The Scottish Government

« Previous | Contents | Next »

Listen

1 INTRODUCTION

In August 2004 Scotland experienced rainfall substantially in excess of the norm. Some areas of Scotland received in excess of three times the 30-year average August rainfall. In the Perth and Kinross area figures of the order of between two and a half and three times were typical. While the percentage rainfall during August reduced to the west, parts of Stirling and Argyll & Bute still received between two and two and a half times the monthly average ¹.

The rainfall was both intense and long lasting and a large number of landslides, in the form of debris flows ( see Section 2), were experienced in the hills of Scotland. A small number of these intersected with the trunk road network, notably the A83 between Glen Kinglas and to the north of Cairndow (9 August), the A9 to the north of Dunkeld (11 August), and the A85 at Glen Ogle (18 August).

While major injuries were avoided, some 57 people were taken to safety by helicopter after being trapped between the two debris flows on the A85 in Glen Ogle ( see picture (© Perthshire Picture Agency, PPA: www.ppapix.co.uk)). The A85, carrying up to 5,600 vehicles per day ², was closed for four days. The A83, which carries around 5,000 vehicles per day, was closed for two days; and the A9, carrying 13,500 vehicles per day, was closed for two days prior to reopening. The disruption experienced by local and tourist traffic, as well as to goods vehicles, was substantial.

Debris flows occur with some frequency in Scotland, albeit affecting the trunk road and the main local road network only rarely. However, when they do impact on the road network the degree of damage, 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 road network. 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.

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 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.

The need to act has been recognised by the Scottish Executive and this initial study (Study 1, Part 1) has been commissioned alongside a second study (Study 2). Study 2 is designed to identify the potential impacts and consequent necessary actions in the light of anticipated climate change and is not considered further in this report, although it is important to note that action has been taken to ensure that the two studies are complementary.

As indicated above, this study, termed Study 1, comprises two parts and it is Part 1 that is reported here. Part 1 deals with the following activities:

• Considering the options for undertaking a detailed review of side slopes adjacent to the trunk road network and recommending a course of action.
• Outlining possible mitigation measures and management strategies that might be adopted.
• Undertaking an initial review to identify obvious areas that have the greatest potential for similar events in the future.

This work will lead to Study 1, Part 2 which will include the development of a system to allow a detailed review of the network to be undertaken to identify the locations of greatest hazard and for those hazards to be ranked and appropriate mitigation and/or management measures to be selected.

The overall purpose of these studies is thus to ensure that the Scottish Executive has a system in place for assessing the hazards posed by debris flows. In addition, the system will rank the hazards in terms of their potential relative effects on road users. This will ensure that the exposure of road users to the consequences of future debris flow events is minimised whilst acknowledging that it is not possible to prevent the occurrence of such events.

A consistent, repeatable and reproducible system is required. This is especially important as a variety of consultants is likely to be involved in the data gathering, analysis and interpretation process. It is apparent at the outset that a unified system acceptable to all of the major players in the industry is required.

It was thus recognised at an early stage of the development of the work that the input of a wide range of experts and stakeholders would be required in order for the studies to be completed successfully.

A Project Workshop was held in order to capture the knowledge vested with individual experts. The Project Workshop was facilitated by Professor Malcolm Horner of the University of Dundee and comprised presentations given by acknowledged experts followed by focussed discussion sessions designed to open out the knowledge base and determine the way forward with the project.

Following the Project Workshop the Report Editors assigned tasks to individuals (including to themselves) in terms of the preparation of a Technical Report (Winter et al., 2005). The individuals were selected in order to enable those most suited to the various tasks to bring their knowledge, expertise and experience to bear on the relevant issues. The individuals involved and their affiliations are detailed on page 6. The work has been funded through a variety of existing contracts with the close and active involvement and support of Scottish Executive engineers as key members of the Working Group.

The Technical Report is intended to assist the professionals charged with taking the work on landslides forward to implementation and also to provide information for the wider professional community. This Summary Report is intended to inform a wider audience of the Scottish Executive's actions both since the events of August 2004 and planned for the future.

Section 2 of the Summary Report gives the background to the Study as a whole. It describes the different types of landslide, focussing on debris flows as recently experienced, and illustrates the recent history of debris flows in Scotland with examples right up to the present. It also deals with climatic issues and identifies data relevant to future work on landslides.

Section 3 describes the proposed assessment methodology in terms of hazard assessment and approach for Study 1, Part 2 and also details the hazard assessment and exposure factors that will form the core of the methodology for the detailed assessment.

Section 4 identifies areas of high hazard that are considered to have the greatest potential for similar debris flow events in the future and sets out opportunities for early actions.

Section 5 presents a brief summary of this report and makes recommendations for the way forward. In terms of management, the logical and sequential approach of Detection, Notification and Action ( DNA) is used. This approach is set out in terms of a response to both precursor conditions, such as intense rainfall, and also to the management of future debris flow events.

2 BACKGROUND

2.1 Landslides

Recent extreme rainfall in Scotland has led to events that have been described in the media using the generic term 'landslide'. These events have intersected with the A83 (between Glen Kinglas and to the north of Cairndow), A9 (to the north of Dunkeld) and A85 (Glen Ogle) trunk roads.

While the recent happenings have been of both high magnitude (in terms of the amount of material moved) and severe (in terms of their impact on the trunk road network and the exposure of its users) it is important to understand that they are by no means unique. Similar events have been observed in recent years by Nettleton et al. (In Press) at Invermoriston, intersecting the A887, and at Stromeferry, intersecting the A890 local road. Other events have been observed at A83 Rest and be Thankful, A9 Slochd, A95 Craigellachie and A84 Strathyre, for example.

Figure 2.1 presents the five distinct types of landslide identified by Varnes (1978), as follows (after Escario et al., 1997).

Figure 2.1 - Types of landslide: (a) falls, (b) topples, (c) slides, (d) flows, and (e) spreads (after Escario et al., 1997).

Varnes (1978) also presented a sixth mode of movement, Complex Failures. These are failures in which one of the five types of movement is followed by another type (or even types).

The recently observed landslide events have been typical of flow-type landslides. The influence of substantial flows of water, the stripping of superficial deposits, and the speed with which debris has both flowed and been deposited have all been apparent. In many cases the initial trigger appears to have been the displacement of relatively small amounts of material, often into a stream channel. This has added a substantial debris charge to already high and potentially damaging water flows. The combination of water with high sediment loadings then has substantial erosive power. In other cases highly saturated materials have slumped rapidly downslope in a manner not dissimilar to that illustrated in Figure 2.1(d).

Such events are typically described as 'debris flows'. They are distinguished from many other types of landslides involving shear by their rapid movement as opposed to relatively slow movements experienced in other failure types. This is an important distinction and not simply an academic nicety. Failure to make such a distinction could very easily lead to inappropriate data being collected and inappropriate approaches being proposed.

Rockfall-type landslides (Figure 2.1(a) and (b)) are already the subject of a hazard assessment and ranking system (McMillan and Matheson, 1977) on the Scottish trunk road network. While it is important that any system to emerge from this work on debris flows is compatible with this system (see Section 3) such falls need not otherwise be considered further here. Of the remaining types it is flows, including debris flows and peat flows, which have most adversely affected the trunk road network in the recent past and are likely to continue to do so in the future.

Flows are largely dynamic in their trigger mechanisms and are generally characterised by rapid erosion and movement with high proportions of either water or air acting as a lubricant for the solid material that generally comprises the bulk of their mass.

Debris flows occur, in the main, because of the character of natural slopes, the deposits of which they are comprised, and the amount and duration of rainfall (and consequent infiltration) to which they are subject. The fact that they impact on a road network is, irrespective of the consequences, coincidental in the phenomenological sense. Debris flows affecting the trunk road network are not caused by its construction and/or management, except in unusual circumstances. However, some aspects of the built environment, including a road network, may contribute to the outcomes of such events.

It is important to note that debris flows are neither a recent phenomenon nor an uncommon occurrence. The first church in the Falkland Islands, for example, was wrecked in 1886 when a "river of liquid peat … roared down from the hills" (Winchester, 1985). Closer to home, a cloud burst in 1744 resulted in the flow and associated erosion of the gulley below the summit of Arthur's Seat known today as the Gutted Haddie (McAdam, 1993). Innes (1983) made a survey of Scotland based upon aerial maps and marked those 10km by 10km grid squares that showed some sign of debris flow activity (Figure 2.2), clearly indicating that such activity is far from unusual.

It is clear that the August 2004 events in Scotland had the potential to cause injury and even death. However, such potential was not on the same scale as the reality that is experienced elsewhere in the world on a regular basis. For example, in September 2004, torrential rain triggered massive floods and landslides in SW China, killing in excess of 170 people and injuring many dozens more ³.

Figure 2.2 - The extent of recorded debris flow activity in Scotland (from Jones and Lee, 1994; after Innes, 1983). Note that the figure does not record any activity in the area around the Rest and be Thankful, for example. It seems unlikely that there was no such activity prior to 1983 when the figure was first published and the data set should thus not be seen as exhaustive.

2.2 Recent Debris Flows

In recent years debris flow events appear to have had an increasing effect on the Scottish trunk and local road network, and the Scottish rail network. At face value this suggests that such events have become more common. Such a conclusion would however be somewhat speculative as comprehensive, detailed records are not generally available for events that do not impact upon man's activities. What does appear clear from simple observation is that many debris flows are initiated on the Scottish hills. However, only a relatively small number turn into major events that impact upon road networks or other forms of infrastructure. This implies that in order to manage the impacts of debris flows it is necessary to understand the preparatory factors (that make a slope vulnerable to debris flows), the trigger factors (that lead to initiation of flows) and any propagation and/or magnification factors.

A number of debris flows have historically occurred in the month of August. One example is an event that intersected the A887 at Invermoriston in 1997 (Figure 2.3). This event was studied in detail and found to have been triggered at a point almost 300m vertically and around 2,000m horizontally from the road, close to the source of the stream which subsequently contained most of the event. A number of contributory factors were established (Nettleton et al., In Press), including the following:

• The lack of water storage volume within the catchment, both above and below ground.
• The ploughing of agricultural land increases and accelerates runoff into streams.
• The presence of downslope bedding planes.
• Low permeability bedrock.
• The presence of forestry bridges which temporarily arrested the flow allowing material to accumulate and subsequently remobilise with greater erosive power.
• The presence of a buried cliff providing a large amount of debris at a point close to the road.
• A steep slope close to the road.

Figure 2.3 - Debris flow at Invermoriston (A887) in August 1997. (Courtesy of Northpix.)

Many of the features of the slope at Invermoriston, such as its convex shape ( i.e. steepening downslope) are characteristic of glacial valleys which are in turn typical of much of the landscape of Scotland. The event was preceded by rainfall of both long duration and high intensity. As a result of the debris flow the road was closed, damage was sustained to vehicles and a local hotel only narrowly escaped substantial physical damage.

Debris flow events have also been observed at other times of the year. They have affected both the A890 and the railway at Stromeferry in January 1999, October 2000 and October 2001 (Figure 2.4). The January 1999 and October 2000 events were characterised by the mobilisation of material from a pre-existent landslide which slipped into a gully thus providing the source material for the debris flow event. The October 2001 event was propagated from a gully that had been infilled with silt, gravel and cobble fractions. In each case disruption to the road and railway was experienced.

Figure 2.4 - Debris flow at A890 Stromeferry in October 2001. (Courtesy of and © copyright Alex Ingram.)

The effects of forestry have frequently been identified as, at least, partial causes or propagators of debris flows in areas such as the Pacific NW of the USA (Brunengo, 2002). Logging or deforestation can have a dramatic effect on the drainage patterns of a slope, reducing root moisture uptake, slope reinforcement due to the root systems, and the physical restraints on downslope water flow for example. Such effects were especially noted as factors in the triggering of a translational landslide (not a debris flow) at Loch Shira adjacent to the A83 trunk road near Inverary in December 1994.

Returning to the more recent debris flows of August 2004, these occurred at three main locations as discussed in the following paragraphs.

The A83 was blocked at two locations in Glen Kinglas and at a point approximately 1km north of Cairndow (Figure 2.5) and the road here was closed for two days. Numerous smaller debris flows were also observed on the hill slopes either side of the glen.

The A9 was blocked by three main debris flows, two of which corresponded with areas of instability adjacent to the old A9 which runs parallel to, and upslope from, the present trunk road (Figure 2.6). In such circumstances both forest roads and minor roads can act to slow and concentrate the downslope flow of water and thus aid its penetration into the slope below. Such a mechanism has been a factor in a number of previous events such as the washout that blocked the A83 Rest and be Thankful in the vicinity of Roadman's Cottage, in 1999.

Figure 2.5 - Debris fan containing boulders (estimated up to 9 tonnes) at A83 Cairndow.

Figure 2.6 - Hillslope flow which has formed its own channel by erosion (A9 north of Dunkeld, August 2004). (Courtesy of Alan MacKenzie, BEAR.)

However, in the A9 Slochd failure of July 2002 it was the presence of the trunk road that contributed to the failure of the old road (used as a cycle path) and consequently to its own failure by undercutting. The presence of forest tracks was also identified as a factor in the debris flow at Invermoriston.

In the A85 incident the road was blocked by two landslides. The southerly slip occurred first and as advice was being offered to motorists by Operating Company staff a second landslide occurred to the north of the first (Figure 2.7). The two landslides effectively trapped 20 vehicles, and 57 occupants were airlifted to safety by RAF and Royal Navy helicopters (see cover photograph). In its latter phases the northerly debris flow surmounted a spur of rock and an unoccupied Operating Company vehicle that had been parked in the lee of the spur was swept over the edge of the road and some distance downslope before it came to rest against a tree.

Figure 2.7 - View of the second and larger of the A85 Glen Ogle debris flows, showing the sharp bend in the channel just above road level.

Since August 2004 further landslides have affected the Scottish road network on the A82 near Letterfinlay alongside Loch Lochy (January 2005). In addition rock falls have affected the A832 near Kinlochewe (December 2004) and the A82 1.5 miles north of the Corran Ferry junction (January 2005).

2.3 Climatic Issues

The climate of Scotland in terms of its rainfall may be very broadly divided into east and west (see Figures 2.8 and 2.9). Data presented by the Meteorological Office (Anon, 1989) indicates the following key comparisons:

• In the east rainfall generally peaks in August while in the west the maximum rainfall levels are reached during the wider period September to January (Figure 2.8).
• Although rainfall levels in the west are relatively low in August they increase from a low point in May.

Both of these scenarios indicate that the soil may be undergoing a transition from a dry to a wetter state at or around August, indicating an increased potential for debris flow and other forms of landslide activity.

The central area, as represented by Pitlochry in Figure 2.8, experiences a mix between the rainfall characteristics of the 'east' and the 'west'.

Figure 2.8 - Annual average rainfall data for points in Scotland.

The soil water conditions necessary for debris flows may be generated by long periods of rainfall or by shorter intense storms. It is however widely accepted that Scottish debris flow events are usually preceded by both extended periods of heavy rainfall (otherwise known as antecedent rainfall) and intense storms.

A number of important points in relation to landslide activity may be drawn from climate change model predictions for Scotland in the 2080s ⁴, as follows:

• In the summer precipitation will decrease but increase in the winter.
• That the models are less good at predicting localised summer storms.
• Such storms are believed to be at least partially responsible for triggering the events of August 2004.
• Climate data may not give a full picture of the relationship between precipitation and landslides.
• As climate models generally predict averages the error limits can be substantial.
• Predicted changes in the number of 'intense' wet days generally indicate a net increase of less than one day per annum by the 2080s, with slightly fewer intense wet days in the summer and more in the winter.
• However, by the 2080s extreme storm event rainfall depths are predicted to increase by between 10% and 30%, with intense winter rainfall increasing slightly more than this and that in spring/autumn by slightly less.
• Summer extreme rainfall depths are predicted to increase by between 0% and 10%.
• Peak river flows are anticipated to increase progressively during the 21 ^st century. Eastern Scotland is expected to experience larger increases than north-west Scotland for example.
• The occurrence of snow and the associated contribution of snowmelt to both river flow and groundwater are, on the other hand, predicted to decrease.
• Reductions in snowfall are predicted to be greater for the eastern and southern parts of Scotland and least for the central upland areas.

Changes in the factors discussed above coupled with increased evaporation and plant transpiration, particularly in the summer, and a longer growing season, leading to increased root water uptake, are expected to have substantial effects on soil moisture. The models predict a 10% to 30% decrease in soil moisture for summer/autumn and an increase of 3% to 5% in the winter. The winter figures reflect the fact that soils can only contain so much water and most Scottish soils are already close to saturation in the winter.

Figure 2.9 - Example of Meteorological Office 30-year monthly average rainfall data for October (image courtesy of the Meteorological Office).

Reduced soil moisture during the summer and autumn months may mean that potentially high suction pressures are generated in the soils that make up some slopes. This may in turn make the slopes more vulnerable to the effects of intense storms as the suction pressures can draw water into the dry slope very rapidly. Cracking of the dry soil may also aid such effects. These issues are explored in more detail by Winter et al. (2005).

The UKCIP ( UK Climate Impacts Programme) report considers three periods: the 2020s, the 2050s and the 2080s. In general terms small changes are noted in the predictions for the 2020s. These changes increase slightly for the 2050s and slightly further still for the predictions for the 2080s, reflecting the temporal trends in temperature and precipitation. Whilst climate models generally predict averages and the associated error limits can be substantial, it is also important to note that inter-annual variability is predicted to increase for many climate factors. This means that average changes, as discussed above, may mask more important variability effects.

2.4 Relevant Information Sources

There is a wide range of information and data available that is relevant to landslide activity in general and to debris flows in particular in Scotland.

In the Technical Report a detailed review of the available data in relation to landslides is presented. Such data includes that relating to the geology, climate, river and streams and land use of Scotland. Also of relevance are traffic data held by the Scottish Executive and commercially available digital terrain models.

A thorough literature review of more than 100 papers, articles, publications, reports and books was undertaken. The results of the review are presented in the main report and are somewhat beyond the scope of this summary report. Similarly the many factors influencing debris flow activity and exposure to such activity identified from the Project Workshop are also presented in the main report. Also included in the main report are reviews of debris flow types and mechanisms and the key contributory factors to debris flows.

The interested reader is referred to Winter et al. (2005).

3 PROPOSED METHODOLOGY

3.1 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 non-peat drift deposits. The second will be based on slope angle. A relatively high slope angle (around 26 ^o to 50 ^o) will be applied to debris flow formation in non-peat deposits with a slope angle of around 8 ^o 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 5 ^o) 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 Winter et al., 2005) 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 ( f') 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.

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.

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 3.3. 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 Geographical Information System-based ( GIS-based) assessment with those areas of high hazard assessed during the Project Workshop and detailed in Section 4.

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 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.

3.2 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 double-counting (or even treble-counting) of the exposure factors.

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 AADT⁵) 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 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.

3.3 Detailed Assessment Factors

A very detailed and comprehensive set of factors was generated at the Project Workshop (see Winter et al., 2005). 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.

3.3.1 Hazard Factors

The first stage assessment, as described in Section 3.1, considers two categories of debris flow hazard assessment.

• The first will effectively seek areas of peat on slope angles of 5 ^o 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 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 8 ^o 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 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.

3.3.2 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.

3.3.3 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.

A detailed assessment (Winter et al., 2005) indicates that the key element that must be addressed in order to ensure compatibility is the means of assessing exposure. In addition ensuring that each system operates with the same number of hazard ranking categories with a broad degree of compatibility between them is also necessary.

4 HIGH HAZARD AREAS AND EARLY OPPORTUNITIES

The early identification of high hazard sites, on a subjective basis by acknowledged specialists at the Project Workshop, 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.

The areas considered to present a sufficiently high hazard to warrant concern was defined. After the Project Workshop, digital Ordnance Survey mapping (1:50,000 scale) was used to inspect the areas identified in both plan form and also using a built in digital elevation model.

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 as follows:

• 29km of the A83 between Ardgarten and Loch Shira.
• 8km of the A84 to the South of Strathyre.
• 6km of the A85 in Glen Ogle.
• 18km of the A87 in Glenshiel (plus a further 17km either end of Glenshiel).
• 29km of the A82 between Fort Augustus and Lochend (plus a further 9km to the south).
• 16km of the A835 between Ullapool and Braemore Junction.
• 22km of the A9 between Dunkeld and Drumochter.
• 1km of the A95 in the Craigellachie area.
• 5.5km of the A86 around Spean Bridge.
• 1.5km of the A87 (Skye) between Gleann Torra-mhichaig and South of the Raasay ferry.

In terms 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: Maintenance of Highway Geotechnical Assets). Critical review of culvert alignments ought to be carried out as part of inspection and reporting procedures.

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 (National Driver Information Centre Scotland), 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'.

5 THE WAY FORWARD

5.1 Early Opportunities

A number of areas of early opportunities have been identified in Section 4. These include the use of the areas of high hazard identified at the Project Workshop as a tool for 'ground-truthing' the outputs from future debris flow hazard assessments using both GIS techniques and site-based methodologies.

Opportunities for ongoing maintenance and construction activities to take account of factors that might lower the likelihood of debris flow activity by the provision of improved and/or more effective drainage have also been highlighted. Similarly the installation of rain gauges at potentially high risk sites and the need to bring NADICS into the management loop with respect to route advice have also been recognised.

5.2 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 5.1 presents a flowchart of the work to be undertaken.

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.

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.

It is also particularly important to note that the site-specific assessment will not be a 'drive-by' 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.

Figure 5.1 - Management and mitigation options within Study 1, Part 2.

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.

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 traffic ⁶ or traffic diversion.

In the case of short-term to medium-term reaction to such occurrences, then this 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 might 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 close consultation with both the Geotechnical Engineering Office in Hong Kong, which has 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. These might include debris shelters, similar to the one illustrated in Figure 5.2, and fences (see Figure 5.3).

Figure 5.2 - Stone shelter on A890 northeast of Stromeferry.

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 4 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.

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.

Figure 5.3 - Flexible catch fence.

Clearly 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.

REFERENCES

Anon. 1989. The climate of Scotland - some facts and figures. London: The Stationery Office.

Anon. 1999. Flood Estimation Handbook. Wallingford: Institute of Hydrology. www.nwl.ac.uk/ih.

Design Manual for Roads and Bridges. London: The Stationery Office. DMRB 4.1.3, HA41/03 Maintenance of Highway Geotechnical Assets, November 2004.

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.

Innes, J. L. 1983. Lichenometric dating of debris flow deposits in the Scottish Highlands. Earth Surface Processes and Landforms, 8, 579-588.

Jones, D. K. C. & Lee, E. M. 1994. Landsliding in Great Britain. London: HMSO.

McAdam, D. 1993. Edinburgh - a landscape fashioned by geology. Edinburgh: Scottish Natural Heritage and British Geological Survey.

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.

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, 11 ^th International Conference on Landslides, Norway, 1-10 September 2005.

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.

Winchester, S. 1985. Outposts: Journeys to the Surviving Relics of the British Empire. London: Penguin.

Winter, M. G., Macgregor, F. & Shackman, L. (Editors). 2005. Scottish Road Network Landslide Study. Edinburgh: Scottish Executive.

« Previous | Contents | Next »