rockhead: e.g. an infilled stream or gully (Figure 4.8).
50
TYPES AND MECHANISMS
Figure 4.7 – Boulder and Tree Trunk Debris Dam containing an estimated 50m3 to 75m3
of debris, which was subsequently broken up to prevent catastrophic failure and
resulting erosional effects. Frenchman’s Burn on the A890 Stromeferry Bypass.
Figure 4.8 – Location of debris flow scour where channel cut down through superficial
deposits over a buried cliff. In excess of 100m3 of material was picked up at this one
location. Note the large boulders and tree trunks in the foreground. Debris Flow on the
A887 road at Invermoriston, August 1997.
51
TYPES AND MECHANISMS
4.3
PRINCIPLES OF RAPID LANDSLIDE DEVELOPMENT
In understanding the mechanisms of debris flows it is helpful first to consider the mechanisms
of landslides more generally, not least as these frequently form all or part of the trigger event.
Fundamentally, all landslides are the result of gravitational forces causing the ground to fail.
Once the failure starts, the debris will travel downhill, sometimes in a highly mobile state due
to mixing with water. There is potential for failure in any sloping ground but, all things being
equal, the steeper the ground the more prone it is to land sliding.
The susceptibility of a particular hillside to failure is expressed as a “Factor of Safety” as
illustrated in Figure 4.9. For any potential failure surface, there is a balance between the
weight of the potential landslide (driving force or shear force) and the inherent strength of
the soil or rock within the hillside (shear resistance). Provided the available shear resistance
is greater than the shear force then the Factor of Safety will be greater than 1.0 and the slope
will remain stable. If the Factor of Safety reduces to less than 1.0 through some change in
conditions, the model predicts failure.
SHEAR
FORCE
SHEAR
RESISTANCE
ALONG POTENTIAL
FAILURE SURFACE
SHEAR RESISTANCE
= FACTOR OF SAFETY
SHEAR FORCE
Figure 4.9 – Landslide development.
The shear force is mostly a component of the weight of the rock/soil making up the potential
landslide. If water gets into the slope however this may have several impacts on the shear
force. Water pressure can actively encourage movement of the landslide downhill. Saturation
can increase the weight of the sliding mass. Other destabilising factors can be vibrations from
nearby traffic, blasting and earthquakes. Damaging earthquakes are rare in Scotland.
Clearly the highest shear forces will be in steeper ground but generally that ground will also
be inherently strong (otherwise it would not stand so steeply) and therefore may have a
similar Factor of Safety as shallower ground. That said, if a failure occurs in steep ground,
then the effects may be particularly severe because the debris may gain momentum quickly
and travel a long way.
52
TYPES AND MECHANISMS
The shear resistance is provided by the natural strength of the soil or rock. This can be very
prone to the effect of water. The resistance along the potential sliding plane depends, among
other factors, upon the weight of the potential sliding mass.
4.3.1 Causes of Debris Flow
Hillside debris flows typically start as a sliding detachment of material (upland debris slide,
peat slide, rock slide etc.), usually initiated during heavy rainfall, which subsequently breaks
down into a disaggregated mass in which shear surfaces are short-lived and usually not
preserved. The failure mass usually combines with surface water flow, which typically results
in high mobility and run-out.
Channelised debris flows may develop as a result of the mobilisation and entrainment of
sediments by extreme flows confined within stream valleys, which may include the collapse
of natural landslide dams that may have partly or completely blocked channels and stream
valleys for some period prior to the event. For this reason, it is particularly important to
investigate entire catchments in respect of channelised debris flow hazard and risk assessment.
From the above, it may be concluded there are two principal causes of debris flows:
• The initiation of a source upland landslide that develops into a hillside debris flow.
• The mobilisation and entrainment of sediments by extreme flows within stream valleys.
With regard to the causes of landslides these are well documented by others (e.g. Jones and
Lee, 1994; Moore et al., 1995). Ultimately, landslides occur when the force of gravity
exceeds the strength of soils and rocks forming slopes. In such circumstances, slope failure
occurs to restore the balance between the destabilising forces (stresses) and the resisting
forces (shear strength) along the surface of rupture or shear surface. Therefore, a landslide
may be regarded as a dynamic process that changes a slope from an unstable to a more stable
state.
The causes of landslides are generally separated into two types:
• Preparatory factors which work to make the slope increasingly susceptible to failure
without actually initiating it, and
• Triggering factors which initiate movement.
As for all landslides, debris flows are caused by a combination of preparatory and triggering
factors. The interrelationship of these factors controls the likelihood and timing of events at
different sites (Figure 4.10).
When considering the actual causes of upland landslides this relative simplicity gives way to
complexity, as there is a great diversity of causal factors. In broad terms, however, they may
be divided into internal causes that lead to a reduction in shear strength and external causes
which lead to an increase in shear stress (Table 4.4). In summary, the main causes of upland
landslides in Scotland are likely to include:
• Reduction of soil and rock strength over time due to weathering and slope ripening,
53
TYPES AND MECHANISMS
• Historical land use changes, including deforestation, road construction, disturbance of
natural drainage, etc,
Shear strength, Shear stress
• Increased rainfall and storm intensity due to climate change, and
• High transient pore water pressures arising from intense rainstorms.
Long term pre-conditioning of
upland slopes reducing material
strength
Short term change in slope
conditions
Change in slope conditions
Change stresses acting on
the slope
Point where failure will occur
(i.e. shear strength = shear
stress)
Short term increase in shear
stress (e.g. intense rainfall event)
Long term increase of shear
stress (e.g. prolonged rainfall)
Time
Figure 4.10 – Long term development of upland slopes and their susceptibility to rapid
landslides.
Table 4.4 – Causes of landslides.
Internal Causes
External Causes
Materials:
• Soils subject to strength loss on contact with water
or as a result of stress relief (strain softening).
• Fine-grained soils which are subject to strength
loss or gain due to weathering.
• Soils with discontinuities characterised by low
shear strength such as bedding planes, faults,
joints etc.
Weathering:
• Physical and chemical weathering of soils causing
loss of strength (apparent cohesion and friction).
• Slope ripening and soil development.
Pore water pressure:
Removal of slope support:
• Undercutting by water (waves and stream
incision).
• Washing out of soil (groundwater).
• Man-made cuts and excavations.
•
High pore water pressures causing a reduction in
effective shear strength. Such effects are most
severe during wet periods or intense rainstorms.
Increased loading:
• Natural accumulations of water, snow, talus.
• Man-made pressures (e.g. fill, tips, buildings).
Transient Effects:
• Earthquakes and tremors.
• Shocks and vibrations.
54
TYPES AND MECHANISMS
Preparatory Factors
Certain conditions are needed for the initiation of upland landslides including some or all of
the following:
• Steep hillsides promoting gravity induced slope failure,
• Weak jointed rocks exposed in rock slopes and cliffs,
• Weak soils, colluvium or peat overlying weathered rock,
• Low vegetation exposing soils to weathering processes,
• Poor drainage, surface water flow and soil piping, and
• Extreme climatic conditions.
In upland environments, winter weather conditions involve freezing and thawing processes
which act to weaken the soil and rock structure. Dry summer conditions may cause
desiccation of soils (particularly peat), opening large cracks and providing routes for the
ingress of surface water. These weathering processes result in weakened soil structures and
loss of material strength.
The products of weathering often form a mantle of weak soils overlying harder rocks which
provide an interface or potential shear surface along which slope failure may propagate.
Where rocks are exposed at surface, deep penetration of weathering along rock joints and
discontinuities can significantly weaken the integrity of the rock mass and provide
detachment surfaces for rock falls and slides. Jacking by tree root forces may open existing
fractures allowing deeper penetration of water and frost penetration.
The long term weathering of soils and rocks make upland slopes increasingly susceptible to
failure. Such processes are often described as the ‘preconditioning’ or ‘ripening’ of slopes.
They are often overlooked as a major cause of landslides given the long timescales over
which they operate but they are a fundamental control in the location and timing of upland
landslide events.
Historical land use changes and construction activities are also important factors in the preconditioning of slopes for upland landslides. The effects of deforestation are well documented
Sidle et al. (1985) and construction activities involving cut and fill and drainage works can
lead to slope failures sometime after the works are completed.
Such processes of deterioration are illustrated schematically in Figure 4.11 (upper right hand).
The gradual deterioration is represented by a curve in which the Factor of Safety reduces over
a period of time which may comprise tens or hundreds of years. The vertical lines represent
the temporary reduction in Factor of Safety caused by relatively short-term, transient events.
In the course of time, the slope will deteriorate to the point where it is vulnerable to a
transient event – causing a reduction in the Factor of Safety to a value below 1.0. Whether
that event results in catastrophic failure or relatively minor movement and distress depends on
the slope, circumstances and severity of event (including how long it lasts).
In general it can be assumed that for any given hillside there will be a whole range of locally
susceptible areas with different current Factors of Safety (as per Figure 4.11, lower diagram).
One might consider the hillside to comprise an inventory of different slopes of different
55
TYPES AND MECHANISMS
susceptibilities. The susceptibility at each location will be a function of the strength (or
weakness) of the soil at that location but also many other factors such as local slope angle
(and therefore shear stress), catchment leading to that location (which will influence water
pressures and erosion potential), local topography leading to concentrations of surface flow,
erosion and undermining and vegetation cover (deep rooted trees will help hold the soil
together).
Shape of F of S vs.Time
curve (Ripening)
Old terrain
Natural Slope
external events cause variations in
F of S over short periods
(earthquakes, storms)
A
Nick point B
C
D
Slope
Stress relief
fractures
General Concept
(deterioration with time and
change in landscape)
dry
F of S
New terrain
1.0
Coastal
steepening
Log Time
Location
A original slope
B is in metastable condition. Minor movement
triggered by relatively small events
– vertical joints open up
– deterioration above eventual
detachment surface
– sediment infill
– piping
Deterioration curve may
be of different shapes
less
or accelerating
deterioration
C
C failed new terrain – relatively stable
D unstable due to coastal steepening
more
susceptible
Natural slope
A
D
F of S
B
Thus the original terrain setting may be
important for the consideration of large
cuttings
1.0
B
Log Time
now
C
“ripe”
most stable
If, for example, a slope is cut at B above, the
upper part may already be in a precarious state
Figure 4.11 – Mechanisms of long term hillslope deterioration.
A minor rainstorm (say a one in 1 year storm) will probably not result in any discrete
landslides although it will contribute to the general deterioration of the hillside which might
be measurable given sophisticated instruments.
A more severe event (say a one in 10 year storm) may cause a few failures in sections of
hillside where the ripened factor of safety is approaching 1.0 (say 1.0 to 1.1).
A much more severe event (say a one in 100 year storm) may cause all slopes to fail within a
much wider range (say 1.0 to 1.3). Not only will the intensity of such a storm initiate discrete
failures but the length of time that heavy rain continues during such a storm will make the
debris more mobile so that it can flow a long way and impact on more structures than would
otherwise be the case – the event may be disastrous.
56
TYPES AND MECHANISMS
Road cuts may be particularly susceptible to triggering events as illustrated in Figure 4.12.
Fundamentally, if the cutting had not been made, the natural slope would have gradually
deteriorated in geological time (100s or 1,000s of years probably). However the process of
cutting the slope leads to a rapid reduction in the Factor of Safety because of increased shear
stress (over-steepening) and probable changes in the groundwater conditions. Such
deleterious effects can be mitigated against by the construction of engineering works such as
retaining walls or in some other way strengthening the soil /rock.
Cut Slope
Original Water
Table
1
3 tends
to 0
2
1
stress redistribution (
2
ground water change (inducing higher
seepage pressures and possible piping)
3
tends to zero)
– zone of accelerating deterioration
– movement measurable
original slope
– geological indicators:
• rupture zone
• opening of joints
• deposition of weathering
products
• piping (exploiting deterioration)
• geophysics
No inherent
weakness
F of S
may
fail on
cutting
1.0
Log time
Figure 4.12 – Mechanism model for cutting a new slope.
Examples of preparatory factors observed in recent Scottish debris flows are shown in Table
4.5.
57
TYPES AND MECHANISMS
Table 4.5 – Examples of debris flow preparatory factors observed in Scotland.
Preparatory Explanation
Factor
Catchment
•
•
Steep Slopes
Drainage
Superficial
Deposits
•
•
•
•
•
Rockmass
•
•
•
Topography
•
•
Landslides
•
Agriculture/
Forestry/
Construction
•
•
•
Catchments with sparse superficial / peat deposits and or significant exposed bedrock are
likely to result in large and “peaky” surface run-off flows following high intensity
rainfall, Figure 4.14.
The aspect of the catchment, with respect to tracking of prevailing weather systems, may
tend to trap and hold rain clouds.
More prone to failure and landslides in the superficial deposits.
Increase the flow rate and, hence, erosive power of water flows.
Capture and convergence of surface water flows by purpose built drains (e.g. forestry,
farming, roads etc.) and “accidental” drains (e.g. footpaths, animal tracks, walls, fences
etc.), Figure 4.14. This may lead to concentration of water flows with associated potential
for scour, piping and pore water pressure rises.
Loose unconsolidated deposits containing silt, sand, gravel, cobbles and boulders are
particularly susceptible to debris flows e.g. Morainic deposits, weathered in situ bedrock,
colluvium, fluvial deposits.
Variations in thickness or permeability of superficial deposits may lead to restrictions of
groundwater flow and associated pore water pressure increases.
Rockhead hollows or channels may have been infilled with superficial deposits and
provide a source of debris, Figures 4.5 and 4.14.
Rockhead hollows or channels may funnel and collect ground water flow, Figure 4.14.
Down slope inclined rockhead or discontinuity surfaces act as “permeability barriers” and
tend to shed water down slope through the superficial deposits, Figure 4.5.
Concave slope profiles may lead to groundwater convergence towards the base of the
concave slope (Wieczorek, 1987).
Convex slopes may give rise to zones of tension at the crest of the convex slope. Zones
of tension may lead to increased infiltration of surface water run off with a corresponding
potential for an increase in pore water pressures, Figure 4.14.
Landslides into stream channels may create “debris dams” which provide a susceptible
debris source.
Changes in vegetation: e.g. felling of forests, forest/vegetation fires, down-slope
ploughing etc. may increase surface water run-off flow rates and transfer “peaky” nature
of intense rainfall events to surface water run-off and groundwater flow.
Disturbance/damage of organic soil horizons and vegetation root mat may render
superficial deposits more susceptible to scour and water infiltration.
Excavation of slopes (e.g. access tracks, road / rail construction etc.) may steepen slopes
and lead to the creation of abrupt changes in slope angle. These locations may be prone
to scour erosion and to the topographic effects described above, Figure 4.1.
Triggering Factors
Triggering events result in the initiation and mobilisation of upland landslides (Table 4.4). In
upland environments, the most significant triggering factor is likely to be the development of
transient high pore water pressures along pre-existing or potential rupture surfaces. High pore
water pressures are typically generated as a result of extreme antecedent (long-duration)
rainfall conditions and intense rainstorms, both of which can result in high groundwater levels
and perched groundwater conditions. If the soil becomes fully saturated surface water flow
may occur which can result in erosion and triggering of hillside debris flows. Examples of
58
TYPES AND MECHANISMS
such features are common in upland Scotland. It is noted that extreme rainstorms of different
intensities, frequency and storm-paths can result in a very different pattern of landslide
initiation and debris flow response.
The permeability of soils and the speed by which surface water can be transmitted to potential
rupture surfaces is a key factor in the initiation of upland landslides. The interface between
permeable soils and relatively impermeable substrate can lead to the development of cleft
water pressures along soil and rock discontinuities and artesian pore water pressures along
potential rupture surfaces. Certain geological situations are particularly prone to the effects of
water infiltration, for example where permeable soil overlies less permeable bedrock. In such
circumstances rapid increases in pore water pressures can trigger slope failure and
mobilisation of landslides.
Debris Flow Propagating Factors
Many debris flows are of a size that would not lead to any significant events, and examples of
such can be seen on many hillsides in Scotland. However, whilst flowing down slope or
channel some these debris flows may encounter particular features that can exacerbate them.
This may lead to even quite modest debris flows escalating into large ones with potentially
significant destructive effects, which are out of proportion to the initial event.
Based upon Scottish experience, it is usually combinations of these propagating factors that
lead to the large debris flows that have a significant impact upon the road network. These
propagating factors are not well documented in the literature and are therefore some examples
are presented in Table 4.6 and Figure 4.13.
Figure 4.13 – Relict rockhead cliff surviving from, in this case, glacial times (a) and
convex slope break (b).
59
TYPES AND MECHANISMS
4.3.2 Mechanisms of Debris Flow
Source Area
Slides in Soils and Peat: Steep upland slopes which are mantled by a cover of unconsolidated
soils or peat are particularly susceptible to debris slides and hillside debris flows. Debris
slides and peat slides involve shear failure of the unconsolidated material or peat at the
interface with the underlying weathered rock, which typically varies between 1m and 5m
below ground surface. Rapid increases in pore water pressure along the interface result in
significant reductions in effective shear strength, leading to rupture or shear failure along the
soil-rock interface.
Table 4.6 – Debris flow propagating factors.
Propagating Explanation
Factor
Debris Dams
•
•
Convex
Slopes
•
•
Rockmass
•
•
•
•
Drainage
•
•
•
Formed when vegetation, landslide debris or previous flows create “dams” behind which
further debris can build up, Figure 4.7. Eventually these dams become unstable, due to
their size or the state of the vegetation, and will fail catastrophically during debris flows.
This additional sediment charge increases the debris flow mass, erosive power and may
create flow pulses.
Tree trunks and branches entrained in debris flows form debris “dams” which are likely to
trap large quantities of debris then fail catastrophically releasing highly erosive debris
flow pulse.
May form zone of tension within superficial deposits may increase water infiltration,
Figures 4.13 and 4.14, leading to increased pore pressures, a decrease in shear strength
and the potential for further landsliding.
At the change in slope a “waterfall” like feature may form leading to scour and the supply
of more debris to the flow, further increasing its mass and erosive power, Figures 4.1 and
4.8.
Rockhead inclined down slope tends to shed superficial deposits relatively easily and
does not tend to hold retain debris flow material, Figure 4.5.
Discontinuities within bedrock may be exploited by debris flow, providing more rock
debris and concentrating the erosive force.
Discontinuities dipping into the slope may form steps on rockhead where debris can
become trapped and lead to the formation of debris “dams”.
Relict rockhead cliffs and infilled gullies may provide a significant source of debris for
the flow and may lead to the formation of a “waterfall” like feature with associated scour,
Figures 4.5, 4.8 and 4.14.
Drainage culverts may become blocked forming debris “dams”, see above.
Inadequate drainage designs may lead to erosion and scour: e.g. inadequate wing walls,
erosion down stream of culverts and bridges due to venturi effect.
Tracks and drainage may concentrate surface water run-off.
Propagation of shear failure may occur in an upslope or downslope direction depending on
the point of maximum strain. Where the slope is undercut or where the interface daylights on
the slope, it is likely that shear failure will propagate upslope due to unloading. Where the
slope is in tension, usually marked by curvi-linear tension cracks defining the potential
headscarp, shear failure will propagate downslope due to loading. Bulging of the landslide toe
is a characteristic of the latter mechanism which marks the location where the shear surface
60
TYPES AND MECHANISMS
ruptures to the ground surface. In both cases, when 50% or more of the shear surface has
developed, rapid failure is likely to develop given a favourable pore water pressure regime.
Figure 4.14 – Location of the October 2001 Debris Flow on the A890 Stromeferry
Bypass. The main preparatory factors are highlighted.
Falls and Slides in Rocks: Upland mountain slopes or rock exposures where slope angles are
close to, or parallel, to the dip of the rock are particularly susceptible to rock falls or rock
slides. They are often characterised by pronounced headscarps and flanks which are relatively
free of debris, and a pronounced scree slope or debris fan at the base of the slope.
Detachment surfaces usually correspond to faults, joints and other structural discontinuities
where rock strengths are considerably less than those of the parent rock due to the effects of
long term weathering and transient pore water pressures. Rapid increases in pore water
pressures along rock discontinuities are a major cause of rock falls and rock slides. Icewedging along joints may also be important.
Debris Flow
Once the fall or slide is in motion and depending on the coherency of the displaced mass, the
failure breaks up on impact and as the slide avalanches downslope. The failure may develop
into a debris flow when the debris comes into contact with surface water and stream flow,
dramatically decreasing the viscosity of the debris-water mix14. As a general rule, where the
constituent particles of the slide debris cease to be in contact and become supported by fluids,
a change in mechanism from debris slide to debris flow takes place. This process is illustrated
in the slope failure that occurred along the A83 in Scotland (Figure 4.15). This transition may
14
Landslides may trap air resulting in ‘fluidisation’ and long run-out as an alternative medium to water, but this
mechanism is not considered further here.
61
TYPES AND MECHANISMS
be very rapid once the slide debris makes contact with surface water or stream flow (Figure
4.16).
Figure 4.15 – Upland debris slide and flow development along at Cairndow on the A83
in 2004.
Debris flows consist of a mixture of fine and coarse material, with a variable quantity of
water, which forms a muddy slurry that flows downslope, often in gravity induced surges.
Debris flows generally mobilise as a result of soil saturation, surface water flow, and high
pore water pressures developed within unconsolidated surface soils. Debris moves as a
combination of viscous flow and mass movement under gravity.
4.3.3 Propagation and Run-out Factors Affecting Debris Flow
Rapid upland landslides and debris flows can develop into large run-out flows. Whether or
not upland landslides develop into hillside debris flows and channelised debris flows depends
to a large extent on a number of conditions or ‘run-out’ factors, these being:
• The supply and mixing of surface water with the landslide mass in motion.
• The erosive capacity of flooded upland streams (channelised debris flows).
• An available source of sediment for entrainment in channelised debris flows.
• Slope steepness and length of slope for gravity induced slides and falls.
• The connectivity between hillslopes and upland stream channels.
62
TYPES AND MECHANISMS
1
Drainage off roads
Overflow causes
washout failure
Blocked
culverts
2
t0
t1
3
Surface erosion
destabilises
boulder
t2
t3
Oversteep slope
with loss of
suction
Wetting band
Local failure due to
surface saturation
to
t3 progressive
positions of wetting band
Recharge zone
upslope
t2 k = 10-4 m/s
t1
t0 k = 10-7 m/s
4
Piezometric
pressure in
confined channel
Rising pressure
k = 10-6 m/s
k = 10-3 m/s
5
to = time wetting band reaches aquiclude
/ aquitard
k of country rock >> k of aquitard
High
k = 10-7 m/s permeability
channel
recharge
10-6 m/s
6
10-6 m/s
-8
t3
t2
t1
t0
7
Dam (e.g.
weathered dyke)
Rising ground
water
10 m/s
Figure 4.16 – Mechanism models for open hillside failure caused by water.
Water plays a major role not only in the initiation of failure but also in the way that the debris
then flows or slides and the distance that it travels. Figure 4.17 illustrates many of the
important factors culminating in the exposure of society to safety and economic consequence
(item 7 on Figure 4.17).
The connectivity between upland landslides and stream channels is a very significant factor in
the propagation and run-out potential of debris flows. For relatively high frequency, shallow,
open hillside landslides, debris typically remains on the hillslopes or is deposited on the lower
valley slopes rather than being directly mobilised as channelised debris flow. However, for
low frequency high magnitude events, debris stored upon hillslopes and within valley floors
provides a source of generally unconsolidated sediment that can be entrained and mobilised
by channelised debris flows. It follows that the accumulation of unconsolidated debris from
numerous hillside landslides over time can provide a large volume of sediment capable of
being mobilised in a single episodic channelised debris flow event.
63
TYPES AND MECHANISMS
1. Rainfall
• intensity
• duration
• return period
2. Sub-catchment Hydrology
• area
• run-off characteristics
3. Susceptibility to Landslide / Erosion
• terrain morphology
• vegetation
• geology
4. Nature of Debris
• mobility
• potential for damming
and subsequent breach
6. Travel Path
• geometry
• potential for
entrainment
5. Basin
Hydrology
• upstream flow
characteristics
7. Consequence
• economic
• life (death / injury)
Figure 4.17 – Debris flow characteristics.
Where open hillside landslides or debris flows deposit directly into stream channels at peak
stage, mobilised sediments will be entrained and transported as a viscous flow (Figure 4.18).
Where landslides deposit into stream channels at other times, landslide dams may form
causing temporary lakes. As the volume of the lake increases, the erosive stresses imposed on
the unconsolidated material forming the dam eventually exceed its holding capacity, leading
to the collapse and break up of the dam. The collapse of landslide dams can be sudden,
releasing surges of water and debris downstream, in the form of debris flows, with destructive
effects.
Debris flows have high erosive energy and are capable of entraining material as they
propagate downslope or downstream (Figure 4.19). The entrainment of slope and valley
deposits often contributes a significant proportion of the volume of debris flows. This is
especially significant in channelised debris flows where colluvial and alluvial deposits occur
as ribbon-like stores along stream channel banks and beds (often as angular and sub-rounded
boulders within a sandy matrix) or broader accumulations in valley floors (valley floor stores).
Moore et al. (2002) indicate that, in steep upland catchments, debris flow run-out volumes
can be as much as eight times greater than the total volume of the source landslides. Five
main stages of debris flow propagation were recognised within the catchment: initiation and
64
TYPES AND MECHANISMS
detachment of material from hillslopes; transport and delivery of this material into the
channel system; storage of material within the channel system (and also, in the short-term, on
hillslopes before delivery to the channels); entrainment and run-out from the catchment; and
deposition on the debris fan. The linkages between these stages are critical.
Figure 4.18 – Entrainment processes increasing the size and nature of run-out
characteristics, Channerwick, Shetland Islands.
Figure 4.19 – Connectivity of Hillslopes and upland streams and their implications on
the run-out characteristics of debris flows, Channerwick, Shetland Islands.
4.3.4 Channerwick Peat Slide and Debris Flow Example
An example is provided of the dramatic peat slide failures triggered by an intense rainstorm in
September 2003 at Channerwick, Shetland Islands. The rainstorm was part of a slow moving
front which pushed south-eastwards across Scotland overnight and anecdotal evidence
indicates an average intensity of 33mm/hr. The intensity of the storm resulted in widespread
flooding of the hillsides (Figure 4.20) and burns and the initiation of rapid peat slides. The
latter developed into hillside debris flows with long run-outs, causing widespread damage to
65
TYPES AND MECHANISMS
roads and other infrastructure. Landslides occurred on slopes with angles between
approximately 7 and 25o.
Figure 4.20 – Upland intense rainfall characteristics at Hoswick Burn Shetland Islands
south mainland, 2003.
A key factor was the timing of the storm which followed a dry summer when groundwater
levels would have been low reducing the load of the peat blanket. Cracks within the peat will
also have formed providing conduits of surface water flow to the peat-weathered rock
interface. During the intense rainstorm, surface water filled the tension cracks and soil pipe
networks that connect the upper slope flushes and bogs to the lower slope peat blanket. The
relative impermeability of the weathered rock interface beneath the basal amorphous peat will
have caused a sudden increase in pore water pressure reducing the effective shear strength of
the overburden. Given the relatively low normal load of the peat overburden it is likely that
the pore water pressures could have resulted in ‘lifting’ (buoyancy effect) of the peat blanket
above the interface. Elsewhere, ‘bogbursts’ are widely reported where artesian groundwater
conditions develop within the soil pipe network and where rupture surfaces break out at the
ground surface.
Once movement is initiated, the partially saturated peat is ‘rafted’ downslope, initially upon
the shear surface and subsequently down steep sodden grassed slopes. The rafted blocks of
peat move rapidly as a debris slide with the peat blocks breaking down into smaller units as
the slide progresses (Figure 4.21).
66
TYPES AND MECHANISMS
Figure 4.21 – Changes in peat transport mechanisms, Channerwick, Shetland Islands.
67
5
KEY CONTRIBUTORY FACTORS TO DEBRIS FLOWS
by A Heald and J Parsons
5.1
HAZARD FACTORS AFFECTING DEBRIS FLOW OCCURRENCE
A wide range of factors may have a part to play in the triggering of particular debris flows in
the Scottish context, and indeed worldwide. Some of these may be considered fundamental
and must be in place, for example steep slopes (except in particular geological circumstances
such as the presence of peat). Others may be considered contributory, for example animal
tracks, but may nevertheless tip the balance of stability. In making a rapid and practical
estimate of the relative hazard of debris flows on a national scale, it is necessary to weigh the
relevant factors and, as a first pass, include only the highly influential factors in a simple
model. Detailed studies at particular sites may then, as a second or later stage, include
verification that more subtle factors are, or are not, in place. It may be that these stages lend
themselves to a GIS-based first pass approach, followed by more detailed examination of
aerial or satellite photographs and by ground truthing.
A number of landslide studies worldwide have used, as a primary indicator, the presence of
pre-existing landslides, determined either from historical records or from geomorphological
features. This has been used to determine both landslide hazard and other variables, such as
magnitude and run-out distances. It is not clear that this approach is appropriate to this case,
as some debris flows take place where there may be no precedent, or in the case of
channelised flows, evidence may have been lost by the more regular processes of erosion.
Furthermore, in assessing the wider subject of risk, experience shows that it is often the
unprecedented event that causes the greatest damage.
Various studies in Hong Kong (Evans and King, 1998) and in Nepal (Hearn and Petley, pers.
comm., 2002) indicate that slope angle and geological unit alone provide good correlation
with landslide occurrence and these factors formed the basis of simple predictive models.
These studies considered a wider range of landslide types than simply debris flows and it may
be that the underlying bedrock geology has a more or less direct influence in relation to the
current study. The following sections discuss each of the factors that may be considered to
influence the occurrence of debris flows in particular and in a Scottish context.
5.1.1 Topographical Factors
Slope Angle
There seems little doubt that slope angle is a fundamentally important factor influencing the
occurrence of debris flows. It should be borne in mind that the slope angle required to trigger
a flow may not be the same as the slope angle required to maintain the mobility of the flow in
its run-out zone. However, there is some overlap since there may be an angle above which
scour will add further material to the flow in its run-out zone.
It appears to be accepted that debris flows may be triggered at angles above about 30º and the
recent flow affecting the A83 at Cairndow (Figure 5.1) appears to be an example of this.
Similarly, the section of A83 leading up to the Rest and be thankful, some 7km south east of
Cairndow, has a history of being blocked as a result of instabilities in the slopes above. It is
interesting to note that the slope angles along this section are also generally 30° to 40°.
68
CONTRIBUTORY FACTORS
There is however some evidence of flows originating at angles as low as 26º, for example, the
A9 Dunkeld flows and possibly also the A85 Glen Ogle flows. It appears that fine granular
superficial materials are likely to flow at lower angles than coarser lithologies or cohesive
soils. Much lower angles are recorded in the special case of peat flows. It may be that the A9
Dunkeld flows rely partially on scour for their origin and may therefore be triggered at a
lower angle but it may also be a function of material type since the A9 Dunkeld (Figure 4.1),
and chiefly the A85 Glen Ogle, flows consisted of finer material than is generally reported
appear in the literature.
Figure 5.1 – Main characteristics of debris flow at A83 Cairndow
Further afield, the 1998 Pachagrande debris flow that destroyed the Macchupicchu power
station in Peru appears to have originated at a gradient of about 1v:2h (26.5º) whilst studies in
Hong Kong (Franks, 1999) concluded that ‘… most landslide sources originate in areas with
slope angles greater than 30°’.
It may be possible to impose an upper limit on susceptible slope angle on the basis that strata
likely to flow do not stand at angles greater than a certain value. There does not seem to be
very much information on this in relation to Scottish conditions but it is considered likely that
a maximum angle would not be greater than 45º to 50º. An angle of 46° is suggested Section
3.1.2, corresponding to the upper limit at which debris accumulates.
It would be recommended that any first pass hazard assessment should include slopes of 26º
to 50º and that this factor is considered to be of primary importance. Where the geological
formation is peat, then a lower minimum slope angle should be adopted.
Slope Height
It is not clear that any correlation exists between slope height and susceptibility to debris flow.
Of the August 2004 events, the vertical height of the main A83 Cairndow and A85 Glen Ogle
flows was similar at around 400m to 500m from source to limit of run-out, but the A9
Dunkeld flows were smaller by an order of magnitude. It may be relevant that the
69
CONTRIBUTORY FACTORS
Pachagrande flow, discussed above, and the Huascaràn stürzstroms of 1962 and 1970 were
almost an order of magnitude greater. Given that the materials involved behave substantially
as granular soils and may be modelled by a c=0 (purely frictional) analysis, then slope
stability theory supports the view that the probability of failure is independent of slope height.
This aside, it is interesting to note that the source of the majority of A83 Cairndow flows did
appear to start at a similar height on the hillside, as did many of the flows at or around the
A85 Glen Ogle, however it is considered that this may be a function of some other factor (e.g.
drift thickness, bedrock, spring line or change in slope angle) rather than a function of height.
It is not considered that slope height should be considered in the hazard model. It may be that,
all other factors being equal, a flow descending from 400m would be more damaging than
one descending from 40m and this could be considered in assessing hazard exposure.
Slope Aspect
Slope aspect relative to the key elements of bedrock structure is often considered an
important factor in landslide prediction. This seems most likely to be a potential preparatory
factor in the initiation of debris flows when the slope aspect and the direction of dip of a
relatively smooth rockhead profile coincide. Similarly it may be that a stepped rockhead
profile, or one with inward facing scarps relative to the slope aspect, are less likely to be
involved in the initiation of debris flows. Any correlation between slope aspect and type or
thickness of drift cover is likely to be too complex and the effect too subtle for incorporation
in the first stage model. Effects consequent upon steep northern slopes compared to gentler
southern slopes, for example, will be picked up by other means.
It is striking that the major flows in each of the August 2004 events all occurred on west
facing slopes and this can be extended to include the Stromeferry event. It may be that the
prevailing south-westerly weather patterns drive a greater degree of rain into the slope
causing a greater degree of saturation. Other contemporaneous flows at Cairndow faced
south and a minority of the smaller 18 August flows in the Glen Ogle/Strathyre district faced
north, south and east. An east-facing flow affected the B898 on the opposite side of the
valley to the A9 Dunkeld flows.
There is limited evidence that slope aspect alone is a reliable predictor of debris flows and it s
use in the model would require careful consideration. This factor, in combination with others,
is explored further in Section 6.
Other Topographical Influences
The presence of active stream channels and gullies tends to focus surface water runoff and
hence make channelised flow more likely. Terraces, ditches (natural or otherwise), and breaks
in slope may have a positive or negative influence on the formation of debris flows depending
upon their form or location. Rock outcrops or other natural or artificial barriers in the source,
transportation or deposition zones may retard the formation or impact of a flow.
These are issues that may prove important at the stage of detailed site appraisal and should be
included in the model at that stage.
70
CONTRIBUTORY FACTORS
5.1.2 Geological, Geotechnical and Hydrogeological Factors
Geological Formation
Since the flows largely mobilise unconsolidated deposits, the influence of bedrock geology
may at first be considered to be limited. Indeed, Vandine (1985) discounted underlying
bedrock as a predisposing factor for landslides in British Columbia. It may be surprising then,
that the three areas affected in August 2004, and the earlier Rest and be thankful instabilities,
were all in areas underlain by Dalradian schists; a rock type often associated with a relatively
low debris flow activity (Section 3.1.3). Similarly, the Invermoriston flow is in an area of
Moinian schist and the Stromeferry flow is underlain by older metamorphic rocks. Thus,
while any direct correlation between susceptibility to flow and bedrock type may not be
entirely clear at this stage, the tendency for schists and similar metamorphic rocks to weather
to produce fine soils consisting of platey minerals, may be significant. Further, the low
permeability of these rock types is likely to limit dissipation of pore water pressures by under
drainage. At least in one case, that of the A9 Dunkeld flows, the false bedded silty fine sand
that flowed does not appear to be locally derived and thus may be attributed to coincidence.
The apparent correlation between debris flows and schist should be considered in the light of
the preponderance of this and similar lithologies in the high relief areas of Scotland. Further
afield, Franks (1999) found that ‘… volcanic rocks were generally more susceptible to
landslides than feldsparphyric rocks’ in Hong Kong, but thought that ‘… this may have been
because the topographic relief is greater where the bedrock is volcanic’.
The presence of a mantle of superficial deposits is of fundamental importance to the
susceptibility to debris flows. It has been suggested that a critical thickness of around 1m to
2m may be most favourable to triggering a flow and this would appear to be supported by the
source areas of the debris flows at the A83 Cairndow, Rest and be thankful, A85 Glen Ogle
and from Hong Kong studies (Franks, 1999). The debris flow materials were predominantly
finely granular deposits, of glacial origin with the exception of the A9 event, which was
fluvial or fluvio-glacial. Given that glaciation affected all of Scotland and that the majority of,
if not all, steep sided slopes are expected to have a partial cover of glacial deposits, it is
unlikely that it will be possible to include this variable as a factor in the model.
In summary, while the solid geological formation is not in itself considered significant, the
lithology of the underlying bedrock is likely to be a secondary influence. The presence and
characteristics of a mantle of superficial materials is of primary importance but, given that
such a mantle may be thin, this information is not readily available in a GIS model and may
be difficult to discern with certainty by any form of remote imagery. It may be more practical
to assume at first pass that everywhere below the maximum slope angle has the requisite
mantle of superficial material and to filter out those cases where this does not apply by
walkover at the second stage.
Landslide History
As discussed above, the pre-existence of landslides is often considered to be a good predictor
of future instability. Although landslide history is an important factor in predicting future
instability, it is not clear that it is as useful in predicting fast moving debris flows as it is in
forecasting more slow moving progressive movements. However, evidence of past debris
flows on a slope is a good indicator that the conditions exist for future flows and this may be
considered an important factor at the stage of a second, more detailed, pass. Further, where a
71
CONTRIBUTORY FACTORS
debris flow has occurred in the immediate past and, for example, the vegetation has been
removed to expose the vulnerable soils beneath, there is no doubt that the area is more
susceptible to remobilisation if the trigger conditions (e.g. rainfall) should recur.
Geotechnical Factors
Soil properties including cohesion, grain size, shear strength, moisture content, void ratio,
relative density and permeability are relevant to the occurrence of debris flows. These are
likely to be known only as a result of a detailed ground investigation and should be picked up
during a second stage detailed site appraisal.
Earthquakes
Although flows worldwide have been triggered by seismic activity (e.g. Huascaràn 1962 and
1970), the occurrence and strength of earthquakes in Scotland is so low that their effect need
not be considered here.
Hydrogeological Factors
Studies in Canada (Vandine, 1985) and California (Reneau and Dietrich, 1987) indicate that
surface drainage is an important factor in controlling debris flow susceptibility, demonstrated
by the fact that most of the landslides studied occurred within or adjacent to significant
drainage lines or hollows. This pattern would appear to hold true for the A83 Cairndow, A83
Rest and be Thankful and the main A85 Glen Ogle (Figure 5.2) events.
Figure 5.2 – Source area of A85 Glen Ogle debris flow event.
The location of the ground water table is important in the prediction of any slope instability
but is difficult to estimate except as a result of detailed ground investigation. However, the
presence of spring lines is an important indicator. It may be possible to identify these
remotely from aerial or satellite photographs and published geological information.
72
CONTRIBUTORY FACTORS
Other hydrogeological and hydrological features that are relevant to the probability of
occurrence of debris flows include runoff coefficients and the size and shape of catchments.
Some of the factors may be obtained remotely and from pre-existing data sets, but others
would only be obtainable from detailed site specific studies.
5.1.3 Meteorological Conditions
Rainfall
There can be little doubt that rainfall is one of the single most important factors in triggering
debris flows in Scottish conditions. It is commonly accepted that the most frequent climatic
trigger for landslides worldwide is a heavy rainfall event following a period of high
antecedent rainfall. Of the August 2004 events, it appears that the A83 Cairndow and A85
Glen Ogle flows occurred after short intense summer storms, albeit against a background of a
wet summer, whereas the A9 Dunkeld flows followed more prolonged heavy rain.
The Meikle Tombane rain gauge approximately 7km from the A9 Dunkeld flows measured
77.5mm of rain on 9 August 2004, two days before the event. This quantity of rain on a
single day has a return period of approximately 50 years. During the three days 9 to 11
August, 171.3mm of rain was measured at Meikle Tombane and such a quantity of rain over
three days has a return period of just over 400 years.
The Lochearnhead rain gauge close to the A85 Glen Ogle event measured 80.8mm of rain on
18 August and this has a return period of 10 to 15 years. It is interesting to note that the
rainfall record indicates that 89mm of rain fell here on 10 August 2004 and this has a return
period of about 20 years. This rain gauge records rainfall only on a daily basis but anecdotal
information suggests that the rain was confined to a relatively short period for the day of 18
August. If the rainfall measured on 18 August fell in only six hours then the return period
would be about 120 years, if in 4 hours then the return period would be 250 to 300 years.
It is also notable that the burns in Glen Ample and the Keltie Water (draining Ben Vorlich
and Stuc A'Chroin to the south east of the debris flows that affected the trunk roads)
experienced much worse flood flow conditions than the Glen Ogle burn. Three bridges in
Glen Ample and five on the Keltie Water were washed away. In terms of return periods for
these two catchments it is estimated, based on observations in the glens, that the floods were
greater than 100 year events. In the Ogle Burn the flood debris indicates a much smaller
event, probably with less than a 10 year return period.
Information has been obtained from rain gauges about 20km away from the Cairndow event.
Their return periods do not suggest an extraordinary event but their distance away from the
site of interest may mean that they did not properly sample the event rainfall where the flows
occurred.
Thus, it seems that rainfall events of both long and short durations should be included in the
model. However, there are currently insufficient rainfall data to determine how much rain
has to fall over what time frame, before the likelihood of debris flows becomes a concern.
Further, it is expected that these ‘trigger levels’ will vary from area to area as soil
composition and other topographical factors come into play.
73
CONTRIBUTORY FACTORS
A major practical difficulty in incorporating rainfall into any model predicting debris flows is
predicting which geographical areas of Scotland may be subject to exceptionally heavy rain
over the lifespan of the model. It may well be considered that all areas of the highlands and
islands, and possibly the whole of Scotland, could be equally subject to this factor. In that
case rainfall distribution is no longer a variable in any predictive model, although, of course,
rainfall level remains a critically important variable. However with more information from
future instabilities, it may be possible to set rainfall ‘trigger’ levels as a short term
management tool.
Other Meteorological Factors
Of the other meteorological influences, snow melt is clearly a source of surface runoff and of
saturation of near surface sediments, thus increasing the likelihood of instability. Conversely,
frozen ground would be expected to be an inhibitor of debris flows. Wind, in addition to the
possible effect discussed above under ‘slope aspect’ in relation to driving rain into the slope,
may also have the secondary effect of uprooting trees with a consequent detrimental effect on
stability. These other meteorological influences are considered either too subtle or too
unpredictable to form a useful basis for a debris flow model.
It should be noted that these comments relate to the long term prediction of the influence of
meteorological conditions on a particular slope over a period of many years. The prediction
that a particular slope has an increased susceptibility due to a storm that is currently occurring
or imminently forecast, is quite a different matter.
5.1.4 Factors Related to Vegetation and Land Use
Vegetation Factors
Different types and densities of vegetation may be more or less retardant to debris flows
depending upon how they affect soil infiltration rates and upon how their root systems serve
to hold the soil in place. Landslides in Hong Kong during 1992/1993, occurred in terrain
with low scrub and grass rather than the dense tropical vegetation typical of the region.
Forestry in particular appears to reduce the probability of debris flows and may be considered
of primary importance. In British Columbia, policy has concentrated on controlling timber
harvesting and encouraging reforestation in the ‘source zone’. Forests may be picked up by
GIS and should be incorporated into the susceptibility model at an early stage. Other types of
vegetation may be considered to be less influential and also less readily identified remotely
and should be incorporated at a later stage.
Land Use Factors
Many land use factors may influence the likelihood of debris flows. These include
agricultural uses, the presence of buildings or other man made features such as hard-standing,
infrastructure or drainage. The influence of the old road in concentrating water flows was
demonstrated at the A9 Dunkeld failure (Figure 5.3) and forest tracks could be expected to
have a similar influence, as in the case of the washout that blocked the A83 Rest and be
thankful in the vicinity of Roadman’s Cottage, in 1999. Conversely, in the A9 Slochd failure
of July 2002, the presence of the trunk road contributed in a similar way to the failure of the
old road (used as a cycle path) and to its own failure by undercutting. In that case, a drainage
channel was another man-made feature that served to concentrate runoff and hence contribute
74
CONTRIBUTORY FACTORS
to the failure. In the case of the Stromeferry flow, it was an old field boundary/deer track that
created a pathway for preferential water flows.
Generally, these features are of local significance and would be difficult to incorporate into a
national model. They should, however, be incorporated at the site specific assessment stage.
Figure 5.3 – Influence of old road on debris flow at A9 Dunkeld.
5.2
HAZARD FACTORS AFFECTING DEBRIS FLOW RUN-OUT
5.2.1 Slope Angle, Height and Magnitude
(Volume of Material Delivered to Deposition Zone)
It is generally accepted that debris-supported flows (i.e. those in which there is particle-toparticle contact) including most or all of those that have affected Scottish trunk roads in
recent years, will flow at slope angles at or above 11º. The 1998 Pachagrande debris flow,
referred to above, is an example of a debris flow that conforms to this limiting slope angle.
Hungr et al. (1987) defines a confined channel as one with a width to depth ratio of less than
five and reported (Hungr et al., 1984) that deposition will occur on slopes of 10° to 14° for
non-channelised flows and 8° to 12° for channelised flows. This agrees well with studies in
Hong Kong (Franks, 1999). Water-supported debris flows (i.e. where the particles are not
generally in contact) often flow at angles at or above 2º. Observations of Scottish debris
flows indicate that they are arrested at angles steeper than 2º.
As discussed in Section 5.1.1 above in relation to probability, there seem to be no limiting
factors related to the height (or length) of run-out.
Along the Cairndow section of the A83, it was observed that, of debris flows originating at a
similar height, ‘smaller’ flows did not tend to reach the A83. Whilst there may have been
subtle differences in the factors affecting the run out channel characteristics (i.e. angle of
slope), it may suggest that there is a certain volume of material required to gain sufficient
75
CONTRIBUTORY FACTORS
momentum to reach the road. However, a more detailed investigation would be required to
confirm this.
5.2.2 Channel Characteristics
In channelised flows, the cross-sectional shape of the stream channel, its width and depth in
particular, may be expected to affect the length and volume of the run-out. Similarly, the
longitudinal shape of the channel may lead to zones of deposition and zones of erosion along
its length and these may vary with the intensity of different stages of the flood. Further, the
smoothness of the channel may promote a longer run-out and this may in turn be a function of
topography, geology (drift thickness, bedrock type and structure), obstructions or
constrictions (natural or artificial), and history of debris flows.
The sinuosity of the channel may absorb the energy of the flood and thus retard it. However,
it may also result in increased erosion on the outer sides of bends and in this way debris may
be added to the flow. Bends in the channel may affect the direction of run-out and thus the
effects of the event. The classical example of this relates to the 1970 Huascaràn stürzstrom in
which the town of Yungay was thought to be protected by a 150m high hill that deflected the
channel to the south. However, one branch of the flow failed to turn the bend, surmounted
the hill and resulted in a reported 18,000 deaths in Yungay. The recent Glen Ogle flow,
though on a much smaller scale and fortunately without casualties, followed a very similar
pattern. The early part of the flow followed a sharp left-hand bend (Figure 5.4) in the stream
channel, thus damaging a culvert and a section of the road. A later pulse did not turn the bend
but had sufficient momentum to continue straight ahead over a rock outcrop, sweeping away
a vehicle that might have been thought to be protected by the outcrop. In this way, the width
of the run-out was increased and a greater length of the road was affected. Conversely, in the
case of the A83 at Cairndow a ridge at the toe of a drainage channel successfully prevented
one debris flow from reaching the road.
Accordingly, the hydrological factors affecting run-out can be seen to be complex and are
thus best reviewed on a site specific basis.
5.2.3 Vegetation and Land Use Factors
The surface conditions in the run-out zone may permit or impede the run-out of the flow.
Afforestation may be particularly important in retarding flows as seen at Cairndow, but other
conditions, such as hard surfacing or pasture land may be much more permissive to flows.
Uprooted trees can contribute to the power of the debris flow. This was seen at the A9
Dunkeld, where trees formed part of the debris that reached the road and trapped vehicles and
at Glen Ogle, where trees were swept into the culverts and formed part of the blockage.
Uprooted trees have caused significant damage in larger scale events in the Himalaya, for
example in the Hinku valley of Nepal and at Punakha in Bhutan where a temporary dam of
trees deflected the flow with a resultant loss of life.
On a Scottish scale afforestation is, in most cases, likely to retard run-out and this may be
considered an important factor in assessing the effects of a debris flow.
76
CONTRIBUTORY FACTORS
5.3
FACTORS AFFFECTING EXPOSURE TO DEBRIS FLOW HAZARDS
The key factor in relation to the exposure that results from a debris flow is whether or not the
flow reaches a vulnerable element. As this study is focused on trunk roads and trunk road
users, this key factor becomes simplified to whether or not the flow is, or is not, expected to
reach a trunk road or associated infrastructure. Clearly, if there is no possibility that the flow
will reach a trunk road (or associated infrastructure) then both the hazard and the hazard
ranking (see Section 6) become, for the purposes of this study, zero.
In cases where a trunk road is present within the modelled run-out zone of a flow, it would be
possible to prioritise actions based on the scale of the exposure as discussed below.
Figure 5.4 – View of the larger of the A85 Glen Ogle debris flows, showing the sharp
bend in the channel just above road level.
5.3.1 Factors Related to Road Usage
Clearly, the potential exposure in relation to death or injury to members of the public are
greater where traffic flows are greater. Debris flows tend to be fast-moving compared to
most other forms of landside and frequently wash down very large boulders, as seen in the
Cairndow (Figure 5.5) and Glen Ogle events. Any washing down of large boulders, or indeed
other large items of debris, has the potential to cause serious injury or fatality.
As trunk roads comprise, by definition, the country’s first level strategic road network, factors
to be taken into account by this study will include traffic flows, sightlines and the availability
77
CONTRIBUTORY FACTORS
and length of diversion routes. Traffic flow relates to the likelihood of a debris flow event
affecting road users, whilst sightlines will determine the potential for the road user to take
avoiding action. The availability and length of a diversion route may be seen as an analogue
for the economic impact of such an event. This may be complicated by the possibility of
alternative routes becoming blocked by other contemporaneous debris flows resulting from
the same weather conditions or other factors. In the cases of both the recent Dunkeld and
Glen Ogle events, minor roads in each area were also blocked by separate but related events.
Figure 5.5 – Debris fan containing boulders (estimated up to 9 tonnes) at A83 Cairndow.
In summary, it may be considered that traffic flows and are a key factor that may be utilised
for prioritisation in a national plan. The other factors discussed here are more subtle and may
be considered on a site specific basis.
5.3.2 Factors Related to Emergency Response
The seriousness of an event may be exacerbated or minimised by the ease of emergency
response. For example, at the A9 Dunkeld event the police were able to attend the scene
within a few minutes and to assist motorists from their vehicles. This may not be the case in
a more remote location. At the A85 Glen Ogle event, BEAR personnel were rapidly on the
scene and provided assistance but, with 20 vehicles and 57 motorists isolated between two
debris flows, the decision was wisely taken to effect evacuation by RAF and Royal Navy
helicopters. The events of August 2004 suggest that when debris flows occur, multiple events
should be regarded as highly likely and thus there is a reasonable chance of the public
becoming trapped or the main emergency becoming inaccessible to emergency vehicles.
Clearly, the use of helicopters can reduce the effect of both remoteness and of multiple debris
flows.
Police and military personnel and trunk road maintenance staff are trained in emergency
procedures and during recent events provided an excellent service. However, assessing the
likelihood and location of any further debris flows is not part of their capability. Depending
78
CONTRIBUTORY FACTORS
upon the location of the emergency, there is always likely to be an interval of several hours
before a geotechnical specialist with experience of landslides can attend the scene to assess
the current and near-future hazards.
In such events, the alarm is often raised rapidly by motorists using mobile telephones. There
are however areas in Scotland where there is no mobile telephone coverage. These may be
areas that are susceptible to debris flow activity and the seriousness of any event occurring in
such locations could therefore be exacerbated by this factor.
It may be considered appropriate to include these factors relating to access and the ability to
rapidly raise the alarm in the determination of hazard ranking of particular routes. However,
such areas are likely to be remote, have lower traffic flows and therefore affect fewer people.
Such actions in the hazard ranking may therefore undermine the need to target resources
where there is the greatest need, typically identified by the greatest traffic levels.
5.3.3 Factors Related to the Local Value of the Asset.
Factors considered here reflect the value of individual assets on the network and the likely
cost of repair, for example damage to structures is likely to be more expensive to repair than
damage to the carriageway surface or to an earthwork.
It is also important to consider the environmental implications of a debris flow. Whilst the
primary concerns of the work here are in ensuring that the exposure of the road using public
to potentially dangerous and adverse economic debris flow events is minimised, clearly some
account of the environmental impact of debris flow is required.
Factors relating to environmental issues and designated areas would need to be assessed on a
site specific basis.
5.3.4 Publicity and Political Factors
There is a potential for adverse publicity to be associated with any event that causes a trunk
road to be closed although this may be diminished if, as in recent events, casualties have been
avoided and the response is timely and efficient. The difficult question would be whether
roads should be closed on the basis of a forecast event in any particular location and how the
non-realisation of such an event would be perceived by the public and the media.
It is considered that the assessment of this factor is beyond the remit of this study.
5.3.5 Secondary Effects
Debris flows may not only have a direct effect on a trunk road but there may also be ‘knockon’ effects. For example, the debris may dam a river causing impounding of water and
inundation upstream. This was the mechanism for destruction of the Macchupicchu power
station following the 1998 Pachagrande debris flow. Subsequent bursting of such a
temporary dam may cause further destruction downstream. Either of these situations could be
damaging to a trunk road or to trunk road users. The potential for secondary effects would
need to be assessed on a site specific basis.
79
CONTRIBUTORY FACTORS
5.4
SUMMARY OF KEY CONTRIBUTORY FACTORS
A wide range of factors may contribute to the likelihood of, and exposure to the effects of, a
debris flow in a particular location. Some of these are widely applicable and others are subtle
but may make a critical difference at a particular location. Furthermore, some are readily
assessed using GIS and/or remote sensing whereas others are only discernible to the expert
after close inspection. It is likely that it will be necessary to base a first stage hazard
assessment and hazard ranking upon the more widely applicable and readily obtainable
factors and then to carry out secondary and subsequent filters using more site specific and
difficult to assess factors.
Rainfall or other source of water is a critical factor but it has been assumed that all parts of
the trunk road network may be subject to excessive rainfall and so this has been excluded as a
differentiating factor.
For a risk or hazard to exist at all, the conditions must allow a debris flow to occur and must
allow the run-out of such a flow to reach a trunk road, a trunk road user or other infrastructure
or feature that can impact upon that road or road user. As a first pass, there are three critical
factors that could be obtained rapidly and remotely from a GIS to assess whether these
conditions are in place:
• A source area where the slope angle is greater than 26º and less than 50º.
• A run-out zone where the slope angle is greater than 8º.
• A trunk road is present within either of the above zones.
It should be noted that peat can flow at much lower angles than these and it would be
appropriate also to perform an alternative first pass in which a search is carried out for all
trunk roads passing through areas of peat.
Perhaps the next most important factors are those that would allow prioritisation of particular
routes or parts of routes, particularly traffic flows, the strategic importance of the route and
the length and viability of diversions.
There are a number of influential factors that should be considered at the second stage and
possibly the most important of these is afforestation. Other significant topographical features
may be considered at this stage along with the lithology of solid and drift geological deposits
and the landslide history. Perhaps the next most critical factors relating to the seriousness of
the event may be the factors affecting the emergency response and possibly the publicity and
political factors.
Other factors such as vegetation and land use, animal and anthropogenic factors, slope aspect,
detailed topography, geotechnical, hydrological and hydrogeological factors, local structures,
environmental implications and secondary effects would need to be considered on a site
specific basis but it may be necessary to bear in mind the possibility of ‘knock-on’ effects at
all stages following the first pass.
80
6
PROPOSED METHODOLOGY FOR DEBRIS FLOW
ASSESSMENT
by M G Winter, F Macgregor and L Shackman
6.1
HAZARD ASSESSMENT
The purpose of the proposed hazard assessment is to determine stretches of the trunk road
network most likely to be affected by debris flow activity. This will involve the sequential
discarding of unlikely areas, at least in the early stages.
Two initial sifts are likely to be undertaken. The first will differentiate between peat and nonpeat drift deposits. The second will be based on slope angle. A relatively high slope angle
(around 26o to 50o) will be applied to debris flow formation in non-peat deposits with a slope
angle of around 8o applied to the run-out zone (between, for example, the area in which debris
flows might form and a trunk road). Soils that are known to exhibit cohesion may have a
hazard reduction factor applied as they may be less susceptible to debris flow activity. A
relatively low slope angle (possibly as low as 5o) will be applied to areas that are covered by
peat. This approach means that most effort in assessing hazard is expended in those areas of
greatest actual hazard rather than an initial ‘blanket’ approach which expends considerable
effort in all areas.
The National Landslide Hazard Assessment (NLHA), based upon NEXTMap and other data,
for landslide hazard zonation (see Section 3.3) can be rapidly adapted to suit the purposes of a
bespoke initial assessment. In addition to the factors described above, some account of
engineering soil type is also incorporated. The NLHA data set incorporates information from
both published and unpublished drift and bedrock geology maps, the unpublished information
not being otherwise readily available to any other form of assessment. Proxy data in respect
of friction angle ( ’) have been developed from drift and bedrock geology descriptions and
are held within the data set.
It is understood that the Meteorological Office have data on regular storm tracks for intense
rainstorms across England and Wales. While it is not entirely clear whether such information
is currently available for Scotland, it would be a very useful means of refining the initial
evaluation of hazard if available. A proxy for antecedent rainfall data could also be delivered
by means of the 30-year average rainfall figures for Scotland. However, some care is required
as such data may not reflect the current rainfall patterns and also ignore the issue of low water
content-high suction slopes that can be vulnerable to intense rainfall events (see also Section
2.3).
It is also perfectly feasible to attach a higher assignment of hazard for conditions relating to
stream channel and catchment areas, thus emphasising the perceived hazard of debris flow
development associated with stream beds.
81
PROPOSED METHODOLOGY
A key issue is the swathe width to be examined and clearly this must correspond to the
catchment adjacent to the road and the location of the watershed may well be the most
appropriate measure to define the relevant swathe. However, the effort required to undertake
the assessment is, to a large extent, independent of the geographical area included. This opens
up the possibility of undertaking the assessment for the entire land mass of Scotland, rather
than simply for land adjacent to trunk roads, thus providing valuable data for Local Authority
roads. This approach also has the advantage that work is not necessary to identify the limits of
all of the catchments adjacent to the trunk road network, albeit that these are available
electronically from the Flood Estimation Handbook (Anon, 1999) which includes details such
as return period, capacity, flow and other data.
In addition to the basic hazard assessment other key outputs have the potential to be sourced
from the GIS, as follows:
• Inventory: An inventory of areas of high hazard in Scotland and, more specifically, areas
that have a potential to affect the trunk road network may be developed. This could be
refined to determine the lengths of the trunk road network subject to hazard from debris
flow activity. Inventories could also be developed for local authority roads as a panScotland approach is proposed.
Some care will be needed to ensure that the inventory is not too restrictive as criticism may
well be levelled at the system if debris flows subsequently occur outside the areas
identified. Notwithstanding that, not all areas identified in the inventory will be selected
for further investigation and/or management measures or mitigation. There remains a
possibility that such areas will be subject to debris flow activity in the future, albeit that
the possibility will be substantially less than those areas that are selected for further action.
• Mapping: The mapping of potential debris flow areas is a potentially valuable tool to
portray the hazard assessment results. Not only could the initial, GIS-based hazard
assessment be illustrated in this way, but results of the subsequent site-specific, more
detailed hazard assessments could also be incorporated. In addition, both the exposure
levels and the hazard ranking (see Section 6.2) could be illustrated in map form. The
foregoing is, of course, subject to the data being available in a suitable format.
Once areas of high potential hazard are identified then more substantive, site specific efforts
may be expended in using a system developed on the basis of the assessment factors
described in Section 6.3. As a first step it is proposed that a ‘ground-truthing’ exercise be
undertaken by making a desktop comparison of the results from the initial GIS-based
assessment with those areas of high hazard assessed during the Project Workshop and
detailed in Section 7.
In terms of the site specific work it is crucial that a range of sites representing as fully as
possible the full range of conditions likely to be encountered relative to debris flows in
Scotland is evaluated. Further, while the evaluation of each site should be undertaken by one
member of the Working Group an independent check should be undertaken by another
member of the Working Group. In each case the results must be compared, evaluated and
audited by the Project Team.
It is important to emphasise that any form of hazard assessment will determine the most likely
areas to suffer debris flow activity. It remains possible that areas identified as having lower
likelihood may experience such events if circumstances come together to provide the
necessary triggers. As has been pointed out on many occasions the precise nature of the
82
PROPOSED METHODOLOGY
ground is uncertain and residual hazards must be expected. However, the judicious use of a
system such as that proposed should ensure that apparently anomalous events are rare and
that the hazards are managed to the best effect within budgetary constraints.
6.2
HAZARD RANKING
The assessment of hazard in isolation simply details the areas most likely to be affected – the
likelihood of occurrence; it does not consider the consequences of such events. In order to
enable the appropriate prioritisation of management budgets for potential debris flow activity
on the trunk road network the exposure due to the interaction of debris flows on the network
must also be evaluated.
Traditionally the product of the hazard and exposure (or consequence) is defined as the risk.
However, there are a number of ways in which exposure might be considered. In an ideal
world the exposure resulting from debris flow activity would be determined in all its contexts.
Such contexts include the exposure to life and limb, social/employment factors (including the
effect upon tourism), environmental factors and economic factors. To include all such factors
would be a major undertaking and is almost certainly beyond current capabilities in terms of
fully understanding the interaction between the factors and ensuring that there is no doublecounting (or even treble-counting) of the exposure factors. A thorough view of risk
assessment in the context of landslides is presented by Lee and Jones (2004).
Accordingly the product of hazard and exposure is referred to in the limited, but direct, sense
in which it is evaluated as a hazard ranking. The hazard ranking may be seen as a
qualitative/semi-quantitative risk assessment as opposed to the fully quantitative conventional
risk assessment approach.
The complexity of the interactions of exposure factors means that many are underpinned by a
few relatively simple measures such as traffic flow, road geometry (especially sightlines), and
the length and, indeed, the existence of a diversion route. These factors are capable of
capturing a simplified assessment of exposure and thus being imposed on the basic
assessments of hazard to provide the hazard ranking described above. The process does not,
however, represent a full risk assessment and nor is such a process either necessary, or
desirable, in this case.
Clearly, debris flow activity on the busy A9 to the north of Perth (traffic flow around 13,500
vehicles per day – all vehicles two-way, 24 hour AADT15) would have a far greater effect due
to the higher traffic flows (and higher number of people dependent upon such traffic
movements) than on the much more lightly trafficked A835 between Ullapool and Braemore
Junction (traffic flow around 2,900 vehicles per day), for example. If two such lengths of road
are found to have the same level of debris flow hazard (i.e. the same likelihood of a debris
flow interacting with the road) then some means of distinguishing between the two and
adopting a prioritisation approach to management and mitigation is required.
Using the simplified exposure evaluation technique described above, it is thus almost certain
that of the two examples cited the A9 would be assigned a higher priority than the A835. This
15
Note that the traffic flow figures are highly variable on a seasonal basis. The figures quoted are the maximum
figures available from 2003 and 2004 records and are generally in either July or August. The minimum figures
were 7,200 for the A9 and 400 for the A835 in either January or February.
83
PROPOSED METHODOLOGY
is entirely appropriate as the interests affected (businesses, commuters, tourists, etc) by such
events would be much greater than on the A835. In addition, the traffic flows on the A9 are
much higher and the chances of personal injury are therefore proportionately higher, albeit
that this aspect is offset to some extent by the presence of generally better sightlines and
geometry on the A9.
Accordingly it may be seen that once the level of hazard has been determined then a further
assessment of the exposure must be applied to a given situation to yield what we describe
henceforth as a hazard ranking. The purpose of this hazard ranking is primarily to distinguish
between areas with similar hazard levels to allow budgetary decisions to be made on an
informed basis. Secondly, as indicated above, it is clear that areas with lower hazard levels
may yield higher hazard rankings than areas with higher hazard levels which may yield lower
hazard rankings. The foregoing, purely hypothetical, comparison of the A9 with the A835
may well be a typical example of such a situation. The effects of an event on the A9 being so
much greater than one on the A835 that the actual level of hazard alone does not determine
the need for action or otherwise.
6.3
DETAILED ASSESSMENT FACTORS
The factors generated at the Project Workshop were very detailed and comprehensive (see
Section 3.2). However, it is clear that many factors have the same, or similar, root. For
example it could be argued that depositional regime is the root factor for others such as
density, relative density, air voids, void ratio, permeability and even saturation.
In this context it is clear that some effort is required in simplifying the factors determined
from the Project Workshop. Indeed, this section does not address how they will be defined,
but merely identifies the most important combined factors. Combining the factors and the
method for doing so is a matter to be addressed at an early stage of Study 1, Part 2.
The factors given below are considered to be a strong reflection of those that must be
incorporated into a hazard assessment and ranking system. Clearly some are likely to be used
at a very early stage (e.g. a GIS-based assessment) while others will be incorporated into a
site-specific assessment methodology. However it is also recognised that some refinement of
these factors will be undertaken as the construction of a working hazard assessment and
ranking system is constructed.
6.3.1 Hazard Factors
In Section 5 key contributory factors to debris flow hazard are discussed in the context of
those that affect likelihood of occurrence and those that affect the consequences of the event.
The first stage assessment, as described in Section 6.1, considers two categories of debris
flow hazard assessment.
• The first will effectively seek areas of peat on slope angles of 5o or greater. While the
presence of streams in the peat will be evaluated these will almost inevitably be present
and further work on a site specific basis will be required.
• The second will deal with all other types of surface deposit. The slope angle, engineering
soil type and presence or otherwise of a stream will all be taken account of. Note that the
84
PROPOSED METHODOLOGY
presence of a trunk road above, not just below, a hazard zone may present a threat to that
road.
Other factors will be incorporated as the availability of data permits.
Once the first stage assessments have been undertaken then more detailed examinations of
areas of high hazard will be required. It is likely that all areas of high hazard in peat will
require site specific assessments. The key issues for further desk-based assessment of areas of
high hazard in non-peat deposits are as follows:
• The presence of a trunk road within the area of high hazard or the presence of a suitable
run-out zone (slope angle 8o or greater) between the area of high hazard and the trunk road.
• The presence of other topographical features that may enhance the likelihood of debris
flow occurrence. These include terraces, ditches (natural or otherwise) or breaks in the
slope which may have either a positive or negative impact on debris flow formation and
transportation and rock outcrops and other natural or artificial barriers that may retard the
formation or passage of a debris flow.
• The existence of a history of landslide activity at the location. Such information is
available from the National Landslide Database and BGS digital maps as well as from
experience and observation.
• Factors relating to bedrock will require some further investigation. Much of the available
research on Scottish debris flows has indicated a limited history of debris flows in areas of
schist bedrock materials for example. However, much on-the-ground experience
contradicts this, especially in localities where the direction of bedrock dip has been found
to approximately coincide with the slope aspect.
• Catchment data such as runoff coefficients and the catchment size and shape (Anon, 1999).
• The presence of spring lines.
• Deforestation and afforestation as factors potentially increasing and decreasing the
likelihood of debris flow activity at a given location. In the case of deforestation the
direction of old planting furrows should be taken into account as these may direct water
into the area of high hazard. Afforestation is a particularly important factor to consider in
the context of arresting or retarding debris flow runout.
• The presence of features such as public or forest roads between the area of high hazard and
the trunk road. These may slow the progress of water and thus increase the deleterious
effects of water ingress immediately below the feature and/or the presence of culverts
passing under such roads may delay the downslope passage of debris and thus increase the
debris load of future events.
Storm track data will be incorporated if available, and 30-year average rainfall data will be
used as a proxy for antecedent rainfall if the advice of the Meteorological Office concurs with
its use in this context.
Slope height, slope aspect, earthquakes and the underlying geological formation are all
considered to be factors that have limited influence on the potential for debris flow
development. However, where slope angle and the dip/direction of bedrock are known to be
coincident then this might be a factor that adds to the perceived level of hazard. In addition
the presence of a layer of drift and/or weathered bedrock deposits is considered vital for the
85
PROPOSED METHODOLOGY
development of debris flows: this must be neither so thin as to provide inadequate material to
develop a debris flow nor so thick as to damp the dynamic flow.
Detailed geotechnical factors, other than as described above and including the location of the
water table, are unlikely to be available at other than a detailed site appraisal stage in which
specific mitigation measures are being evaluated.
6.3.2 Exposure Factors
There are three main factors that would ideally be incorporated into the assessment of
exposure for the system. These are as follows:
• Traffic flows which not only give an estimate of the likely number of vehicles that will be
delayed due to an event, but also give an, admittedly indirect, evaluation of factors such as
the potential for personal injury and indeed the potential damage to the local economy.
• Factors related to road geometry, such as sightlines and carriageway width, determine the
forward visibility available to drivers at a given location. This, in turn, describes the
potential visibility of a hazard and therefore the potential for the driver to see it in time to
stop or take other appropriate avoiding action. Clearly sightlines will become less relevant
at night when the distance that a driver can see will be determined by the efficacy of the
vehicle lighting.
• Diversion length improves the estimate of the potential damage for the local economy,
albeit still in an indirect sense. This will be improved if the suitability of the diversion for
the disrupted traffic levels, see item (a) above, and for HGVs can be assessed. Clearly if
there is no diversion then the hazard ranking will need to reflect this fact.
6.3.3
Compatibility with Existing Systems
Having discussed these factors it is also clear that the other landslide hazard assessment and
ranking system in use on Scotland’s trunk road network needs to be taken into account. The
Rock Slope Hazard Index system (ROSHI), developed by McMillan and Matheson (1997) for
the Scottish Executive’s use on trunk roads, considers only rock slopes. However, it gives a
hazard ranking for the purposes of budgetary prioritisation of management and mitigation
measures. Clearly having the two systems running in parallel and on an entirely different
basis would severely restrict the ability of the Executive to make rational decisions on
expenditure and to compare rock fall hazards with debris flow hazards. As such it is
important that the end results from the two systems can be compared.
It should, however, be recognised that the approach to the ROSHI is specific to rock slope
instability and it is not the most desirable approach for the development of a system for debris
flows. For example, the assessment of debris flow lends itself to a GIS-based initial
assessment. This means that areas that satisfy specific, multiple criteria are identified. In
contrast the ROSHI takes a sequential approach to building up a hazard ranking number (see
below). As such it is proposed that the present debris flow system adopts the same number of
hazard ranking categories and that efforts are made to ensure that these are as comparable as
possible with the ROSHI. Not least among these efforts should be that the assessment of
exposure used in the ROSHI categories is adopted for the debris flow system, with changes
only as required to reflect the different nature of the hazard.
86
PROPOSED METHODOLOGY
The exposure factors used in the ROSHI are as follows:
a) Sightline.
b) Road type (single track, single carriageway, wide single carriageway, dual carriageway,
two-lane motorway, three-lane motorway) – carriageway width and NESA/COBA speedflow relationships.
c) Traffic flow (24-hour, 2-way, AADT).
d) The existence of services and/or other structures above the road.
e) The existence of a downwards slope, river or loch immediately to the opposite side of the
road from a potential failure.
Factors (a) to (c) are closely related to those described in Section 6.3.2.
Essentially, the possible range of values of each factor is split into sub-ranges and the
intermediate range assigned a parameter value of unity. Factor values in higher and lower
ranges are assigned higher and lower parameter values, respectively.
Sightline parameters (item (a) above) are based upon stopping distances from the Highway
Code. Thus for single track and single carriageway roads a parameter value of unity is set to
the sightline range of 40m to 60m, indicating that vehicles travelling within the speed limit
are as likely to hit a block on the road as not 16 . For sightlines less than 40m a greater
proportion of vehicles are likely to hit the block and the parameter is increased accordingly.
Similarly for sightlines above 60m a greater proportion of vehicles will stop before hitting the
block and the parameter value decreases. The equivalent sightline range for dual carriageways
and motorways, corresponding to a parameter value of unity, is 60m to 100m.
Carriageway width (item (b) above) gives an indication of the potential space available for
avoidance manoeuvres in the event of a rockfall both at the time and if the rockfall has come
to rest on the road. The ROSHI parameter values range from 0.7 to 1.2 with unity set for the
6m to 8m width range. However, as a debris flow event is likely to cover the full road width
in a very short period of time the opportunity for avoidance manoeuvres is severely limited
and carriageway width is thus less pertinent to debris flows. A typical debris flow is likely to
close the road for a longer period of time than a typical rockfall and, as previously observed,
diversion issues come to the fore in place of avoidance.
Traffic flow (item (c) above) is used along with other data in the ROSHI to derive traffic
parameter values. Speed flow relations ships combined with AADT 2-way flows are used to
obtain an indication of speed from, which parameter values for each of the six different road
types in item (c) above are derived. Although it is also claimed that design speed is
incorporated into the assessment of traffic parameter values, it is not clear from McMillan
(1995) how this achieved.
16
This approach appears to allow for shorter stopping distances than those given in the Highway Code (source:
www.highwaycode.gov.uk) for the speed limit of such roads. It is assumed that this approach has been adopted
to allow for a range of actual vehicle speeds. For example, taking speeds of 40mph to 60mph gives stopping
distances of 36m to 73m. Allowing for some rounding the range of 40m to 60m seems reasonable for cars in dry
conditions with good visibility; given that at most hazard sites speeds close to the speed limit are highly unlikely.
However, for debris flows which are likely to be encountered in wet conditions with poor visibility some
adjustment may be necessary for conditions and possibly also for vehicles with longer stopping distances.
87
PROPOSED METHODOLOGY
Factors (d) and (e) have relatively little influence on the overall outcome and are used to
make small percentage adjustments to the overall hazard ranking score.
Factor (d) does however raise the question of whether the Scottish Executive should expend
additional effort in including statutory undertakers’ services in the hazard ranking assessment.
This could potentially divert the assessment from its primary aims of protecting life and limb
of road users and the economic activity to which Scotland’s road network is so vital. As the
presence of statutory undertakers’ services is not immediately apparent and they will
potentially be protected by any mitigation works undertaken it is suggested that such services
not be included in the evaluation of exposure.
The way in which ROSHI combines factors related to exposure and hazard is complex. Data
and their effects are amalgamated by data type. Thus factors such as sightlines and
carriageway widths are combined into the assessment along with slope angle, slope height
and other factors relating to slope geometry in a section entitled, not surprisingly, geometry.
This approach makes it very difficult to assess exactly how each factor contributes to the
assessment of exposure. Notwithstanding this it is clear from an assessment of McMillan
(1995) that while this approach may be unusual it does not introduce errors, only
difficulties in comparing the outputs with other systems. In the proposed debris flow
system it is recommended that the assessment of hazard remain separate from that of exposure
in order to ensure that the development of each is clear.
The proposed evaluation of exposure for debris flows thus becomes
i) Sightline.
ii) Carriageway width (small percentage change to the hazard ranking).
iii)Traffic flow (24-hour, 2-way, AADT) and road type/speed-flow relationship as an
indicator of relative traffic speed.
iv) Diversion existence, length and suitability.
v) The existence of vulnerable structures in the path of any potential debris flow (small
percentage change to the hazard ranking).
vi) The existence of a downwards slope, river or loch immediately to the opposite side of the
road from a potential failure (small percentage change to the hazard ranking).
The associated categories of hazard ranking and their scoring in the ROSHI are follows:
• No [immediate] action required (score 0 to 1).
• Review in five years (score >1 to 10).
• Detailed inspection (within two years) (score >10 to 100).
• Urgent detailed inspection (score >100).
It is recognised that the above descriptive categories are not entirely suited to debris flows.
However the important issue will be to ensure that the two systems are as comparable as
possible while recognising that the hazard evaluated and therefore the approach to their
evaluation are different. Thus measurement of exposure must be as similar as possible for the
two systems, as described above, and there must be the same number of, broadly comparable,
hazard ranking categories.
88
PROPOSED METHODOLOGY
Once a hazard ranking is established for a range of sites then priorities can be set within the
context of planned maintenance and capital works. Areas and sites viewed as the most
vulnerable can then be subjected to well-targeted and justified management and mitigation
actions as discussed in Section 8.
89
7
HIGH HAZARD AREAS AND EARLY OPPORTUNITIES IN
SCOTLAND
by M G Winter, F Macgregor and L Shackman
With the bringing together of such a range of geotechnical specialists, the majority of whom
have a detailed knowledge of the Scottish trunk road network, the Project Workshop
presented an excellent opportunity to identify what might be termed ‘at-risk’ sites on the
network, where early investigation of potential debris flow occurrence would be likely to be
productive.
The early identification of high hazard sites, on a subjective basis by acknowledged
specialists, would serve the joint functions of assisting prioritisation of areas for action under
Part 2 of the study, whilst providing, in parallel, a shortlist of sites appropriate for validating
the debris flow hazard model in its development phase.
A listing of the areas considered to present a sufficiently high hazard to warrant concern was
set out. Subsequent to the Project Workshop, Digital Ordnance Survey mapping at the
1:50,000 scale was used to inspect the areas identified in both plan form and also using a
digital elevation model built into the software. The identified areas have been described in
terms of their layout relative to adjacent steep slopes, watercourses, lochs and other features.
Approximate distances between significant locations along the road have also been given.
The sites identified (in the order in which they were suggested at the Workshop, and not in
any order of perceived hazard or hazard ranking) are set out in the following section.
7.1
AREAS OF HIGH PERCEIVED HAZARD
A83 Ardgarten to Loch Shira (29km)
This includes the stretch of road between the Rest and be Thankful and Cairndow and is
characterised by the steep slopes above and below it, to the north-east and south-west
respectively.
From Ardgarten (NN 27500 03150) the road ascends to the Rest and be Thankful (NN 22960
07450), a distance of around 7km, first in the base of the valley and then, as it gains elevation,
on sidelong ground high above and to the north-east of the valley floor. The road then passes
above Loch Restil and along the valley floor, before running on sidelong ground along the
north side of Glen Kinglas. With the steep slopes of the Binnein an Fhidhleir rising above,
this is where debris flows occurred in August 2004. The road then follows Glen Kinglas
down to its mouth on to Loch Fyne (NN 18420 11300). This latter section of the route is
some 9km long.
By this point the road is within 10m of sea level and close to the Loch Fyne shore, with steep
slopes still rising above. At NN 19400 12500 the road turns sharply onto the low-lying
ground around the head of Loch Fyne for a distance of less than 1km. The shore line is then
rejoined (NN 18700 12600) and the route continues, generally no more than 20m above sea
level, on sidelong ground with steep slopes above, until Loch Shira - an area of known
landslide potential. This lochside stretch of the road covers approximately 13km.
90
HIGH HAZARD AREAS AND EARLY OPPORTUNITIES
A84 South of Strathyre (8km)
Heading south from Strathyre, the A84 follows the valley floor to the head of Loch Lubnaig.
The road then follows a course just a few metres away from the eastern margin of the loch
(NN 56350 15300) for 7km with the steep slopes of Beinn Each and associated hills above.
At the end of the loch (NN 58600 10700) the slopes slacken near St Bride’s Chapel before
steepening again through the Pass of Leny (NN 58790 09160 to NN 60200 08650) for a
further 1km. This is where the river runs through the Falls of Leny, immediately below the
road.
A85 Glen Ogle (6km)
Running north-west, the A85 leaves Lochearnhead (NN 58870 23820) and follows Glen Ogle
for a distance of just under 6km. (NN 55830 28400) Along this length the road runs on
sidelong ground alongside the river. Running first to the west of the river and then to the
north-east, the road runs up to 40 m above the river in some places. The August 2004 debris
flows occurred in this section some 3km out of Lochearnhead.
Some debris flow activity has been observed to the north of Strathyre (18 August 2004)
between the two preceding sections. It may be prudent to include this section in any early
evaluations.
A87 Glenshiel (18km, plus a possible further 17km)
The A87 in Glenshiel is characterised by the steep mountain slopes on either side of the route.
Running south from Kintail Lodge (NG 94450 20230) the road runs initially along the
northerly shore of Loch Duich, turning into the mouth of Glenshiel (NG 93300 19100) after
1.5km. From this point the road crosses Shiel Bridge and then runs first along the south-west
side of the small Loch Shiel and then the River Shiel, in Glen Shiel itself, as far as the
Glenshiel battle site (NG 99150 13170). This section of around 7.5km runs mainly on the
valley floor, but occasionally ascends to around 30m above the river. At the battle site the
road once more crosses the river and then follows the north side of the glen as it turns to run
eastwards. Through this approximately 8.5km stretch the road is mainly within a few metres
of the valley floor as far as the pass that separates Glenshiel from Loch Cluanie (NH 03890
11700).
There may well be a case for adding the further 8.5km stretch of the A87 alongside Loch
Cluanie (NH 09200 12060 to NH 16320 10700) to the list of high hazard areas. This length of
road is backed by steep slopes to the north and runs close to and just above the loch
throughout much of this stretch. A similar case could also be made for the 8.5km stretch of
the A87 that runs alongside Loch Duich from just south of Eilean Donan (NG 88500 25550)
to south east of Inverinate (NG 94400 21150) near Morvich. Although, relatively little drift
material is present in parts of this area close to the road, debris flows have been known in this
area with source material from high level slopes. It is however recognised that additional
assessments may be required using ROSHI (see Section 6.3.3).
A82 Fort Augustus to Lochend (29km, plus a possible further 9km)
As the A82 runs north out of Fort Augustus (NH 38250 10200) it follows a route close to the
westerly side of Loch Ness, rarely rising more than 20 m above the loch until it turns inland
91
HIGH HAZARD AREAS AND EARLY OPPORTUNITIES
towards the bridge over the river at Invermoriston (NH 41950 16500). This section is around
8.5km long. From Invermoriston, (NH 42050 16800) the road returns close to the lochside
and runs parallel to it for some distance, again rarely rising more that 20m above the level of
the loch itself. At Achnahannet (NH 51100 25750) the road then begins to follow a route
further from and higher above the loch before turning inland at Urquhart Castle for
Drumnadrochit (NH 52880 28450) some 17km from Invermoriston. After Drumnadrochit
(NH 52650 30000), the road then rejoins the lochside, remaining close to it as far as Lochend
(NH 59650 37950). This section is some 13.5km long.
The section from Drumnadrochit to Lochend has been the subject of recent inspections by the
Operating Company. These are believed to have revealed only limited drift deposits close to
the road and this section may be more suitable for assessment using ROSHI (see Section
6.3.3). However, it is not clear how far up the slopes of the adjacent hillsides the inspections
reached and therefore how relevamnt they are to debris flow hazard. Similar inspections are
planned for the section south of the Drumnadrochit in 2005.
While there are noticeable variations in the steepness, extent and ground cover of the hills
bounding this section, these generally rise sharply above the road right along the lengths
detailed. For this reason, there is insufficient information on the basis of a simple map-based
survey to rule out any of the sections from being of high hazard. Indeed there is a strong
argument for including the 9km length of road alongside Loch Lochy which runs down from
Laggan Locks (NN 28750 96150) to Letterfinlay (NN 24750 90800). Debris flows were
experience in this area in early 2005 (see Section 2.2).
A835 Ullapool to Braemore Junction (16km)
To the south of Ullapool (NH 15100 92050) the A835 follows a line that is frequently close to
and just above the shore of Loch Broom and then latterly the River Broom all the way up the
Corrieshalloch Gorge to Braemore Junction (NH 20920 77720), a distance of some 16km.
Steep slopes are in evidence above this entire length of road.
A9 Dunkeld to Drumochter (22km)
From a point just north of Dunkeld, where the road crosses the River Tay (NO 00450 43900),
the A9 runs for approximately 3km (NO 00200 47150) with the river close by below and the
old A9 on the steep slopes above. To the north, the slopes above the road slacken and the
hazard level diminishes. It is not until just to north of Pitlochry that the slopes steepen again
as the road enters the Pass of Killicrankie (NO 91620 60750). After 2km the end of the pass is
reached (NN 91920 62350) and the slopes slacken to lessen the hazard levels once more. In
this area the slopes beneath the A9 are also steep and lead down to the old A9, the railway
and the River Tay. The slopes above the road steepen once more at Shierglas (NN 88480
64320) and do not slacken for around 3.5km, beyond Blair Atholl, until Balnansteuartach is
reached. At the Pass of Drumochter the slopes above the road steepen once more at The Wade
Stone (NN 69142 71730) only slackening in steepness 13km later in the locality of North
Drumochter Lodge (NN 63000 79700).
A95 Craigellachie (1km)
The hills in this area are significantly less steep and less high than in other areas identified in
this listing. However, there may be a relatively high hazard level locally where the A95
92
HIGH HAZARD AREAS AND EARLY OPPORTUNITIES
between Maggieknockater and Craigellachie passes over a hill (NJ 30150 44750) and then
sweeps downwards on sidelong ground through Birchbank downwards to the Spey at
Craigellachie itself (NJ 29380 45150).
A86 Spean Bridge (5.5km)
Debris flows are known to have occurred in the area of the National Trust for Scotland site at
Achaneich/Inverroy (NN 24600 81600) around 1999/2000. This area is characterised by
relatively shallow slopes (for the area), but a high density of streams which could carry debris
flow. Above the spring line the slopes steepen significantly. In addition the 3km length of
road to the east of Spean Bridge exhibits particularly steep slopes (between a point just to the
east of Roybridge, NN 27970 80900, and a second point to the east of Achluachrach, NN
31000 81200). It is suggested that the entire section between NN 23100 82000 to the east of
Spean Bridge and NN 31000 81200 to the east of Achluachrach be considered, with the
exception of the short stretch on the flood plain at Roybridge.
A87 (Skye) Gleann Torra-mhichaig to South of Raasay ferry (1.5km)
This section commences about 1.5km north of the Sligachan Hotel (NG 49650 30550),
running generally southwards. After passing the junction with the A863, the A87 then runs
north-east, skirting the base of the very steep slopes of Glamaig round past Sconser, and then
heads southwards into Gleann Torra-mhichaig terminating where the road crosses the river
Abhainn Torra-mhichaig (NG 53750 30700). For the initial part of the section in question the
road runs just above the shore line, thereafter entering the glen, where it runs above the river.
7.2
EARLY OPPORTUNITIES
After availability of the GIS assessment data (see Section 6.1) during Study 1, Part 2, a
comparison will be made with the sections of road identified above. Such an exercise will
enable a selection of different types of potential failure to be used in the evaluation and
validation of the system for hazard ranking which is to be developed as a key objective of
Part 2.
In addition to identifying the sites as listed above, the Project Workshop also gave an ideal
opportunity to consider actions which could be carried out in the short term, either to
minimise the build-up of potential factors which might give rise to unstable slope situations
on the network, or to improve systems to collect and use meaningful data which might assist
in the assessment or prediction of slope failure events in the future.
In the realm of minimising potential contributory factors, some retargeting of maintenance
actions could be productive. Checking of gullies, ditches and catchpits, with a wider view to
that of merely keeping the roadway itself clear of water, could be undertaken as part of
regular inspections. Where ineffectiveness of the system, or underperformance under updated
drainage criteria, is suspected, this should be considered in conjunction with the inspection
regime for the roadside side slopes and remedial action addressed via an appropriate
structured asset management plan. The principles of such a management approach are set out
in HD 41/03 (DMRB 4.1.3). Additionally, critical review of the alignment of culverts and
other conduits close to the road ought to be carried out as part of inspection and reporting
procedures.
93
HIGH HAZARD AREAS AND EARLY OPPORTUNITIES
Certain monitoring measures are already under consideration – for example, the installation
of a rain gauge close to the A85 – but the use of any such data gained, in conjunction with
longer-duration data available from the Meteorological Office, needs to be managed
appropriately to serve a worthwhile and consistent function. At a later stage, informed
selection of locations for discrete placement of additional rain-gauging facilities could be
productive, and should be considered in the light of experience of managing the information
from current sources.
An important action which could be introduced on an early basis is bringing NADICS,
including both the current and proposed future network of variable message signs, into the
management loop with regard to route advice when weather conditions conspire to create
situations where sections of the network might be considered ‘at-risk’.
94
8
DEBRIS FLOW MANAGEMENT AND MITIGATION OPTIONS
by A Sloan, L Shackman, F Macgregor and M G Winter
8.1
MANAGING THE ASSET
The trunk road network comprises a long linear asset, much of which passes through hilly
terrain, with varying potential for the development of debris flows and other disruptive events.
The roads themselves carry different traffic flows and therefore the resulting consequences or
losses from such events are variable.
This section explores the various management and mitigation practices that have been
adopted in both the UK and overseas. It then goes on to recommend some of these as
potential techniques for use on the Scottish trunk road network. It is suggested that a threestep management tool in Study 1, Part 2 be adopted, as follows:
• Detection: The identification of the occurrence of an event or the precursor conditions that
could lead to an event.
• Notification: The dissemination of information relating to the hazard(s).
• Action: The proactive process by which intervention reduces the exposure of the road user
to the hazard.
This Detection-Notification-Action (or DNA) process has been developed as a tangible
approach to reducing the hazards to which the public are exposed when using the trunk road
network. It is feasible to introduce this on parts of the network, particularly at locations where
the hazards posed by debris flows are recognised as being real and present. It is considered
that the DNA process should be an intrinsic element of a fully developed Asset Management
System.
8.2
APPROACHES TO LANDSLIDE MANAGEMENT
In developing management processes for problems in Scotland it would be prudent to learn
lessons from countries where landslide management has been practised for some time and
where losses arising from landslides are significant. It is to be noted that in an international
context the scale, frequency and consequential losses from landslides in Scotland have
fortunately been relatively minor to date. The following sections briefly examine procedures
developed for use in the United States of America (USA) and Hong Kong although it is worth
noting that substantial work has been carried out in Australia (AGS, 2000) and elsewhere in
the world.
8.2.1
United States of America
Landslides in the USA, for example, collectively constitute a serious hazard and result in
significant losses. The US Geological Survey (USGS) estimates an average of 25 to 50 deaths
annually are attributable to landslides as well as financial costs of between $1 billion and $3
billion per annum and untold environmental and societal disruption. To this USGS has
developed a management system (Spiker and Gori, 2003) based upon the following elements:
• Research.
95
MANAGEMENT AND MITIGATION OPTIONS
• Hazard mapping, the benefits of which are particularly emphasised by Anon (2004).
• Real time monitoring.
• Loss assessment.
• Information collection, interpretation, dissemination and archiving.
• Guidelines and training.
• Public awareness and education.
• Implementation of loss reduction measures.
• Emergency preparedness, response and recovery.
Amongst other conclusions drawn from an assessment of the USGS proposals (Anon, 2004) it
was also concluded that:
• The dissemination of collected information on landslide hazards was of critical importance
in the implementation of an effective risk reduction programme.
• Finally it was concluded that the wish to implement a loss reduction programme would
cost money and that a budget should be set for both the development and the
implementation of such a process.
8.2.2
Hong Kong SAR
The benefits of introducing a slope management system in the developed world are perhaps
most dramatically observed from the experience of Hong Kong. In 1972 and 1976 landslides
occurred that killed 100 and 18 people respectively. In response to this the precursor of the
Geotechnical Engineering Office (GEO) was established with the basic mandate to improve
public safety.
The GEO has developed slope management systems that are rigorously adhered to throughout
Hong Kong. These involve defining maintenance requirements, examinations, risk analyses,
real-time warning systems in relation to intense rainfall, increased community awareness
through education programmes, and direct engineering works. This process has reduced
dramatically the risk to public safety despite the rapid growth of Hong Kong (Chan, 2000).
Early stages of the Hong Kong programme concentrated on hazard definition through the
developments of asset inventories, mapping and geological/geotechnical assessments. It was
realised that such a technical based approach was insufficient alone to reduce the risk posed
by landslides. Consequently developmental work was carried out into inspection and
maintenance regimes, risk analysis, warning systems in relation to heavy rainfall, education
of the community and increasing public awareness to the presence of slope hazards. This
included the setting of regulatory instruments to control the construction of new slopes and
the remediation of existing slopes17.
The foregoing is simply a small sample of the wide range of systems implemented around the
world for slope management. The examples do however serve as useful reference systems,
albeit covering geographical areas more prone to landslides than Scotland.
17
The processes involved in the day-to-day management of slopes in Hong Kong can be studied through the
following web link; http://hkss.ced.gov.hk/hkss/eng/whatsnew/index.htm.
96
MANAGEMENT AND MITIGATION OPTIONS
8.2.3
United Kingdom
In the UK other managers of long linear infrastructure are Network Rail and the Highways
Agency in England. Both these organisations have asset management systems specifically
dealing with slopes. However, the Highways Agency in particular only really deals with
slopes in the near-field relative to their road infrastructure. It does not, in general, have to
contend with slopes that extend some hundreds of metres vertically and many more metres
horizontally from their infrastructure as does the trunk road authority in Scotland.
Of particular relevance for comparison purposes is the system employed on Scotland’s
railways. This is of relevance because the railway in Scotland is faced with similar problems
to those experienced by the country’s trunk road system. The railways in Scotland run in the
same topography and quite often along the same glens as the trunk road system. The rail
network in Scotland has suffered from some significant landslides in the past. Partly as a
consequence of these failures Network Rail has a system whereby all of the earthworks
(embankments and cuttings) on the network are examined on a regular basis as a risk
management procedure. The process is summarised in Figure 8.1.
It is important to note that the primary response of Network Rail to a perceived heightening
of risk is often to close the railway and use buses to transport their customers by public road.
This is clearly not an option for the trunk road authority.
Inventory of Al Slopes
Examination and Assessment
of Potential for Failure at
Each Slope
Engineered Remedial Works
Monitoring
Categorisation of Slopes as
Poor
Marginal
Serviceable
Re-Examination of All
Slopes on a Cyclical Basis
Reduce Risk to Acceptable
Levels
Prioritisation of Poor Slopes
in Order of Perceived Risk
and Develop Business Case
Poor Slopes: Every Year
Is the Risk to the Safe
Operation of the Railway at
Each Slope Acceptable?
Marginal: Every 5 years
Seviceable: Every 10 years
Figure 8.1 – Summary of the Network Rail slope management procedures.
The elements of the process within the red outline on the above diagram are those that are
specified in Network Rail’s published standards. The management system revolves around
the visual examination of slopes on a cyclic basis after an initial assessment. Those posing the
greatest hazard are examined every year whilst those that are more benign are examined
every 10 years. Those slopes of marginal hazard rating are scheduled for re-examination
97
MANAGEMENT AND MITIGATION OPTIONS
every five years. The processes by which the slopes have to be examined are specified in
Network Rail company standards along with the level of competence required of the
examiner. Physical monitoring or engineering works are carried out on slopes where the risk
is considered to be such that visual inspection alone cannot be relied upon to manage the risk.
It is clear that management systems developed in relation to the potential for landslides
recognise that it is not possible to prevent such events from occurring at every location. The
aim is to manage the situation by understanding the hazards and potential losses arising from
landslides and reducing losses to acceptable levels.
Key factors which define a system to be developed for dealing with debris flow hazards can
be summarised as follows:
• The relatively small scale of the problem in Scotland.
• The emphasis on the trunk road network.
• Particular topographical and climatic conditions in Scotland.
• The particular emphasis on debris flows on Scotland’s mountains.
Reporting Landslide Events on the UK Rail Network
The initial report of an incident on the railway, including a landslide, can come from a train
driver, a member of the public or a railway worker. The nature of the incident and its location
is reported to Network Rail ‘Control’, a 24 hour a day control centre for all aspects of the
entire network. All rail workers are briefed as to the telephone number for ‘Control’ and it is
displayed prominently on rail structures such as bridges.
Upon receipt of a notification of a landslide, ‘Control’ will notify the on-call engineer from
the engineering department with responsibility for earthworks. This engineer makes a
decision as to the best course of action. This, more often than not is to call Network Rail’s
Earthworks Examination engineers. This is a firm of independent consultants appointed to
undertake the cyclic examinations of the network. This firm also provides 24 hour cover for
call-outs to landslide events. The purpose of the call-out is primarily to undertake an
examination and record the nature of the landslide.
Engineers on site assess the magnitude of the problem. Management decisions are made
based upon the evidence collected from the site. The line may be already closed, or open but
in a dangerous condition. In either case emphasis is placed on re-opening the line for full
speed train operation in as expedient manner as possible, whether engineering works are
required prior to re-opening or not.
8.3
ASSET MANAGEMENT FOR TRUNK ROAD SLOPES
8.3.1 The Concept of Loss
Asset Management requires knowledge of the degree of risk associated with the nature of the
hazard, the likelihood of debris flows developing and knowledge of the consequences if the
debris flow occurred. A set of criteria developed to understand and recognise hazards,
although it is also recognised that it is highly unlikely that a stage of predicting debris flows
or any other landslides with any degree of certainty will be reached.
98
MANAGEMENT AND MITIGATION OPTIONS
Defining the consequences of such an event to allow a meaningful and complete assessment
of risk is extremely complex. The identification and quantification of loss associated with
landslides is as important an element of risk assessment as hazard identification. Take, for
example, the debris flows that occurred at the same time and very close to those that closed
the A83 in August 2004 but did not reach the road. While the hazard was essentially the same,
within discernible limits, the losses were significantly different. This fact has been recognised
in Section 6 and the term hazard ranking used as shorthand for an assessment that is
qualitative or semi-quantitative.
Accepting that it will not be possible to reduce the risk to zero in each and every case
introduces that realisation that there must be a level of loss that is acceptable and that to
define this one must identify a boundary between acceptable loss and unacceptable loss.
This concept is useful in determining risk as it gets away from the need to accurately define
loss, instead the losses can be rated as either acceptable or unacceptable. For example it may
transpire that on certain routes it is unacceptable to have a road closure at all or for no more
than a short period of time. In other locations the level of acceptable loss may be drawn at
personal injury (i.e. road closure is acceptable for a period of days but injury to the road user
is unacceptable).
Care should be taken not to set the aspirations of loss reduction to a level that cannot be
justified in financial terms due to either the magnitude of the problem and/or inherent
technical complexity. Stage 2 of this study will need to broadly define acceptable and
unacceptable losses in the context of the exposure assessment discussed in Section 6.3.
8.3.2 Asset Management Strategies
The options available for the development of an asset management strategy fall into two
groups defined herein as follows:
• Category 1: Reactive Approach.
• Category 2: Proactive Approach.
The Category 1 approach accepts that debris flows will occur and seeks to formalise the
response to such events. This would involve little or no change to the current arrangements
whereby the Operating Company is mobilised in the event of a landslide occurring on the
trunk road network. Technical and practical assessments are made at the scene and actions
such as diversion and road re-opening decided upon and implemented appropriately. With the
absence of a debris flow hazard assessment model, this type of approach is one which
requires to be adopted where such types of event can potentially occur.
A reactive approach may be appropriate where small non-life threatening slips are expected.
In the terminology defined above this equates to situations where the level of consequential
losses are small. The magnitude of the debris flows experienced in the summer of 2004 were
such that the reactive approach is considered to be inappropriate in that the consequential
losses could so easily have been significantly greater.
To this end it is considered that the better approach is that of the proactive Category 2 type.
This can be summarised as the development and implementation of a step-by-step approach
to managing the risk posed by debris flows. This would involve identifying risk and reducing
99
MANAGEMENT AND MITIGATION OPTIONS
this where the levels were determined as being unacceptably high, the aim being to minimise
losses due to debris flows.
The Category 2 approach necessarily comprises several stages to be worked through in the
development and implementation of management procedures to minimise potential losses.
The complexity of each stage and the magnitude of the effort required will be dependent on
the magnitude of the problem as a whole and the aspirations of the Scottish Executive.
The steps that require to be addressed are summarised as follows:
• Develop an inventory of slopes likely to pose a hazard to the road network.
• Carry out hazard assessment of each slope.
• Debris flow mapping.
• Assess the acceptable loss at each site.
• Assess the hazard posed at each site.
• Assess whether this hazard is acceptable or not.
• In cases where hazard is unacceptable decide on hazard reduction measures.
• In cases where hazard is acceptable decide how to measure change in hazard.
• Education and knowledge dissemination.
The options available to deliver each stage are discussed in more detail below. The purpose of
the discussion is not to present technical detail rather to discuss the options from a
management perspective and to reinforce the manner in which these elements of a system as
largely described in previous sections fits into an asset management strategy.
The development of an inventory of slopes does not lend itself well to natural slopes, not least
due to the likely very high number of low risk slopes that would be categorised. More
appropriate is to ensure that areas likely to pose a hazard to the road network zones are
delineated. The determination of such areas is described in Section 6 along with the approach
to hazard assessment and the development of maps of debris flow hazards. The results of
these will need to be taken forward into a loss, hazard ranking and loss acceptability
assessment.
8.3.3 Assessment of Loss, Risk and its Acceptability
Without an assessment of loss it will not be possible to assess risk in a fully quantitative
manner. However, as discussed in Section 6 it may be that the potential for loss can be
summarised on a route by route or part route basis and a semi-quantitative/qualitative
assessment may not only be acceptable but potentially desirable. In managing the asset, as
discussed below, it will be necessary to compare the risk at individual slopes to determine
which ones should be considered in more detail for risk mitigation measures. If losses along
sections of road are deemed to be the same, then the comparison of risk distils to a
comparison of geotechnical hazard.
The concepts of acceptable loss and unacceptable loss are discussed in forgoing text. An
action for the next stage of the study is to ascertain on a route by route basis the degree of
100
MANAGEMENT AND MITIGATION OPTIONS
acceptable loss and thereafter acceptable risk. As a minimum this will be set at a level less
than personal injury to the road user. It is to be noted that the commissioning of this study
indicates that the degree of loss suffered in the summer of 2004 was unacceptable. In these
events no one was injured but several road users had fortunate escapes and it may be
considered that it is the realisation of what could have happened that is unacceptable.
The key test of the risk assessment process as an element of the overall management of the
network is to assess whether the risk determined is acceptable or unacceptable. This single
element is probably the most critical aspect of the asset management process. Up until this
point the process has been steered by determining risk. In comparing slopes and deciding
whether or not to recommend risk reduction measures a degree of objectivity and experience
is required. It requires consideration of not only the level of risk but other factors as well, not
least amongst these being financial constraints.
In the case where the risk is considered to be acceptable the option exists to do no further
work. However there is always the possibility of unknown factors in the condition of a slope
that could introduce an unsuspected hazard. In addition there is always the inherent, and in
the case of debris flow assessment the very real, possibility of uncertainty in assessing the
true extent of the hazard and, indeed, the risk. Consequently it is likely that some form of
repeat examination will be required after the initial categorisation of hazard and risk. The next
stage of the study will address the need for repeat examination and reassessment of slopes.
Further to this it may be the case that risk reduction measures cannot be put in place at all the
slopes identified as posing significant risk. In this eventuality it may be that repeat
examination is required in lieu of risk reduction measures.
8.3.4 Risk Reduction Measures
In the case of slopes that are categorised as being of sufficiently high risk that the situation
cannot be managed by repeat examinations it will be necessary to take action to reduce the
risk or, in this case, hazard ranking. Risk, being the product of hazard and consequence, can
be managed by tackling either or both of these two elements. Examples of these are discussed
in detail in Section 8.4, albeit in terms of the hazard, exposure and hazard ranking approach
described in Section 6.
8.4
MITIGATION TECHNIQUES
The foregoing details the processes of managing slopes to understand the potential for debris
flows. The process culminates in a decision on whether the hazard ranking, in the context of
the safe operation of the road network at any location, is acceptable or not. At those locations
where the hazard ranking is unacceptable it will be necessary to undertake some form of
mitigative action to either reduce the hazard or to reduce the exposure of the road user.
To reduce the hazard to the road user either the magnitude of the hazard and/or the potential
exposure or losses that are likely to arise as a result of debris flow must be reduced. To reduce
the exposure of road users, the debris flow event is taken as a given and either the number of
people exposed to the hazard must be reduced, for example by closure of the road, or they
must be warned to exercise caution at appropriate times and places.
101
MANAGEMENT AND MITIGATION OPTIONS
8.4.1 Exposure Reduction
The reduction of exposure lends itself to the use of a simple and memorable three-part
management tool, as follows:
• Detection: The identification of either the occurrence of an event, by instrumentation (e.g.
tilt meters or acoustic sensors) or observation (e.g. Closed-Circuit Television, CCTV,
monitoring or visual patrols during high likelihood periods), or by the measurement and/or
forecast of precursor conditions (e.g. rainfall).
• Notification: The dissemination of information relating to the hazard(s) and exposure(s),
by for example Variable Message Signs (VMS) including NADICS signs, media
announcements (radio, TV, traffic guidance systems and the web) and “landslide patrols”
in marked vehicles.
• Action: The proactive process by which intervention reduces the exposure of the road user
to the hazard, by for example road closure, convoying of traffic18 or traffic diversion.
This DNA approach can be operated for either precursor conditions that potentially lead to
landslide events in high hazard ranking areas, namely rainfall, or to actual landslide events
that have taken place.
Precursor (Preparatory or Trigger) Conditions
Detection: Debris flows are initiated, in the main, by heavy rainfall in combination with other
conditions. Forecast and real time rainfall data for an area with adverse topographic or other
conditions is extremely useful information. If high rainfall is forecast or recorded in such
areas then the potential for debris flows will be higher. In certain parts of the world weather
forecasting and thereafter rainfall monitoring in real time are two of the controlling factors in
landslide management. For example the very successful system run in Hong Kong and that
trialled in California both pass information on the heightened likelihood of landslide
development to the public as a result of rainfall monitoring. This is achieved in a not
dissimilar fashion as that in which information on extreme weather events is passed to the
public in the UK.
In the case of Hong Kong19 a comprehensive network of automatic rain gauges covers much
of the region to record and send data to a central control point for analysis in real time. This is
combined with short-term forecast data to enable managers to monitor the rainfall situation as
it develops and make informed decisions in an expedient fashion.
If such predictive capability was installed in Scotland then it would be possible to develop
systems to reduce the exposure of the road user to the effects of debris flows. However, it
must be understood that in Hong Kong more than 20 years of experience have been acquired.
This means that a sound knowledge of the relationship between rainfall and landslides is in
place relating to the local climate and geology. It is clear that some considerable time would
be required to build a similar knowledge base for Scotland, possibly a minimum of five years.
A significant investment in instrumentation, data analysis and maintenance would however be
required.
18
Note that while this moves traffic past a potential hazard rapidly if a convoy is hit the losses would be greater
than might otherwise be the case.
19
Source: http://hkss.ced.gov.hk/hkss/eng/index.htm.
102
MANAGEMENT AND MITIGATION OPTIONS
Notification: In Hong Kong if the conditions for a ‘Landslip Warning’ are met then the public
are alerted to reduce their exposure to possible danger from landslides. The issue of a
Landslip Warning also triggers an emergency system within various Government
Departments that mobilizes staff and resources to deal with landslide incidents. A Landslip
Warning is issued when it is predicted that numerous (more than about ten) landslides will
occur. Nonetheless it is accepted that isolated landslides may occur from time to time when a
Landslip Warning is not in force and that Landslip Warnings will occasionally be issued and
not be followed by landslides. Landslip Warnings are issued by means of website notices,
media announcements and notices prominently displayed in public buildings and areas.
In Scotland it is clearly important that a variety of public announcements are used when there
is a heightened likelihood of landslide development in an area. This might involve a variety
of systems including websites (e.g. NADICS), variable message sign systems and media
(radio and TV) announcements notifying drivers that their potential exposure to the hazards
posed by landslides is real and present. Announcements could also be linked into traffic
guidance systems such as TrafficMasterTM.
Action: If such a system were devised and implemented for Scotland and warnings were
received that heavy rain was falling in an area or was approaching an area recognised as
being of high hazard ranking then a number of options are available for action. First the road
length (or lengths) deemed to be threatened could be closed. This might be effected by
installing barriers such as the snow barriers present on some of Scotland’s roads. The decision
to reopen the road would need to be taken after the intense rainfall had passed and an
inspection of the route had taken place. The dis-benefit to this approach is that given the
relatively rare occurrence of debris flows, at least those that interact with the trunk road
network, and the high levels of rainfall that Scotland receives, a number of false alarms could
be expected. The public at large could, potentially, become disillusioned at what could be
seen as a very conservative approach.
Alternatively trained operatives could be deployed on high hazard ranking sections of road
during periods of predicted or actual high rainfall. These operatives could escort people
through the high hazard ranking sections of road.
An alternative approach could be to simply inform the public of the heightened informed of
the heightened likelihood of landslide development in an area, as described above, and to take
no further action until an event occurred.
Event Occurrence
Detection: The movement of slope material can be monitored and the resulting information
used in a similar way to rainfall data. The data is measured in real time and used as a
management tool. Monitoring instruments can be located such as to record movement from
potential debris flow or positioned such that notification is received if debris reaches or gets
close to a road.
In relation to the former, the seeding area for debris flows can be very large and high on the
hillside. This introduces difficulty in pinpointing the optimum location for the installation of
the monitoring system and doubt as to whether the debris will reach the road. Installing
instrumentation to indicate whether debris has reached a road has precedence: there is a
103
MANAGEMENT AND MITIGATION OPTIONS
location on the Scottish rail network at Glen Douglas where an instrumented fence has been
installed. The purpose of this is to recognise when a fall of ground impinges the line.
Similarly the railway through the Pass of Brander above the A85 at Loch Awe has a system
whereby any rolling rocks or debris flows trigger signals on the railway that shut the line and
stop trains.
It is likely that any instrumentation would be electronic with remote reading of data sent back
to a central control point.
Whether such a system is sufficient in isolation is questionable but it is considered that in
conjunction with rainfall monitoring and possibly the deployment of operatives the likelihood
of road users being affected by debris flow events could be reduced significantly.
A range of possible instrumentation types is presented briefly, as follows:
• Borehole or Shallow Inclinometers: Instrumentation installed in the ground that monitors
ground movement. Of use when movement is know to be occurring and is to be monitored
over a period of time.
• Tilt Meters: Instrumentation installed on a structure to determine rotation of the structure.
The rotation is measured by electronic tilt switches. To be installed in road or hill side
barriers to indicate movement of the ground or impact of a debris flow.
• ‘Trip Wire’: Instrumentation to be installed along the strike of the slope that records
whether debris has moved on the hillside. In this case a cable is physically moved by
debris either as material strikes it or the fixed ends or fixed points of the cable move
relative to one another. Movement can be detected by a change in electrical resistance in
the cable.
• ‘Ball of String’: Generally used to detect movement broadly along the dip of a slope. A
fixed point is placed to stable ground and a freely rotating drum of wire is attached. The
free end of the wire is attached to a point on potentially unstable ground. Movement is
detected by the rotation of the drum as the movement causes additional wire to be paid out.
• Telemetry: Movement of the slope at discrete locations is recorded remotely by measuring
distances form a set point.
• Acoustic Meters: Instrumentation that detects small amounts of noise/vibration caused by
small movements preceding larger landslide events.
• Remote Sensing: Remote assessment of slope instability using techniques ranging from
CCTV monitoring to satellite imagery. Monitoring might initially be by human/visual
means while automatic movement detection systems are developed.
An alternative approach is to use operatives to detect debris flow events by introducing
landslide patrols during periods of high rainfall. As previously noted it is essential that such
operatives are trained in what to look for and that patrols should operate in pairs for safety
reasons.
Notification: In the first instance, a landslide event having occurred, notification must be to
the Operating Company and the infrastructure owner. The decision must then be made rapidly
to close the road or to keep it open. The nature of debris flows is such that in most cases the
104
MANAGEMENT AND MITIGATION OPTIONS
road will be blocked and therefore closed to all intents and purposes. Secondly the public
must be warned by media announcements.
Action: In terms of positive actions that may be taken after a debris flow event the range of
actions is similar to that available for Precursor Conditions and described above. However, it
is important to note that closing the road in the area immediately adjacent to the event is not
an adequate response. Debris flow propensity is generally believed to affect long lengths of
hillside and an evaluation of the vulnerable area must be performed in order to ensure that
and appropriate length of road is closed.
In all cases re-opening of the road must only occur after a thorough inspection of the road and
the adjacent slopes has been undertaken to ensure that the likelihood of further debris flow
events is at an acceptable level. Current practice is to undertake ground-based inspections
only when the adverse weather has abated and only to reopen the road once such inspections
indicate that the residual hazard and exposure are at an acceptable level.
8.4.2 Hazard Reduction
The challenge with hazard reduction is in identifying locations that are of sufficiently high
hazard and exposure to warrant spending significant sums of money on engineering works.
The lengths of road that have already been identified in Section 7 are significant. The costs
associated with installing remedial works over the entirety of such lengths are almost
certainly both unaffordable and unjustifiable. Moreover the environmental impact of such
engineering work should not be underestimated, having a lasting visual impact at the least and
potentially more serious impacts. It is considered that such works should be limited to
locations where their worth can be proven.
Notwithstanding the foregoing, simple measures can be taken such as ensuring that that
channels and gullies are kept open can be effective in terms of hazard reduction. This requires
that the maintenance regime is fully effective both in routine terms and also in response to
periods of high rainfall, flood and slope movement.
Typically, the reduction in hazard will entail physical engineering works to change the nature
of a slope or road to reduce the potential for either initiation and/or the potential for a debris
flow to reach the road once initiated. As described in earlier sections of this report such slides
tend to be dynamic and are quite often initiated some distance above the road. When the
slides reach the road they are relatively fast moving high energy flows. The energy of these
systems has a significant impact in the nature of the engineering works that can be used to
reduce the hazard to the road and its user. Hence, there are three broad approaches to
selection of hazard mitigation works:
• Accept that debris flows will occur and protect the road.
• Carry out engineering works to reduce to opportunity for a debris flow to occur.
• Realign the road.
In relation to the first option there are not many examples of such engineering works in
Scotland or the rest of the UK, but in upland areas of mainland Europe such engineering is
relatively common place. The energy of the debris flow is such that a rigid barrier constructed
to protect the road would have to be designed for very high loads. The problem with a rigid
105
MANAGEMENT AND MITIGATION OPTIONS
barrier is that the debris flow has significant momentum and to bring the slide to a sudden
stop, as is the case with a rigid barrier, requires the dissipation of a lot of energy,
instantaneously imparting very high loads.
Road Protection
Debris Flow Shelters: Stone shelters or ‘avalanche shelters’ are engineered structures that
form canopies over a section of road prone to rock fall or debris flows. These structures are
usually formed from reinforced concrete. There is an example of such a structure on the A890
north-east of Stromeferry in the north-west highlands. This structure straddles both the road
and railway at that location (Figure 8.2).
Figure 8.2 – Stone shelter on A890 northeast of Stromeferry.
In these structures energy is dissipated by placing a depth of granular material on the roof on
which the debris flow lands.
Debris Flow Overshoots: In situations where the energy is anticipated to be very high,
modifications can be made to the above to allow the debris flows to pass over the top of the
structure. This is done by shaping the top of roof of the shelter such that the falling material
passes over the structure without dissipating much energy. This shaping or profiling involves
constructing a ski-jump type reinforced concrete structure. Material falling simply slides over
the roof and continues down the hillside.
Barrier Fences: Fences can be constructed to act as effective barriers to halt debris flows.
Such fences are designed to be flexible so that the energy of the debris flow is dissipated over
a short period of time thus reducing the forces that the structure has to cater for. These
systems have been shown to work well. Figure 8.3 shows such a fence installed on the
Inverness to Kyle of Lochalsh railway in Scotland. Such fences do require maintenance after
the impact of a debris flow. A related approach has been taken to the arrest of rockfalls using
highly flexible fences with fixed end-posts only (e.g. Winter et al., In Preparation).
106
MANAGEMENT AND MITIGATION OPTIONS
Flexible fixed position fence structures are common place in upland areas of mainland Europe
and while the UK does not have engineering design standards for such structures experience
is available and formalised procedures do exist, particularly in Switzerland.
Improving Channel Flow: In certain circumstances it may be possible to improve channel
flow down to the road and beneath the road by for example widening culverts. Such works
would improve the potential for debris flows to avoid the road.
Debris Flow Prevention
In relation to the second option, preventing the slide happening in the first place, applicable
engineering solutions will vary depending very much upon individual circumstances. Debris
flows can have a relatively large source area and in the case of recent examples in Scotland,
be located very high up on the hillside above the road. In most circumstances the potential for
carrying out conventional remedial works to restrain the material before it starts to move is
considered to be very limited. There may be particular conditions where a combination of
techniques such as gravity retaining structures, anchoring or soil nailing may be applicable.
However, in general terms the cases where these are applicable and economic are likely to be
limited.
The link between debris flows and intense rainfall has been established previously in this
document. As a result water management can reduce the potential for debris flow initiation.
In the circumstances of the large debris flows that occurred in the summer of 2004 it is
considered that on hill drainage improvement would have had little impact due to the scale of
the events. In other locations positive action to improve drainage may well have a beneficial
effect. This would include improving channel flow and forming drainage around the crest of
certain slopes to take water away in a controlled manner.
Figure 8.3 – Flexible catch fence.
107
MANAGEMENT AND MITIGATION OPTIONS
Road Realignment
Road realignment can be used as part of the Scottish Executive’s structural management
activities in order to improve the road in terms of both alignment and junction layout, in
particular to reduce accidents and also to ensure compliance with current design standards. In
cases where the hazard ranking from debris flows is high and other factors indicate that some
degree of reconstruction is required, road realignment may be a viable option. Similar
expedients have historically been used on the Scottish rail network, for instance at
Stromeferry, Penmanshiel and Dolphinston, where hazards have been sufficiently significant
to justify the high cost of such realignments.
8.4.3 Partnership, Education and Knowledge Dissemination
It is recognised (USGS and GEO: Malone, 1998) that widespread public awareness of
landslide hazards enables individuals to make informed decisions as to where to live what
property to buy where to locate businesses and so on. For local decision makers such
knowledge allows for better town planning and the locating of critical facilities. In this study
the decisions that people would have to make are limited to road usage. However, it is
considered that a more inclusive approach will result in a wider understanding of losses, for
example road closures. In parallel, the dissemination of such knowledge widens the base of
decision making responsibility and as such public acceptance of loss is likely to increase.
To this end it is suggested that the next stage of the study addresses, the most appropriate
ways to increase public awareness, evaluate the effectiveness of different types of message
and messaging systems, and the most appropriate methods to disseminate information.
It should be recognised that different users will look for differing levels of information. This
could range from roadside warnings in its simplest form to more detailed data presentations,
possibly web site based where information detailing the management procedures is posted.
It is not only the public that should be engaged in debris flow education, but road maintainers
and local authorities as well. In a similar way any alterations road layouts or new road
developments should have the assessment of the potential impact on debris flow and land
slips in general as an integral requirement of the Approval in Principle process.
108
9
SUMMARY AND RECOMMENDATIONS FOR DEBRIS FLOWS IN
SCOTLAND
by M G Winter, F Macgregor and L Shackman
9.1
SUMMARY
In August 2004 a series of landslides in the form of debris flows occurred in Scotland. Some
of these affected the A83, A9 and A85, which form part of the trunk road network. These
incidents were well reported in the media.
While debris flows occur with some frequency in Scotland, they only rarely affect the trunk
road network or for that matter the main local road network. However, when they do impact
on the road network the degree of damage they do, in terms of the infrastructure and the loss
of utility to road users, can have a major detrimental effect on both economic and social
aspects of the use of the asset. Additionally, there is a high potential for such events to cause
serious injury and even loss of life although, fortuitously, such consequences have been
limited to date.
The events of August 2004 followed a sustained period of heavy rainfall and, in addition,
intense localised storms contributed to the triggering of at least some of the resulting debris
flows. Rainfall of up to 300% of the monthly average fell in certain parts of Scotland during
August 2004.
Within the recent past, debris flow activity in Scotland has occurred largely in the periods
July to August and November to January, but there is no certainty that such a pattern will be
continued in the future. However, eastern parts of Scotland do receive their highest levels of
rainfall in August. Additionally, climate change models indicate that rainfall levels will
increase in the winter but decrease during the summer months but that intense storm events
will increase in number. These factors, therefore, may change both the frequency and the
annual pattern of debris flow events.
The impacts of such events, when they do happen, are particularly serious during the summer
months due to the major contribution that tourism makes to Scotland’s economy.
Nevertheless, the impacts of debris flow events during the winter months should not be
underestimated.
Following the events of August 2004, this study was commissioned to take stock of the
present situation on the trunk road network and to determine a sustainable approach to the
management to such occurrences in the future. The process chosen involves the assessment
and ranking of hazards with a system of management and mitigation also being proposed.
This system is based upon the principles of Detection, Notification and Action (DNA) applied
both to the response to landslide events and to precursor rainfall conditions.
109
SUMMARY AND RECOMMENDATIONS
9.2
9.2.1
RECOMMENDATIONS
Early Opportunities
A number of areas of perceived high hazard were identified at the Project Workshop. The
lengths of the road and the slope lengths they involve are substantial. Accordingly, it is
considered unrealistic to undertake suitably prioritised further evaluations at this stage. The
proposal is for the outputs of the GIS-based assessment to be used to corroborate the
identification of the localities identified at the Project Workshop and, in addition, as a
validation tool for the site specific assessment methodology.
In the meantime it is important that maintenance and construction projects currently in design
take the opportunity to limit any hazards or exposure by incorporating, where suitable,
measures such as higher capacity or better forms of drainage, or debris traps. Peer group
consultation in the form of the involvement of the Overseeing Organisation and its
Independent Geotechnical Checker, the corresponding specialists within the Operating
Companies, design organisations or other appropriate organisations is an essential part of this
process.
In the realm of minimising the potential impacts of debris flows on the network, some
retargeting of maintenance actions could be productive. The checking of gullies, ditches and
catchpits, with a wider view than that of merely keeping the roadway itself clear of water,
could be undertaken as part of regular inspections. Where ineffectiveness of the drainage
system, or underperformance under updated drainage criteria, is suspected, this should be
considered in conjunction with the inspection regime for the roadside side slopes and
remedial action addressed via an appropriate structured asset management plan. Additionally,
critical review of the alignment of culverts and other conduits close to the road ought to be
carried out as part of inspection and reporting procedures.
Certain monitoring measures are already under consideration – for example, the installation
of a rain gauge in the A83 Rest and be Thankful area, where debris flows are generally small
but relatively frequent, potentially yielding more comparable data in a short time frame. The
use of any such data gained, in conjunction with longer-duration data available from the
Meteorological Office, needs to be managed appropriately to serve a worthwhile and
consistent function. At a later stage, informed selection of locations for discrete placement of
additional rain-gauging facilities could be productive, and should be considered in the light of
experience of managing the information from current sources.
An important action which could be introduced on an early basis is bringing NADICS into the
management loop with regard to route advice when weather conditions conspire to create
situations where sections of the network might be considered ‘at-risk’.
9.2.1 Study 1, Part 2
The initial stage of Study 1, Part 2 will be to develop the methodology for the assessment of
hazard and exposure to provide a hazard ranking, together with the selection of an appropriate
management approach. The second stage will be to test the methodology before applying it
more widely to the trunk road network.
Figure 9.1 presents a flowchart of the work to be undertaken.
110
SUMMARY AND RECOMMENDATIONS
The initial stage of this work is itself divided into four elements and can be summarised as
follows:
• Development of a debris flow hazard and exposure assessment system to provide a hazard
ranking of ‘at-risk’ areas of the road network.
• Undertaking a computer-based GIS assessment as a first stage in the hazard assessment
process.
• Undertaking site specific hazard and exposure assessments of areas identified by the GIS
as being of higher hazard.
• The identification and development of appropriate management processes for each
category of hazard ranking.
Study 1, Part 2
Comparison with
Sites of Known
High Hazard
(see Section 7)
GIS Hazard
Evaluation
(see Section 6)
Methodology
Validation via
Sites of Known
High Hazard
(see Section 7)
Development and Implementation of
Site Specific Hazard Ranking
Evaluation Methodology
(see Section 6)
Elimination of Sites
with Extremely Low
Hazard
Identification of Very
High, High, Medium and
Low Hazard Ranking Sites
LOW
MEDIUM
HIGH
VERY HIGH
Do-Nothing
No Action
Reactive to Events
Do-Minimum
Assess Diversions
and Implement Plans
Do-Something 1
Exposure Reduction
(see Section 8)
Do-Something 2
Hazard Reduction
(see Section 8)
Monitoring and
Feedback
Figure 9.1 – Management and mitigation options within Study 1, Part 2.
The GIS-based assessment will be used as a first stage in the hazard assessment process.
This will enable site specific assessments to be targeted in order to obtain better value from
such relatively resource-intensive activities. It will also allow the elimination of large areas of
the network having minimal hazard.
111
SUMMARY AND RECOMMENDATIONS
It is also particularly important to note that the site-specific assessment will not be a ‘driveby’ survey; it will require a highly specialised detailed site examination which will need to be
carried out using an overall consistent approach. Prior to undertaking any site surveys it is
important that the system is established for consistently describing and identifying hazards
and the associated exposure. Some of the factors that will need to be incorporated in such a
system, such as slope angle and the broad nature of the geology, will be incorporated into the
GIS assessment. Other, more detailed, factors such as the effects of forestation will need to be
incorporated into the site-based survey. Once a hazard assessment has been completed it may
be combined with an assessment of the exposure of the road user to that hazard to give a
hazard ranking. This will allow, in-turn, an appropriate management option to be selected
from the range of options to be developed.
There are a number of potential options which could be applied to the management of debris
flows. These are addressed in the following paragraphs.
The ‘Do-Nothing’ approach is intended to be applied to sites of low hazard ranking for which
substantial expenditure is inappropriate. For such sites, whilst it is not possible to eliminate
the chance of a landslide event affecting such areas it is seen as unlikely, largely
unforeseeable and/or the exposure is less serious than at other locations where resources may
be better expended.
The ‘Do-Minimum’ option, with the potential to mitigate the impacts of debris flows to some
extent involves simply ensuring that forward plans are in place to ensure that diversion routes
are available and may be exploited in an expedient and well organised manner. Diversion
route maps and contingency plans are currently held for many areas of the trunk road network.
Whilst it is not possible to eliminate the chance of a debris flow event affecting such areas
any occurrence is seen as unlikely and largely unforeseeable and any residual exposure
cannot readily be quantified and is unlikely to justify the commitment of additional resources
which may be better expended at other locations.
‘Do-Something 1’ is the first management option where site specific action is contemplated.
Such action is essentially exposure reduction by managing the access to and/or actions of the
road-using public on the network at times either when events occur or precursor rainfall has
indicated a high likelihood of landslides occurring.
In the case of short-term to medium-term reaction to such occurrences, then the DetectionNotification-Action (DNA) approach can be implemented by pre-planned actions such as
issuing an advisory warning or closing the road. There may be a case for reacting to
extremely heavy rainfall events in a similar fashion, especially with warnings. A caveat to
this is the need to consider carefully at what levels the triggers should be set, in so far as the
relationship between rainfall and landslides in Scotland is by no means fully understood.
Considering the longer-term approach, precursor triggering conditions (i.e. rainfall) may
enable many of the actions described above to be taken prior to the occurrence of major
events. Either an extensively enhanced network of rain gauges installed across Scotland or
access to data derived from radar and of sufficient resolution would be required. Such work
initially be concentrated on known storm tracks, if these are available from the
Meteorological Office, and vulnerable slopes. Clearly, if this approach is taken then a close
consultation with both the Geotechnical Engineering Office in Hong Kong, which has
112
SUMMARY AND RECOMMENDATIONS
extensive experience of operating such a system albeit in different climatological and
geological conditions, and the UK Meteorological Office would be highly desirable.
It is fully expected that it will take some considerable time and effort to ensure that sufficient
data has been obtained and analysed so as to be able to introduce a warning system. Even
then it must be expected that atypical events, which are not the subject of warnings, may
occur. Also a number of false alarms may inevitably be expected. A programme of public and
media education and awareness-raising is also likely to be desirable to minimise any potential
adverse reaction to such scenarios.
‘Do-Something 2’ involves more major works in order to achieve hazard reduction (as
opposed to exposure reduction in the ‘Do-Something 1’ case). The approaches involved entail
physical measures such as the protection of the road, reduction of the opportunity for a debris
flow to occur or realignment of the road away from the area of high hazard. Such options
need to be considered in the context of the policy governing the Scottish Executive’s overall
trunk road maintenance and construction programme. In general, these are likely to be of high
cost necessitating their restriction to the very few areas of highest hazard ranking.
Clearly, and as illustrated in Figure 9.1, Monitoring and Feedback is fundamental to the
success of the system and key to deriving best value from the arrangements proposed. The
system developed is an active one and lessons learned from future landslide events, whether
they occur in areas of high or very high hazard ranking or not, will produce valuable data
which needs to be taken into account in adjusting the parameters that form the cornerstone of
the assessment methodology.
There exists a need to ensure that actions identified by the existing Rock Slope Hazard Index
system (as developed in the early 1990s) are carried out on a priority budget basis. These will
include both maintenance works and re-inspection activities. While the rock slope system and
the proposed landslide system have very different structures, great efforts have been made to
ensure that the critical exposure evaluation and the output categories are capable of being
mutually compatible.
113
REFERENCES
Anon. 1989. The climate of Scotland – some facts and figures. London: The Stationery Office.
Anon. 1999. Flood
www.nwl.ac.uk/ih.
Estimation
Handbook.
Wallingford:
Institute of
Hydrology.
Anon. 2004. Partnerships for Reducing Landslide Risk, Assessment of the National Landslide
Hazards Mitigation Strategy. Washington D.C.: National Research Council.
Australian Geomechanics Society. 2000. Landslide risk management concepts and guidelines.
Australian Geomechanics, 35(1), 51-92.
Bagnold, R. A. 1954. Experiments on a gravity-free dispersion of large solid spheres in a
Newtonian fluid under shear. Proceedings, Royal Society of London, Series A, 225, 49-63.
Ballantyne, C. K. 1981. Periglacial landforms and environments on mountains in the
Northern Highlands of Scotland. Unpublished PhD Thesis. Edinburgh: University of
Edinburgh.
Ballantyne, C. K. 1984. The Late Devensian periglaciation of upland Scotland. Quaternary
Science Reviews, 3, 311-343.
Ballantyne, C. K. 1986. Landslides and slope failures in Scotland: A review. Scottish
Geographical Magazine, 102(3), 134-150.
Benda, L. & Dunne, T. 1997. Stochastic forcing of sediment supply to channel networks from
landsliding and debris flow. Water Resources Research, 33(12), 2849-2863.
Berti, M., Genevois, R., Simoni, A. & Rosella Tecca, P. 1999. Field observations of a debris
flow event in the Dolomites. Geomorphology, 29, 265-274.
Bingham, E. C. & Green, H. 1919. Paint, a plastic material and not a viscous liquid; the
measurement of its mobility and yield value. Proceedings, American Society for Testing
Materials, 19, 640-664.
Bovis, M. J. 1993. Hillslope geomorphology and geotechnique. Progress in Physical
Geography, 17(2), 173-189.
Brunengo, M. J. 2002. Who knows/who cares/why bother about debris flows? Washington
State’s experience translating information into policy. Proceedings, Geological Society of
America, Cordilleran Section, 98th Annual Meeting.
Bryant, E. 1991. Natural hazards. Cambridge: Cambridge University Press.
Caine, N. 1980. The rainfall intensity-duration control of shallow landslides and debris flows.
Geografiska Annalerum, 62A, 23-27.
114
REFERENCES
Chan, R. K. S. 2000. Hong Kong slope management system. Proceedings of the Symposium
on Slope Hazards and Their Prevention, The Jockey Club Research and Information Centre
for Landslip Prevention and Land Development. Hong Kong: The University of Hong Kong.
Coussot, P. & Meunier, M. 1996. Recognition, classification and mechanical description of
debris flows. Earth Science Reviews, 40, 209-227.
Cruden, D. M. & Varnes, D. J. 1996. Landslide types and processes. In: Special report 247:
Landslides: Investigation and Mitigation (Eds: Turner, A. K. & Schuster, R. L.), 3675.Transportation and Road Research Board, Washington, D. C.: National Academy of
Science.
Design Manual for Roads and Bridges. London: The Stationery Office.
DMRB 4.1.3, HA48/93 Maintenance of Road Earthworks and Drainage, May 2001.
DMRB 4.1.3, HA41/03 Maintenance of Highway Geotechnical Assets, November 2004.
EPOCH. 1993. The temporal occurrence and forecasting of landslides in the European
community (Ed: Flageollet, J. C.). Contract No. 90 0025, 3 Volumes.
Erismann, T. H. & Abele, G. 2001. Dynamics of rockslides and rockfalls. Berlin: Springer.
Escario, M. V., George, L.-A., Cheney, R. A. & Yamamura, K. 1997. Landslides: techniques
for evaluating hazard. Report of PIARC Technical Committee on Earthworks, Drainage,
Subgrade (C12), 12.04B. Paris: PIARC, World Road Association.
Evans, N. C. & King, J. P. 1998. The natural terrain landslide study: debris avalanche
susceptibility. GEO Technical Note TN 1/98. Hong Kong: Geotechnical Engineering Office.
Fisher, R. V. 1983. Flow transformations in sediment gravity flows. Geology, 11, 273-274.
Franks, C. A. M. 1999. Characteristics of some rainfall induced landslides on natural slopes,
Hong Kong. Quarterly Journal of Engineering Geology, 32, 247-259.
Hearn, G. J. & Petley, D. N. 2002. Unpublished proceedings of Mountain Hazards Research
and Risk Minimisation Seminar. Nepal: Kathmandu.
House of Commons. 1984. Roads (Scotland) Act 1984. (Commencement No.3) Order 1990:
Statutory Instruments 1990 262 (C.72)(SI.217). London: The Stationary Office.
Hungr, O., Morgan, G. C. & Kellerhals, R. 1984. Quantitative analysis of debris torrent
hazards for design of remedial measures. Canadian Geotechnical Journal, 21, 663-676.
Hungr, O., Morgan, G. C., Van Dine, D. F. & Lister, D. R. 1987. Debris flow defences in
British Columbia. In: Debris Flow/Avalanches: Process, Recognition and Mitigation (Eds:
Costa, J. E. & Wieczorek, G. F.). Reviews in Engineering Geology Volume VII, 201-222.
Boulder, CO: Geological Society of America.
Hutchinson, J. N. 1988. General Report: Morphological and geotechnical parameters of
landslides in relation to geology and hydrogeology. Proceedings, Fifth International
Symposium on Landslides (Ed: Bonnard, C.), 1, 3-35. Rotterdam: Balkema,
115
REFERENCES
Innes, J. L. 1982. Debris flow activity in the Scottish Highlands. Unpublished PhD Thesis.
Cambridge: University of Cambridge.
Innes, J. L. 1983a. Lichenometric dating of debris flow deposits in the Scottish Highlands.
Earth Surface Processes and Landforms, 8, 579-588.
Innes, J. L. 1983b. Debris flows. Progress in Physical Geography, 7, 469-501.
Johnson, A. M. 1970. Physical processes in geology. San Francisco, CA: Freeman-Cooper.
Johnson, A. M. & Rodine, J. R. 1984. Debris Flow. In: Slope Instability (Eds: Brunsden, D. &
Prior, D. B.). John Wiley & Sons Ltd., Chichester, UK. 257-361.
Jones, D. K. C. & Lee, E. M. 1994. Landsliding in Great Britain. London: HMSO.
Kirkby, M. J. 1963. A study of rates of erosion and mass movements of slopes, with special
reference to Galloway. Unpublished PhD Thesis. Cambridge: Trinity College.
Larsen, M. C. & Parks, J. E. 1997. How wide is a road? The association of roads and masswasting disturbance in a forested montane environment. Earth Surface Processes and
Landforms, 22, 835-848.
Lee, E. M. & Jones, D. K. C. 2004. Landslide risk assessment. Thomas Telford, London.
Malone, A. W. 1998. Risk management and slope safety in Hong Kong. Proceedings,
Seminar on Slope Engineering in Hong Kong, 3-17. Rotterdam: Balkema.
McAdam, D. 1993. Edinburgh – a landscape fashioned by geology. Edinburgh: Scottish
Natural Heritage and British Geological Survey.
McMillan, A. A. & Powell, J. H. 1999. BGS Rock Classification Scheme, Volume 4:
Classification of artificial (man-made) ground and natural deposits - applications to
geological maps and datasets in the UK. British Geological Survey Research Report No. RR
99-04. p 65.
McMillan, P. 1995. Rock slope risk assessment. Project Report PR/SC/22/95. Edinburgh:
TRL Limited. (Unpublished report available only on direct personal application to TRL
Limited, Edinburgh.)
McMillan, P. & Matheson, G. D. 1997. A two stage new system for road rock slope risk
assessment. International Journal of Rock Mechanics & Mineral Science, 34:3-4, Paper No.
196.
Middleton, G. V. & Hampton, M. A. 1976. Subaqueous sediment transport and deposition by
sediment gravity flows. In: Marine sediment transport and environmental management (Eds:
Stanley, D. J. & Swift, D. J. P.). London: J Wiley & Sons.
Moore, R., Lee, E. M. & Clark, A. R. 1995. The undercliff of the Isle of Wight: a review of
ground behaviour. Newport, Isle of Wight: Cross Publishing.
116
REFERENCES
Moore, R., Lee, E. M. & Palmer, J. S. 2002. A sediment budget approach for estimating
debris flow hazard and risk: Lantau, Hong Kong. Instability, Planning and Management:
seeking sustainable solutions to ground movement problems (Eds: McInnes, R. G. &
Jakeways, J.). London: Thomas Telford Publishing.
Nettleton, I. M., Tonks, D. M., Low, B., MacNaughton, S. & Winter, M. G. In Press. Debris
flows from the perspective of the Scottish Highlands. Proceedings, 11th International
Conference on Landslides, Norway, 1-10 September 2005.
Pierson, T. C. 1985. Initiation and flow behaviour of the 1980 Pine Creek and Muddy River
lahars, Mount St. Helens, Washington. Geological Society of America Bulletin, 96, 10561069.
Pierson, T. C. 1995. Flow characteristics of large eruption-triggered debris flows at snow-clad
volcanoes: constraints for debris flow models. Journal of Volcanology and Geothermal
Research, 66, 283-294.
Pierson, T. C. & Costa, J. E. 1987. A rheological classification of subaerial sediment-water
flows. In: Debris Flow/Avalanches: Process, Recognition and Mitigation (Eds: Costa, J. E. &
Wieczorek, G. F.). Reviews in Engineering Geology Volume VII, 1-12. Boulder, CO:
Geological Society of America.
Pierson, T. C. & Scott, K. M. 1985. Downstream dilution of a lahar: transition from debris
flow to hyperconcentrated streamflow. Water Resources Research, 21, 1511-1524.
Reneau, S. L. & Deitrich, W. E. 1987. The importance of hollows in debris flow studies;
examples from Marin County, California. In: Debris Flow/Avalanches: Process, Recognition
and Mitigation (Eds: Costa, J. E. & Wieczorek, G. F.). Reviews in Engineering Geology
Volume VII, 165-180. Boulder, CO: Geological Society of America.
Sidle, R. C., Pearce, A. J. & O'Loughlin, C. L. 1985. Hillslope stability and land use. Water
Resources Monograph. 140pp. Washington, D.C.: American Geophysical Union.
Smith, G. A. 1986. Coarse-grained non-marine volcaniclastic sediment: Terminology and
depositional process. Geological Society of America Bulletin, 97, 1-10.
Smith, G. A. & Lowe, D. R. 1991. Lahars: Volcano-hydrologic events and deposition in the
debris flow - hyperconcentrated flow continuum. Sedimentation in volcanic settings (Eds:
Fisher, R. V. & Smith, G. A.). SEPM Special Publication, 45, 99-106.
Spiker, E. C. & Gori, P. L. 2003. National Landslide Hazards Mitigation Strategy – A
Framework for Loss Reduction. USGS Circular 1244.
Strachan, G. J. 1976. Debris flows in Wester Ross. Unpublished MA (Hons) Dissertation,
Aberdeen: University of Aberdeen.
Takahashi, T. 1978. Mechanical characteristics of debris flow. Journal of the Hydraulics
Division, American Society of Civil Engineers, 104, 1153-1169.
117
REFERENCES
Takahashi, T. 1980. Debris Flow on prismatic open channel. Journal of the Hydraulics
Division, American Society of Civil Engineers, 106, 381-396.
Takahashi, T. 1981. Debris Flow in Dyke. In: Annual review of fluid mechanics (Eds: Van,
M., Wehausen, J.V. & Lumley, J. L.), 13, 57 – 77.
Tivy, J. 1962. An investigation of certain slope deposits in the Lowther Hills, Southern
Uplands of Scotland. Transactions of the Institute of British Geographers, 30, 59-73.
Vandine, D. F. 1985. Debris flows and debris torrents in the Southern Canadian Cordillera.
Canadian Geotechnical Journal, 22, 44-68.
Varnes, D. J. 1978. Slope movement types and processes. In: Special Report 176: Landslides:
Analysis and Control (Eds: Schuster, R. L. & Krizek, R. J.). Transportation and Road
Research Board, National Academy of Science, Washington D. C., 11-33.
Watters, R. J. 1972. Slope stability in the metamorphic rocks of the Scottish Highlands.
Unpublished PhD Thesis. London: Imperial College of Science and Technology.
Wieczorek, G. F., Mossa, G. S. & Morgan, B. A. 2004. Regional debris-flow distribution and
preliminary risk assessment from severe storm events in the Appalachian Blue Ridge
Province, USA. Landslides, 1, 53-59.
Wilford, D. J., Sakals, M. E., Innes, J. L., Sidle, R. C. & Bergerud, W. A. 2004. Recognition
of debris flow, debris flood and flood hazard through watershed morphometrics. Landslides, 1,
61-66.
Winchester, S. 1985. Outposts: Journeys to the Surviving Relics of the British Empire.
London: Penguin.
Winter, M. G., Nettleton, I. M. & McMillan, P. In Preparation. A full-scale destructive trial of
a rockfall catch fence. Paper for submission to the Quarterly Journal of Engineering Geology
and Hydrogeology.
WP/WLI. 1990. A suggested method for reporting a landslide. Bulletin of the International
Association of Engineering Geology, No. 41, 5-12.
WP/WLI. 1991. A suggested method for a landslide summary. Bulletin of the International
Association of Engineering Geology, No. 43, 101-110.
WP/WLI. 1993. A suggested method for describing the activity of a landslide. Bulletin of the
International Association of Engineering Geology, No. 47, 53-57.
WP/WLI. 1995. A suggested method for describing the rate of movement of a landslide.
Bulletin of the International Association of Engineering Geology, No. 52, 75-78.
Yarnold, J. C. 1993. Rock-avalanche characteristics in dry climates and the effect of flow into
lakes: Insights from mid-Tertiary sedimentary breccias near Artillery Peak, Arizona.
Geological Society of America Bulletin, 105, 345-360.
118
APPENDIX – PROJECT WORKSHOP AGENDA
Debris Flow Risk Assessment and Mitigation on the Scottish Trunk Road Network
A Project Workshop held by the Scottish Executive
Tuesday 28 September 2004 at Pentland Suite, Corus Edinburgh North Hotel
Facilitator:
0900
Professor Malcolm Horner, Dundee University.
Coffee
0930 Welcome, Professor Malcolm Horner, Dundee University.
Introduction, Forbes Macgregor, Scottish Executive and Dr Mike Winter, TRL
Limited.
0945 Debris Flows on the A9, A85 and A83, Andy Heald/Julie Parsons, BEAR.
1000 The Potential for Debris Flows on the SE/SW Network, Paul McMillan, Amey.
1015 Shetland Peat Flow Case Study, Stewart Martin, Halcrow.
1030 Coffee and Informal Discussions
1100 Debris Flows at Stromeferry Bypass, Ian Nettleton, EDGE Consultants.
1115 Debris Flow Indicators, Dr Mike Winter, TRL Limited.
1130 Risk and Hazard Assessment Techniques, Alan Forster, BGS.
1145 Remote Rapid Assessment Techniques, Matthew Willis, Arup.
1200 Asset Management, Andy Sloan, Donaldson Associates.
1215 Discussion Session 1: Data Needs for Risk Assessment and Mitigation.
1300 Lunch and Informal Discussions
1400 Discussion Session 2: Risk Assessment Techniques.
1500 Discussion Session 3: Mitigation and Risk Management Strategies.
1530 Tea and Informal Discussions
1600 Discussion Session 3 (Cont’d): Mitigation and Risk Management Strategies.
1630 Discussion Session 4: Obvious Areas of Greatest Potential Risk on the Trunk Road
Network.
1700 Close
Other Attendees:
Lawrence Shackman, NMD, Scottish Executive;
Polly Griffiths (Reporter), TRL.
119
colour green
24/5/05
© Crown
1:08 pm
Page 1
copyright 2005
ISBN: 0-7559-4649-9
This information is available on the Scottish Executive website
www.scotland.gov.uk
Astron B41378 06/05