Chapter 2 Mass-movement GCR sites in Precambrian and Cambrian rocks

Introduction to the mass movements in the older mountain areas of Great Britain

D. Jarman

The older mountain ranges of Britain — the Scottish Highlands, the Southern Uplands, the Lake District, and the northern half of Wales (Figure 2.1) — have long been prized for both their exceptional landscape value and their scientific interest. They were fashioned during the Caledonian Orogeny 480–390 million years ago, mainly in metamorphic rocks of Precambrian, Cambrian, Ordovician, and Silurian age, but with some contemporaneous igneous intrusions. Mass movements in these ranges differ considerably in character, cause, mechanism, and geomorphological effect from those on the Devonian, Carboniferous, Mesozoic and Cenozoic cliffs, escarpments and valley slopes described in following chapters of the present volume.

This chapter focuses on deep-seated mass movements in bedrock (rock slope failure' — RSF). Mass movement in superficial deposits and mass wasting of rockfaces, both active and relict, is of course prevalent in many parts of these mountains, and may become more significant again as climate change favours more extreme events. However manifestations such as screes, debris cones, solifluction terraces, and rock glaciers are well represented by sites in the Quaternary volumes of the GCR series, Quaternary of Scotland (Gordon and Sutherland, 1993), Quaternary of Wales (Campbell and Bowen, 1989) and Quaternary of Northern England (Huddart and Glasser, 2002).

Paraglacial rock slope failure

Rock slope failure is principally a 'paraglacial' phenomenon in the British mountains. In other words it is not a direct product of glacial or glacio-fluvial activity, but is a geomorphic response to glaciation and deglaciation, a process of stress-release and re-equilibration (Ballantyne, 2002a). Nearly all of the recorded rock slope failures are on the flanks of glacial troughs (including those submerged as fjords), or in the corries/cwms which feed them. Non-paraglacial failures can also be found on sea cliffs (e.g. Faraid Head, Durness; the Banffshire coast), and with instructive rarity above fluvial gorges (e.g. Arnisdale by Loch Hourn (Figure 2.2), Craig Maskeldie in the Angus glens [NO 386 795]).

A special case arises with the Tertiary [Palaeogene] Volcanic Province of the Western Highlands and Inner Hebrides (Figure 2.1). Here numerous extensive mass movements occur where massive basalts overlie less competent strata. They are more comparable in behaviour and in expression with the coastal sites of England, and following the scheme of this volume are dealt with in Chapter 6. Many of them are still active or could easily be re-activated, whereas nearly all of the rock slope failures in the old hard rocks of the mountains are long dormant or 'metastable'. The failure sites selected at the Trotternish Escarpment in Skye (see Chapter 6) also represent the extensive plateau-rim failures formed on Devonian and Carboniferous lavas in the Midland Valley of Scotland (Evans and Hansom, 1998; see also Craig Rossie GCR site report in Stephenson et at. (1999)).

Extent of rock slope failure and state of knowledge

By contrast with mass movement in the sedimentary slopes of Britain, there has been remarkably little research into rock slope failures in the mountain areas. This is partly because of their isolation and low geohazard status, and partly because they are enigmatic features, difficult to address within the canons of either geomorphology or engineering geology.

Scotland

Most is known about the Scottish Highlands. Here, a systematic but unpublished study of aerial photographs yielded 364 extant rock slope failures (Holmes, 1984), and together with incomplete British Geological Survey mapping and other sources gave Ballantyne (1986a) a total population of 495 in the mainland Highlands. Detailed field survey indicates that in some areas this might be augmented by 20% (Jarman, 2003a). Three unpublished theses provide information on collections of sites including some valuable geotechnical analyses (Watters, 1972 — 20 cases; Holmes, 1984 — 27 cases; Fenton, 1991 — 44 cases). A few individual sites have been recorded in the literature, but these have generally not been of seminal status (e.g. Beinn nan Cnaimhseag; Sellier and Lawson, 1998).

In the Southern Uplands, rock slope failure is sparse and low-key, with isolated cases, for example in the Galloway mountains (Cornish, 1981), north-east of Sanqhuar, and in the Cheviots (W. Mitchell, University of Durham, pers. comm.). Ballantyne (1986a) recorded 24 sites.

England

Systematic investigation of rock slope failure in the Lake District is beginning to emerge, with around 50 sites affecting at least 5.5 km² in total already identified (Wilson et al., 2004). Some are quite substantial, with one at Robinson–Hindscarth (Buttermere) being large (1.7 km²) and significant in UK terms (Wilson and Smith, 2006). Others display bold features, notably the antiscarps on Kirkfell, Wasdale (Wilson, 2005) (Figure 2.3) and the Fairfield complex (see (Figure 2.10)).

Wales

In northern and mid Wales, there is no systematic survey of paraglacial mountain rock slope failure, although active failure in coastal old hard-rock exposures is of continuing interest (cf. Nichol, 2002). A few individual sites have been reported in Snowdonia (e.g. Curry et al., 2001; Rose, 2001), and some in the Berwyns such as that damming Llyn Moelfre [SJ 180 285] (Hutchinson, unpublished data). One substantial site selected here at Cwm-du (see Chapter 3) represents behaviour in weakly indurated Silurian metasediments, but is most noted for the uncertainty surrounding its origins. Another site at Tal-y-llyn near Cader Idris has been thoroughly investigated (Hutchinson and Millar, 2001, fig. 48) and is notable as the largest landslide dam in Britain. A kilometre of glacial trough wall cut in Ordovician metasediments collapsed, with 50 x 10⁶ m³ of debris impounding the lake of Tal-y-llyn, once 2.5 km and now 1.6 km long. A substantial extension to the failure scar has become arrested after short travel. This site is of international significance (Nichol, 2002).

GCR site selection

When the original shortlisting of mass-movement sites was made in 1982, only three sites were put forward in the mainland Highlands, reflecting the dearth of published investigations. Of these, two were major discoveries arising out of unpublished PhD theses — Glen Pean (de Freitas and Watters, 1973) and Beinn Fhada (Holmes and Jarvis, 1985). The third Scottish site at Coire Gabhail (see Chapter 4) was famous as the landslide-blocked Lost Valley of Glencoe; it is the only known rock slope failure on high-strength Devonian lavas in the Highlands.

Three GCR sites selected for the Quaternary of Scotland GCR Block (Gordon and Sutherland, 1993) are also relevant to the subject of the present volume. One is in Dairadian quartzite and two are in Precambrian Torridonian sandstone: Beinn Shiantaidh on Jura (Gordon and Mactaggart, 1997) and Baosbheinn in Torridon are now regarded as more probably rock slope failures rather than rock glacier and protalus rampart cases respectively, while Beinn Alligin has been described in the present volume (as well as in Gordon and Sutherland, 1993), as cosmogenic dating (Ballantyne and Stone, 2004) has largely resolved the controversy over its mode of emplacement (although it still does not fully merit the designation of 'sturzstrom).

Systematic characterization of rock slope failures in several parts of the Highlands (Jarman, 2003a,b; Hall and Jarman, 2004), and of all 140 larger failures in the Highlands (Jarman, 2006) has demonstrated their great diversity. It has also underlined the previously overlooked importance of their contribution to erosion and landscape shaping over Quaternary times. It became clear that rock slope failure in the older mountain areas could not adequately be represented by just six sites, of which three were in lithologies where failure is exceptional and relatively small-scale. Of the three in metasedimentary rocks, Cwm-du (Chapter 3) has been studied mainly as a quasi-glacial deposit; and while Glen Pean and Beinn Fhada are two of the largest and most impressive failures in Britain, they are of rather similar character and setting, and both occur within similar geological contexts in the North-west Highlands.

In reviewing the Highland rock slope failure sites in 2003, to ensure that the GCR encompassed the full spectrum of characteristics (including geological context, type of failure, landshaping effects), eight additional GCR sites were proposed and are described in the present chapter — Beinn Alligin, Ben Hee, Benvane, Cant Dubh, The Cobbler, Druim Shionnach, Glen Ample and Sgurr na Ciste Duibhe. The Glen Pean site was recommended for deletion (where the original interpretation is now found to be implausible, c.f. Jarman and Ballantyne, 2002); this type of rock slope failure is better represented by Beinn Fhada.

This introduction sets out the general context for understanding the diversity and significance of rock slope failure in the older mountain areas, starting with the main characteristics that the selected sites seek to represent.

Representing the diversity of geology and structure

Lithological controls

The vast majority of rock slope failures occur in the metamorphic lithologies (Ballantyne, 1986a). This is unsurprising given that most of the older mountain ranges are constructed from them. But significant failure can occur in every lithology, including Torridonian sandstone (e.g. Beinn Alligin), volcanic lavas (e.g. Coire Gabhail, Chapter 4), and granite (e.g. Lundie, [NH 164 114]. A notable complex on granite affects 3 km on both sides of Strath Nethy, beside Cairn Gorm (Hall, 2003). This complex falls within the Cairngorms GCR site, selected for the Quaternary of Scotland GCR Block (Gordon and Sutherland, 1993). Current investigations at this site by the British Geological Survey suggest an unusual combination of glacial, periglacial and paraglacial activity.

Metamorphic rocks in the British mountains range in age from dominantly Precambrian in the Scottish Highlands and Islands, to mainly Ordovician and Silurian in the Southern Uplands, Lake District and North Wales. In the Highlands, the Moine and Dalradian Supergroup rocks are mainly composed of metamorphosed and deformed sandstones, siltstones and mud-stones. Here the term 'schist' has been applied commonly to the more indurated Highland rocks, with the dominant psammitic (i.e. sandy) variants being termed 'quartz schists'. In the Palaeozoic ranges the metasedimentary rocks are often less indurated, with the generic term 'slate' including friable greywackes (see Cwm-du GCR site report, Chapter 3). But rock slope failure occurs across all metamorphic types and grades, including those of igneous origin such as the ancient Lewisian gneisses and the Borrowdale volcanic rocks. It tends to be more widespread in slaty and interbedded strata, where mica-rich cleavage and foliation surfaces facilitate sliding, but can equally well operate in blocky to massive and relatively uniform psammite terrains such as the central Grampian Highlands (Hall and Jarman, 2004).

Structural controls

Structurally, the metamorphic rocks are more prone to develop deep-seated failure planes, by virtue of profound tectonic activity during the Caledonian Orogeny and (in the North-west Highlands) earlier orogenies. As well as the foliation (schistosity) surface, three or more joint-sets are commonly present (Watters, 1972) (Figure 2.4), so that potential sliding surfaces, depending on their pervasiveness, can be available on most slope aspects. By contrast, granite is generally more sparsely jointed, and it 'springs' on shallow fracture surfaces that develop parallel to the present or original slope. Torridonian sandstone also tends to fail at joint-block scale, although slices of cliff have collapsed on near-vertical joints, and mass creep has occurred on gently dipping bedding planes (e.g. Beinn Bhàn, Applecross [NG 800 450]).

Mountain-building processes have left the metasedimentary rocks inclined at all angles from sub-horizontal to sub-vertical, and in all scales and intensities of folding. The textbook ideal for large-scale sliding is a smooth surface inclined close to the peak or residual friction angle (Hoek and Bray, 1981), typically 20°–40° for schists. Any gentler, and friction prevents sliding, any steeper, and the surface becomes less likely to have been undercut by glacial trough steepening; ultimately it becomes a self-supporting wall. However, rock slope failure occurs freely in rocks inclined at all angles and showing every degree of folding and contortion, if not on the foliation or bedding surface then on joint-sets that cut through the contortions, and if not by sliding alone then by creep, sag, buckle or topple, or any combination.

The sites selected here show that while geological controls can be direct and obvious (e.g. Ben Hee), rock slope failure can develop in a wide range of contexts, and in some cases without an obvious relationship to any observable structures (e.g. Beinn Fhada, The Cobbler). The relatively straightforward analyses and predictions that can be made for failures in the regular sedimentary strata of Britain become more elusive in the older mountain areas.

Representing the diversity of rock slope failure types

The general introduction to this volume follows the classification of Hutchinson (1988), but observes that any attempt to classify mass movements is unsatisfactory because firstly they are on a continuum, and secondly most are complex, embodying several modes of failure. This is especially true of the older mountain areas. Characterization of rock slope failures in the older mountain areas like the Scottish Highlands ((Table 2.1); (Figure 2.5)) has adapted the Hutchinson schema to reflect prevailing modes there, with five broad categories spanning the continuum (Jarman, 2006):

• compressional deformation
• extensional deformation
• arrested translational sliding
• sub-cataclasmic slide/collapse
• cataclasmic slide/collapse

Slope deformation

The first two categories (compressional and extensional deformation) cover slope deformations where the lateral margins are diffuse, and downslope movement is limited. They tend to be extensive, and account for 64% of the larger (> 0.25 km²) Highland rock slope failures (Jarman, 2006). Extensional deformations display creep, sagging, or bulging usually with some headscarp development and tension features such as furrows, fissures and trenches. They include the 'sackungen' first described in the Alps. Compressional deformations are less well attested, but correspond to Hutchinson's 'rebound' category. They are virtually in situ, lacking in tension features, and evidenced mainly by disrupted drainage (dry slope with basal springline). The distinction between these types is subtle, and often best attested by the character of their antiscarp arrays (see below). Many slope deformations combine both compressional and extensional indications.

(Table 2.1) Characteristic types of large rock slope failures (RSFs) in the Scottish Highlands and Lake District. Adapted from Jarman (2006) and Wilson et al. (2004). See (Figure 2.5) for explanation of terms.

Varying degrees of deformation are represented at Beinn Fhada, Benvane, Sgurr na Ciste Duibhe, and Glen Ample. These are very large sites, affecting at least a square kilometre of valley-side; Beinn Fhada and Ben Our (Glen Ample) are two of the largest rock slope failures in the older mountain areas, and are of exceptional significance for their clarity of surface expression and scope for research into slope deformation processes.

Arrested translational sliding

Arrested translational sliding is the prevalent mode of failure in the mountain areas, accounting for 60% of all rock slope failures in the Southwest Highlands (Table 2.2), although only 25% of the larger sites across the Highlands. Here a well-defined mass has failed and travelled some distance downslope before becoming arrested in a more-or-less coherent state. There is usually a well-defined source cavity, which may be wedge or armchair-shaped. The failed mass may have antiscarps, open fissures, detached megablocks, and basal debris-lobes or block piles. The reasons for arrestment are poorly understood, but probably reflect inadequate lubrication, rough or corrugated failure planes, and lack of vertical relief; such arrested slides seem less apparent in higher mountain ranges of the world.

Arrested slides are represented at The Cobbler (very short travel from a headscarp but much dislocated), Benvane (long travel from a classic source bowl, low-key lobe) and Ben Hee (short travel from a corrie headwall). Good examples of armchair-source slides are at An Sornach in Glen Affric (Holmes, 1984; Jarman, 2003c), Glen Fintaig in Lochaber (Watters, 1972), and Beinn Tulaichean at Balquhidder (Jarman, 2003a) (Figure 2.6), while Tullich Hill above Glen Douglas (Luss Hill) is an instructive wedge-source complex (Jarman, 2003d) (Figure 2.7).

A hybrid category between the in-situ compressional deformation and the translational slide is common in the Cluanie area of the Western Highlands (Jarman, 2003b), and elsewhere on steeply inclined high-competence lithologies, such as the Mamores and Grey Corries. It is represented by Druim Shionnach.

Cataclasmic and sub-cataclasmic failure

Cataclasmic rock slope failure is so termed because it is not literally 'catastrophic', but describes where the failed mass has disintegrated and travelled to the slope foot or beyond (Jarman, 2002, 2006). Fully cataclasmic events are very uncommon in Britain, and are here represented by Beinn Alligin, and by Coire Gabhail (Chapter 4). The term 'sturzstrom' has occasionally been borrowed from high mountain areas to indicate rockslides with excessive runout, but even Beinn Alligin scarcely warrants this designation, lacking the vertical relief and the jump-off bench normally required.

Sub-cataclasmic failure is commoner, where although substantially disintegrated, the debris lobe has barely reached the slope foot, and remains only conditionally stable. Carn Dubh (Ben Gulabin) represents this mode with two small debris-tongues. Other tongues analysed include Mam na Cloiche Airde (Watters, 1972) and Gleann na Guiserein (Bennett and Langridge, 1990; Ballantyne, 1992), both in Knoydart, Mullach Fraoch-choire in Glen Affric (Holmes, 1984) (Figure 2.8), and Burtness Comb in the Lake District (Clark and Wilson, 2004). Moelwyn Mawr in Snowdonia (Rose, 2001) may well be similar.

A variant of sub-cataclasmic failure is the crag collapse leaving a coarse blocky debris-pile on the slope. In the South-west Highlands, 50% of sites display an element of (sub-)cataclasmic behaviour, the massive schistose grit lithology being more conducive to it than in areas of slabby metasediments. A fine example is on Beinn an Lochain North (Watters, 1972; Holmes, 1984) (Figure 2.9).

Complex failures

Many rock slope failures demonstrate several modes of failure, as at Sgurr na Ciste Duibhe. Hell's Glen (Holmes, 1984) dramatically illustrates a progression from extensional deformation, producing a 15 m-deep antiscarp trench slanting across the midslope, with large translated masses breaking away into detached megablocks up to 60 m high, and some cataclasmic collapse debris reaching the glen floor.

(Table 2.2) Rock slope failure (RSF) incidence, character, landshaping effect, and association with breaching in the Southern Highlands and Kintail area (including clusters 1, 5 and 7 in (Figure 2.13)). Updated from Jarman (2003a,b). Note: sites may be in more than one character or landshaping category. See (Figure 2.15) and (Figure 2.18).

Representing the landshaping effects of rock slope failure

Paraglacial rock slope failure has played a significant, if generally unremarked, role in shaping the present mountain topography. Its geomorphological impacts are . numerous (Jarman, 2003a,b; (Table 2.2)), and may appear as isolated incidents (e.g. Carn Dubh (Ben Gulabin)) or, where its occurrence is dense, as a more generic contribution to Quaternary landscape evolution.

Arêtes and horns

The most striking effects of rock slope failure can be seen where summit ridges have been narrowed and incidented. This has been recognized in attributing mites in the ranges around Ben Nevis to rock slope failure (Bailey and Maufe, 1916), and in interpreting the horn of Streap near Glenfinnan as a summit sliced by a slide scar (Watters, 1972). The Cobbler has been selected to display these summit ridge effects, while at Sgurr na Ciste Duibhe the summit mass has been lowered bodily by about 10 m, leaving a fretted arête. Good examples in the Lake District are the horns of Helm Crag (Grasmere [NY 325 090], and the Cofa Pike arête on Fairfield ([NY 355 129]; (Figure 2.10)).

Corrie development

It has been speculated that rock slope failure may play a part in 'seeding' corrie development (Clough, 1897; Peacock et al., 1992; Turnbull and Davies, 2006; see Cwm-du GCR site report, Chapter 3). Certainly in the minority of cases where failure occurs within a corrie, it is contributing to corrie enlargement and ultimate destruction, whether by reduction of the enclosing arms (e.g. The Cobbler) or breaking through the headwall (e.g. Ben Hee).

Valley widening and ridge reduction

Most rock slope failures contribute to the general processes of valley widening and spur truncation (Table 2.2), both directly and indirectly, by providing weakened slopes and debris masses ready for erosion and evacuation by glaciers in the next cycle (Bentley and Dugmore, 1998). By the same token, rock slope failures encroach into ridges and pre-glacial plateau remnants, with source scarps and fractures often daylighting by as much as 50–100 m behind the crest or rim (e.g. A Chaoirnich in the Central Grampians) (Jarman, 2004a). Such effects are seen particularly well at Beinn Fhada, and in lower-relief contexts in Glen Ample and at Benvane.

Where mountain ranges are dissected by breaching, the isolated mountains become more vulnerable to concerted attrition by rock slope failure, until their crests are lowered sufficiently to permit glacial over-riding and reduction to subdued relief. Examples of mountains at various stages of isolation by breaching and where rock slope failure is encroaching on several fronts include Beinn an Lochain in Argyll (Watters, 1972), Ciste Dhubh in Kintail (Jarman, 2003b), An Dùn in the Gaick Pass (Jarman, 2004a), and Na Gruagaichean in the Mamores (Figure 2.11).

Antiscarps

Antiscarps are one of the most conspicuous indicators of rock slope failure in the mountain landscape. These uphill-facing scarplets occur both in translational slides, as the moving mass begins to disaggregate (The Cobbler), and in slope deformations, where they may extend laterally for hundreds of metres (e.g. the exceptionally fine array on Beinn Fhada). They may develop intricate lattices (Benvane) and platy structures (Glen Ample). In the Lake District, a notable case occurs on Kirk Fell (Wilson, 2005; (Figure 2.3)). Antiscarps typically reach up to a few metres high, and where only decimetric may be hard to see on the ground. In Glen Shiel and Kintail, and a few other isolated cases (including Kirkstone Pass in the Lake District — Wilson et al., 2004), they can attain 5–10 m, with Beinn Fhada having the highest classic midslope antiscarps. Druim Shionnach is exceptional, with a 14 m antiscarp, but this may have developed as part of a graben structure opposite the source scarp.

Research is needed to understand why and how antiscarps develop. Their incidence and character may illuminate the extent to which deformation is either extensional in unsupported steep slopes, or compressional, driven by differential glacio-isostatic rebound stresses between valley floor and summit ridge. The factors governing antiscarp height are especially unclear. It could be an indicator of rebound stress intensity, but will also depend partly on rock-type and strength, and on length of exposure to weathering. Extant heights may be much reduced from their original levels, partly by crest degradation but mainly by trench infilling. It is rare to find datable bedrock exposures on an antiscarp, or a fault-like fracture in the few exposed cross-sections. Some antiscarp arrays may have been subdued by the Loch Lomond Stadial (LLS) glaciers, or re-emerged after them (Beinn Fhada; Beinn Odhar Bheag, Glenfinnan, [NM 850 775]).

A special case of apparent antiscarp occurs where the failed mass includes the summit ridge, and the fracture plane 'daylights' behind the crest. This creates the phenomenon known as a 'split ridge', common in the Alps as a 'doppelgrat' (Crosta, 1996). Sgurr na Ciste Duibhe displays aspects of this tendency, but a remarkably clear example extending intermittently for over 1.5 km and attaining a source scarp height of 10–15 m has recently been recognized on Aonach Sgoilte in Knoydart (Figure 2.12).

The spatial distribution and root causes of rock slope failure

Perhaps the most purling aspect of the rock slope failure phenomenon is its irregular spatial distribution. Even on mountain ranges most conducive to it, failure is far from endemic: it may be abundant, sparse, or absent on valley-sides of similar scale, steepness and geologi'cal character. Where rock slope failure is common, it is seldom obvious why one section of valley slope should have failed rather than adjacent sections; local factors are clearly important. Given that most failures appear to date from early or mid-Holocene times, this suggests that at periods of maximum rebound and climatic stress, extensive areas of valley-side were close to the limits of stability.

Association with ice limits

Two studies have addressed the wider distribution of rock slope failure, though regrettably neither has been published to expose the debate. Holmes (1984) found that in nearly all cases, some force, augmenting gravity, was required to mobilize translational sliding. He sought to demonstrate a close correlation between failure locations and the upper limits of the Loch Lomond Stadial (LLS) glaciers, as identified by Sissons and his school. He envisaged that excessive meltwater pressures at deglaciation would trigger a spate of failures. Unfortunately, where the LLS was a valley-full glaciation with nunatak ridges exposed, there is an inevitable co-incidence between steep valley-sides where rock slope failure is most likely to occur, and the upper limits of the glaciers. Where the LLS achieved near-icecap coverage at its centres, as is now thought likely (Golledge and Hubbard, 2005), such a correlation obviously cannot occur. In fact, failure can be found at all levels within any glaciated valley system, both within and well beyond the LLS outer limits. Furthermore, this model does not explain adequately the absence of rock slope failure from apparently suitable terrain well within the LLS outer limits such as Ardgour (see (Figure 2.13)). Nor does an engineering model developed for discrete, compact translational block-sliding account for the large proportion of diffuse and laterally extensive slope deformations. Finally, the model cannot account for the rock slope failures that have been dated (directly or inferentially) to several thousand years after deglaciation.

Association with neotectonic activity

Fenton (1991, 1992) took the spatial incidence of rock slope failure in the North-west Highlands as an indicator of significant seismic movements after deglaciation. This built upon a model that assumed that plate-tectonic stresses suppressed by the weight of the icecap would undergo a period of accelerated 'catching-up', triggering earthquakes along ancient faultlines (Davenport et al., 1989). Unfortunately, while many failures are located on faults, that is mostly because valleys have been eroded along them; conversely, many failures are as distant from main faults as is possible in the Highlands (Jarman, 2003a). Furthermore, the derivation of earthquake magnitudes from rock slope failure size and proximity was based on evidence from contemporary earthquakes in tectonically active mountain ranges, and therefore subject to interpretation in its wider application.

It is rather more likely that a phase of modestly elevated seismic activity followed deglaciation, as rebound stresses varied in relation to the local thickness of ice cover and intensity of erosion. The concept of blocky isostatic recovery was first developed at Glen Roy (Sissons and Cornish, 1982) and may retain some validity (Stewart et al., 2000), although firm evidence for the 'neotectonic fault scarp' displacements found by Fenton is still lacking (cf. Firth and Stewart, 2000). Work in alpine ranges suggests that 'topographic amplification' of seismic shocks can enhance their shaking effects at crests and peaks (Ashford and Sitar, 1995). This might be explored in cases such as Sgurr na Ciste Duibhe and The Cobbler.

Association with glacial breaching

A possible explanation currently being explored views rock slope failure as a response to locally exceptional slope stresses, on top of those generically induced by glaciation/deglaciation. A spatial association can be observed in every failure cluster with glacial breaches of main and secondary watersheds (Jarman, 2003a,b; 2006; (Table 2.1) and (Table 2.2). The Central Grampian cluster demonstrates a near-100% association between rock slope failure and the breaches of Loch Ericht, Garry, and Gaick (Hall and Jarman, 2004) (see (Figure 2.14)). Conversely, failure is sparse or absent in mature troughs and glens that have long adapted to efficient ice discharge, and on plateau rims away from breaches (see (Figure 2.15)). The inferred cause of rock slope failure is that breaching has involved concentrated glacial erosion of bedrock either in the breach or within a few kilometres downstream (by augmentation of ice catchment area), or in rejuvenated tributary valleys. Bulk erosion by glaciers streaming through such breaches could be an effective way of over-steepening slopes, setting up rock-mass stresses, and day-lighting fallible discontinuities (see (Figure 2.16)). On deglaciation, the rebound stresses are significantly greater than those from generic glacio-isostatic recovery alone, provoking failure in the most susceptible locations.

The extent of glacial breaching was recognized by Linton (1949), and its importance in the dissection and reduction of the mountain areas by Haynes (1977a). It can only occur when the iceshed becomes offset from the watershed. The main watersheds of the Highlands have shifted east and north by 5–30 km over the Quaternary Period along much of their length as a result of glacial breaching (Jarman, 2006; see (Figure 2.13)). The spatial association of paraglacial rock slope failure with this dense incidence of breaching is therefore of considerable interest for future research, although it must be emphasized that many failures are not in close proximity to obvious breaches, and that there is an inevitable degree of self-correlation between rock slope failure and steep, narrow passes (Figure 2.17).

Rock slope failure clusters

Most rock slope failures in the Highlands fall within the main clusters identified for the larger sites (Jarman, 2006; (Figure 2.13)):

It may be that these clusters indicate where glacial breaching and erosion have been most active in Devensian times, and thus where shifts in ice centres and/or dispersal patterns have been most marked. Work is required to scope and calibrate deglaciation slope stresses of all kinds, and to refine icecap models in light of rock slope failure information. The intensity of dissection by breaching increases dramatically from east to west across the Highlands, as in microcosm in the Lake District and North Wales (Clayton, 1974, after notes by D. Linton; Haynes, 1977a, 1995). Failure clusters may indicate where dissection is further intensifying, and where the 'Clayton Zones' of dissection intensity are migrating eastwards (see Gordon and Sutherland, 1993, fig. 2.1).

The sites selected here mostly fall within the two densest rock slope failure concentrations first recognized by Holmes (1984) — the Southern Highlands, and Affric–Kintail (Figure 2.15) and (Figure 2.18). A cluster around Glen Roy is touched upon under that GCR site report in the Quaternary of Scotland GCR volume (Gordon and Sutherland, 1993), an important site, not least because the sequence of Parallel Roads enables close dating of the different rock slope failures (Fenton, 1991; Peacock and May, 1993). Two sites in Glen Shiel are notable in interpreting the major breach west from Glen Cluanie: Sgurr na Ciste Duibhe is immediately above the 'canyon of adjustment' on one of the highest steep slopes in Britain, while Druim Shionnach is in one of the side troughs rejuvenated by it. Beinn Fhada is just downstream of the next breach north, and on the inferred pre-glacial watershed. By contrast, The Cobbler is in the heart of an area intensely dissected by primary and secondary breaching, with failure ramifying into the rejuvenated side valleys. The Ben Hee rock slope failures may be a response to local northwards breaching. Benvane and Glen Ample are close to immature breaches of a major secondary divide, in lower relief near the Loch Lomond Stadial outer margin.

Some of the major Lake District rock slope failures are adjacent to breaches, such as Honister, Grisedale (Figure 2.10), and Kirkstone, or above troughs possibly augmented by transfluent ice such as at Wasdale. Conversely, failure is almost absent from more mature valleys such as Langdale, Borrowdale and Ullswater. In North Wales, the Tal-y-llyn trough (into which the Graig Goch landslide descended) appears to have been augmented by ice breaching across the main divide.

It is probable that the actual triggers for individual rock slope failures were a combination of factors such as elevated water pressure, seismic shaking, freeze-thaw, and progressive failure (cf. Ballantyne, 2002a). These factors have probably all applied in varying intensities and combinations. The sites selected here represent locations where each factor has been invoked, and where further research is merited. Some of the sites described here are simple, others complex in mode and probable triggering mechanism. Some of the mass-movement features are relatively recent, dated as mid-late

Holocene (Beinn Coire Gabhail (Chaper 4)); others probably occurred around final deglaciation or even earlier, or have evolved in stages (Sgurr na Ciste Duibhe).

Scale and international comparisons

Rock slope failure sizes display a typically skewed size distribution ((Table 2.1) and (Table 2.2)), with many small cases of purely local impact. The depth to which failure extends is seldom known, except in rare cases where a cavity has been largely evacuated, but is generally assumed to be in the order of tens of metres for smaller rock slope failures, rising to 100–150 m for larger cases (Holmes, 1984; Fenton, 1991). Inferred volumes reach 10 x 10⁶ m³ for translational slides, and may exceed 100 x 10⁶ m³ for the very largest deformations (Table 2.3).

The scale of rock slope failures in the mountain areas of Britain is of course limited by virtue of the available relief being less than in ranges such as the Scandes and the Alps. Even so, the largest cataclasmic rock slope failure in Britain (Graig Goch in North Wales) is of the same order of magnitude as the largest yet reported in Scandinavia at Kärkevagge (Jarman, 2002); albeit at approximately 50 x 10⁶ m³ these are an order of magnitude smaller than the ten largest in the Alps, let alone the extraordinary Flims failure which reaches 12 000 x 10⁶ m³ (Abele, 1974).

With slope deformations, the typical scale in the Alps involves relief of 1000 m, a depth of 100 m, and a volume of 100 x 10⁶ m³ (Brückl and Parotidis, 2005). It is interesting that such dimensions are approached by the largest or highest rock slope failures in the Highlands (Table 2.3).

In a European context rock slope failures such as Beinn Fhada and Glen Pean (de Freitas and Watters, 1973) are of significant scale and bear instructive comparison as steep mountain deformations. Sgurr na Ciste Duibhe is the highest and steepest failed slope in Britain, and since it displays some progression from in-situ deformation to sliding, offers a valuable benchmark for international studies of critical stability thresholds. By contrast, Ben Our (Glen Ample) is exceptionally extensive in very low relief, and no international comparators have yet been found for it. In terms of exhibiting the direct impact rock slope failure can have in shaping mountain scenery, The Cobbler can hold its own with any international comparison.

Incidence and impact of rock slope failure during the Quaternary Period

Clearly this paraglacial contribution to erosion in mountain areas can be very substantial locally. Where rock slope failures encroach significantly into corries, ridges or pre-glacial land surfaces, the local rate of scarp retreat can be vastly greater than by all other means of erosion over a glacial cycle.

With many such glacial-paraglacial cycles over the Quaternary Period, the cumulative impact of rock slope failure will have been very substantial, especially in the most geologically susceptible areas (Evans, 1997; Hall and Jarman, 2004; cf. Trotternish Escarpment GCR site report, Chapter 6). However, the total contribution of paraglacial rock slope failure to the huge volumes of erosion experienced by the higher mountain ranges of Britain is as yet difficult to assess. Extant landslides yielding debris ready for evacuation by the next glaciers have relatively small volumes (typically less than 1 x 10⁶ m³), whereas almost intact slope deformations such as Ben Our (Glen Ample) with much larger volumes may not be much more vulnerable to erosion than the surrounding terrain.

The incidence of rock slope failure may have changed as Quaternary landscape adaptation to ice progressed, and as glaciations increased in severity. It seems likely that paraglacial failure would have been fairly endemic in the relatively weak early Quaternary glacial-interglacial cycles, as fluvial valleys with interlocking spurs became adapted to efficient ice discharge as straightened and deepened troughs. In the great ice-sheet glaciations of mid-Quaternary times, transfluence across watersheds would have initiated the process of glacial breaching, provoking a fresh round of intensive but less widespread failure.

(Table 2.3) Large rock slope failures (RSFs) in the Scottish Highlands for which data are available. After Jarman (2006). Sites are listed from the north, with the Great Glen separating the North-west Highlands from the Grampians. Note the disproportionate number of large RSFs studied north-west of the Great Glen, where foliation (F) is rarely as conducive to sliding as in the Southern Highlands. Most studies are of (sub-)cataclasmic RSFs or slope deformations, rather than conventional arrested slides.

Many of these breaches are now relatively mature. It is therefore likely that the incidence of failure has diminished into the last (Devensian) glaciation. The extant population of rock slope failures may thus be relatively modest, and its spatial distribution probably indicates where glacial erosion and other destabilizing factors were concentrated latterly.

Rock slope failure commonly occurs with a delayed reaction time of hundreds or even several thousands of years after deglaciation. Thus although many failures are within the boundaries of the Loch Lomond Stadial (LLS), this was short-lived (c. 12 900–11 500 years BP) and carried out little fresh erosion. These rock slope failures are probably responding to stresses induced during the Last Glacial Maximum, which peaked approximately 22 000 years BP and deglaciated approximately 15 000 years BP

However, paraglacial responses as delayed as at Beinn Alligin (7000 years after deglaciation) and Coire Gabhail (9000 years after deglaciation) may be exceptional. No large rock slope failure movements are known within the mountain areas during recorded history, with even conspicuous cases such as Glen Kinglass [NN 190 096] of c. 1700 AD (Clough, 1897) only amounting to 70 000 m³ (Holmes, 1984). Progressive failure in one area of Norway is still leading to catastrophic collapses (Bjerrum and JOrstad, 1968) but this is unknown in Britain, the creeping failure in fjord-type cliffs affecting road and rail at Attadale, Loch Carron (NG 914 377; Watters, 1972) perhaps being the nearest approximation.

Significance of rock slope failure in the older mountain areas

Understanding mass movements in lowland Britain is of critical importance for civil engineering, geohazard awareness, and active geomorphological processes such as coastal retreat, but there are different reasons for studying paraglacial rock slope failure in the mountain areas, and for conserving key sites:

• It is a significant if overlooked contributor to glaciated landscape development, especially as an agent of selective linear erosion (notably breaching), where rates of valley incision and widening can be orders of magnitude more rapid than normal.
• It has played a remarkable role in shaping many mountain summits and ridges, and is a potential key component in Earth science and landscape interpretation and geotourism development (cf. Brown, 2003).
• It links with other nature conservation and environmental history interests, in that failed and deformed slopes greatly increase habitat niche availability, ameliorate micro-climate, and thus enhance biodiversity. Because these slopes are typically drier and warmer, and more fertile and sheltered, they have played an important part in the colonization of the mountains by early man and in the survival of subsistence farming and contemporary land utilization in otherwise inhospitable terrain.
• Finally, the spatial distribution of rock slope failure has potential to inform and calibrate efforts to inform the analysis of shifting ice centres and dispersal patterns over the Devensian glacial period, with possible benefits for palaeoclimate change studies.

The sites selected here represent all of these aspects of rock slope failure to varying degrees, but by comparison with well-studied fields such as coastal landslips and glacial deposits, there is considerable scope to augment the site coverage as further failure research proceeds. In particular, sites in Lochaber, the Lake District, and North Wales would reflect local diversity of expression, while no sites have yet been subjected to state-of-the-art geotechnical investigation, or to slope-stress modelling in relation to underlying causes and trigger events.

References