Paraglacial rock slope failure as an agent of glacial trough widening

David Jarman
Geological Society, London, Special Publications, 320, 103-131, 22 July 2009, https://doi.org/10.1144/SP320.8

Abstract

Rock slope failure (RSF) generates the largest single erosional events in the glacial–paraglacial land system, leaving numerous obvious cavities and less obviously weakened valley walls. Its contribution to trough widening in a mountain range has not previously been systematically quantified. Map-based measures of RSF ‘depth of bite’ are applied to five sample areas in the Scottish Highlands, and a comparator area in north Norway, all in metasediments structurally conducive to mass deformation and block sliding. Problems in applying map-based measures include bedrock cavities remaining partially occupied by failed debris or subsequent infill, and multiple planes of reference. The most practical measure is of maximum recess depth on any single contour (DMAX). This is a standardizable single-point indicator of visible impact, not a measure of actual cavity depth, nor an average applying to the whole RSF. In four of the five areas, average DMAX is consistent at 40–45 m. RSF breadth averages 270–600 m over the five areas. RSF affects 9% and 14% of total valley wall length in the two densest RSF areas, rising to 47% and 52% on two specific valley sides. The depth:breadth ratio in areas dominated by slope deformation can be twice that in areas of translational sliding. An evolutionary model of glacial–paraglacial cycling proposes a ‘zone of paraglacial relaxation’ in which RSF is intense in early cycles as fluvial profiles adjust to ice discharge, diminishing with maturity as trough walls become stress-hardened, and reviving in response to neotectonic and glaciological perturbations, notably ice piracy via transfluent breaching. However, a major unknown is the efficacy of glacial exploitation of RSFs: if it takes several cycles to evacuate debris and pare back cavity angles, cumulative RSF impact is lessened. Glacial–paraglacial cycling is a classic positive feedback loop, promoting valley widening beyond the parabolic norm. Preferential exploitation of structure by RSF promotes asymmetrical trough profiles. RSF acts both as a scarp retreat process, and as a slope reduction counterpoint to glacial slope steepening. In landscape evolution, it is a powerful agent in destruction of paleic relief, notably around watersheds that are undergoing breaching by transfluent ice, where trough development and widening is still vigorous.

Paraglacial rock slope failure (RSF) is the most dramatic mode of paraglacial activity, as scoped by Ballantyne (2002). It is widespread, but not endemic, in glaciated mountain ranges. It is an episodic (high-magnitude–low-frequency) process, occurring in susceptible locations where deglaciation stresses are sufficient to rupture bedrock slopes over extents from tens of metres to kilometres. It tends to relax glacially steepened and deepened slopes towards lower angles and conditional stability. It occurs in diverse modes, with landshaping effects including trough widening and scarp retreat (Jarman 2003a, 2006).

RSF is now becoming recognized as a major contributor to erosion in young mountain ranges, which are often glaciated (Hewitt 1988; Korup et al. 2007). In these areas the paraglacial signal is enmeshed with tectonic, fluvial incision and gravitational effects. In older glaciated ranges, such as Britain and Scandinavia, the paraglacial RSF contribution to their gross landscape evolution is clearer but barely acknowledged (Evans 1997; Jarman 2002). Glacial geomorphology has focused on the classical parabolic cross-section (Harbor 1992; Augustinus 1995), and on processes and rates of deepening of the floor (Glasser & Hall 1997; Hebdon et al. 1997), rather than on trough widening. Sugden & John (1976, p. 209) noted that ‘the role of sub-aerial slope processes in modifying and widening troughs has not been examined in any depth’. Even with recognition that ‘slopes are likely to be unstable and prone to collapse, thus modifying the form of the trough by paraglacial reworking’ (Benn & Evans 1998, p. 352), there has been little systematic attempt to quantify it over wider areas, least of all with respect to RSF.

Yet the cavities left by RSFs are often obvious (Fig. 1) and represent the highest-impact single erosional events in the glacial–paraglacial land system. Debris may fall onto the valley glacier (e.g. Gordon et al. 1978; Sigurđsson & Williams 1991; Evans et al. 2006), or be emplaced at or near the slope foot for export by the next glacier, e.g. at Coire Gabhail, Glencoe (Ballantyne 2007b), or remain partly or wholly within the failure cavity. Since many RSFs are not fully evacuated, the term ‘bite’ is adopted to embrace both visible and occupied cavities. It also expresses the erosional dimension to RSF, which is too often considered only as a process or deposit. The practical and conceptual challenges of measuring it are explored in a Scottish Highlands context:

  • Can the ‘depth of bite’ of extant RSFs be defined and consistently measured?

  • Can typical/maximal rates of paraglacial scarp retreat by the RSF process be obtained? And how do they vary across a range of geomorphic contexts?

  • How might RSF incidence have varied and cumulated over the Quaternary? And can a model of RSF contribution to landscape evolution over many glacial–paraglacial cycles be inferred from its present incidence?

  • How does the glacial–paraglacial process actually work? And can it really facilitate wholesale trough widening, beyond the contribution visibly achieved by extant RSFs?

Fig. 1.
Fig. 1.

Streap RSF, Glenfinnan, Western Highlands [NM 944860], a 0.30 km2 arrested translational slide with progression to subcataclasmic failure (located in Cluster 2, Fig. 6). (a) Sliding is on the angled joint-set exposed in the left source scarp. The slide cavity has created bounding arêtes and a horn summit. The pre-failure summit is inferred (from Watters 1972) to have been some 10 m higher and a more rounded dome. The arrested failed mass impounds a lochan, and has a prominent extruded pinnacle (point 750). DMAX (see Fig. 5) is 155 m, one of the largest recorded in the Highlands (see Table 2). View from SW, photograph by Hamish Johnston. (b) Long section through the RSF derived partly from Watters (1972).

Paraglacial RSF types, ages and geotechnics in the Scottish Highlands

A simple RSF typology (Jarman 2006) distinguishes: (a) cataclasmic and (b) subcataclasmic failures, where the debris has reached the slopefoot or lower slope, thus fully evacuating the cavity; (c) arrested translational slides, where the slipped mass still partly occupies a reasonably discernible cavity; and (d) extensional and (e) compressional slope deformations characterized by creep features and antiscarp arrays, where margins are diffuse and no readily definable cavity may exist. Individual cases may display compound character, with evolution of mode across a slope or by nested subsequent events. Cavities may be rectilinear (armchair), acute or obtuse wedges (often multiple), or planar slices. RSF locations range from fjord flanks at sea level to high summits at 1200 m asl (m above sea level). Many weak or modified candidate sites are not confirmable without geotechnical investigation, and remain probable/possible.

Very few RSFs in the Scottish Highlands have been dated, and only the small minority of (sub)cataclasmic cases with fragmented debris lend themselves to it (Ballantyne 2007a, b). It is assumed for this study that all extant Highland RSFs are paraglacial, with many occurring at or soon after deglaciation, and a diminishing tail (Cruden & Hu 1993) by progressive failure or delayed reaction. A handful of cases appear to be triggered by (glaci-)fluvial incision; no substantial cases are known within the last millennium. RSF interpretation is complicated by the minimally erosive Loch Lomond (Younger Dryas) Stadial. Although the majority of extant RSFs are within Loch Lomond Stadial limits, all are arguably responding primarily to slope stresses engendered by the Last Glacial Maximum and its deglaciation. Present RSF incidence thus excludes a lost population of immediately post-maximum RSFs exported by the Loch Lomond glaciers. Some of their cavities may survive as ‘debris-free scarps’ inventorized by Holmes (1984) but these are excluded from this study.

Most Highland RSFs are in the prevalent Neoproterozoic metasediments, with varying degrees of tectonization yielding extensive through-going discontinuities; RSF is rare on the granite intrusions. This paper does not consider the geotechnics of the sample areas or case studies (see Jarman 2007a and associated site reports, with photographs of the full range of RSF types; see also Watters 1972; Holmes 1984). Generally, structure and lithology in the sample areas are conducive to RSF, but actual incidence and absence respond to differences in joints, faults, foliation surfaces, dip, aspect, incidence of fallible rock units, etc. Likewise, whether failure is triggered by periglacial, hydrological or seismic events is less important than slope preconditioning by glacial erosion.

Methodology for measuring RSF ‘bite’

The key parameters required to quantify an RSF cavity are its depth and breadth (Fig. 2). This is straightforward in homogenous terrain such as Icelandic basalt-plateau rims (Bentley & Dugmore 1998), but is more difficult in complex Scottish geology. Breadth is usually evident, although slope deformations may have diffuse margins. Depth of bedrock bite is unclear in the majority of Highland RSFs, which are not fully evacuated and/or have subsequent fill (talus, solifluction). Cataclasmic cases with rockwall cavities such as Beinn Alligin (Ballantyne 2007a) are exceptionally rare. Further complications arise with irregular source geometries such as multiple wedges (e.g. Fig. 3), or where source scarps are degraded and indefinite, or with compound sites of different modes or ages of failure.

Fig. 2.
Fig. 2.

Terminology for scarp retreat by paraglacial RSF. ‘Cavity breadth’ is used rather than ‘width’ since it is orthogonal to ‘trough width’ and must not be confused with valley widening. ‘Depth of bite’ can also be confused with ‘trough depth’, but no suitable alternative term exists. The depth of bite depends on the planes of reference adopted (see Fig. 4). Further complications arise where RSF daylights behind a ridge crest, since the bite is both widening the valley laterally and lowering the ridge vertically.

Fig. 3.
Fig. 3.

Beinn Dubh [NS338948] in Glen Luss (for location see Figs 7 and 14). (a) View downvalley from the west. Three main translational slide wedges have developed within a large (0.5 km2) deformation. The several bedrock cavities (marked with dots) remain partly obscured by non-evacuated debris and subsequent infill, so that a representative depth measure is difficult to derive. The ridge crest has been lowered by approximately10 m for 350 m between the bars. (b) The depth measures defined in Figure 5 give DMAX of 50 m in the main cavity, and DAVE of 20 m across the site. Two of the wedges break the crestline, daylighting downslope towards Loch Lomond with DBAY of 180 m and 140 m. This incipience creates a weakness likely to be exploited by glacial parafluence, breaching the ridge and isolating its lower end as a ‘pap’, a common form in the Highlands. Vertical air photograph © RCAHMS.

This occupied-cavity problem vitiates any attempts at accurate measurement of topographic depth parameters, whether by field survey, photogrammetry or DEM (digital elevation model). Calculating scarp retreat rates from debris volumes (e.g. Holmes 1984) only applies to the minority of (sub)cataclasmic RSFs. Geophysical surveys of RSF subsurface extent have not been attempted in British mountains.

There are at least five possible pre-failure planes of reference from which to measure RSF depth (Fig. 4). The map-based measure DPLAN is selected for this study as most relevant for landscape evolution, but measures into the rockmass are also reported since deep-seated failures will be more influential in trough widening than shallow ones.

Fig. 4.
Fig. 4.

Five possible planes of reference from which RSF depth of bite could be measured. DPLAN, orthogonal to the valley axis (as a plan measure of lateral valley widening); DRIM, a variant of DPLAN, from plateau rim (as a measure of scarp retreat); DSLOPE, parallel to the valley-side slope; DVERT, vertical depth to base of failure; DGEOL, parallel to the geological failure surface.

Map-based measures of depth of bite

Other than in the simplest rectangular evacuated cavity, no single measure of DPLAN can fully represent RSF depth of bite. Ideally, a composite measure would be generated from the average bite along each contour. However, with many RSFs having irregular contours such sophistication is unwarranted. A more robust method simply identifies the deepest contour recess; this recess may well occur where retained debris or infill is shallowest and where it may most closely approximate true cavity depth. Four DPLAN variants employed here bracket trough widening with a range of values (Fig. 5).

  • Maximum contour recess (DMAX). This measure best represents the visual impact of the failure. The deepest contour recess from a projection of the pre-failure contour across the site is taken. At Ordnance Survey 1:25 000 scale, contours are at 10 m vertical interval, but the bold 50 m contours are most reliable for reconstruction. The DMAX contour can be at the top of the failure, but is more usually in the upper third.

  • Average contour recess (DAVE). A prevailing figure (not a calculated mean) is taken along the same contour. In a 90° wedge, it would be half DMAX. It is a corrective for broad deformations where DMAX can be a local aberration, but is difficult to apply consistently.

  • Bay depth (DBAY). This is the generalized plan depth from the mouth of the failure bay where it interrupts the valley wall to the source scarp or fractures (the term ‘bay’ is preferred to ‘cavity’ as that refers to the bedrock shape). DBAY can reach an order of magnitude greater than DMAX, especially on long or gentle slopes (and thus does not usually represent actual scarp retreat).

  • Rim bite (DRIM) (Fig. 4). This applies in a minority of cases where RSFs cut into a plateau, and to incipient failure behind a rim or brow; it can coincide with DMAX.

Where no measurable contour recess or bay exists, as with failures on convexities and some slope deformations, the site is excluded from the depth analysis, but its breadth is still included in the total. Note that breadth and depth data cannot be multiplied to obtain failure volumes. Statistical analysis of the data is inappropriate given the diversity of RSF shapes and settings.

Fig. 5.
Fig. 5.

Three measures of DPLAN obtained in this study (DRIM is shown on Fig. 4). The difficulties of applying them meaningfully to irregular RSF slips and slope deformations are illustrated on the right.

Sample areas in Scotland and Norway

The mainland Scottish Highlands displays the most extensive array of primarily paraglacial RSFs in the British Isles. There are at least 550 definite and probable sites (>0.01 km2), their average area being 0.21 km2 (Jarman unpubl. data). Five sample areas have been selected (Fig. 6 and Table 1). Three are within the main clusters of large RSFs (>0.25 km2) identified by Jarman (2006), and two are in areas of sporadic or smaller-scale RSF. They span the Linton zonal typology of increasing glacial trough dissection from east to west (in Clayton 1974). All are in the prevailing Neoproterozoic metasediments, traversing their full diversity of lithology and structure. Slope angle and aspect have not been measured: while RSF occurs on lower- and higher-angle slopes alike, it is more frequent on valley sides with geological dip conducive to failure, or with open S-to-W aspects (cirque dissection reduces scope for RSF on N-to-E faces).

Fig. 6.
Fig. 6.

The five sample areas in the Scottish Highlands, in relation to clusters of larger RSFs and the Linton zones of increasing glacial dissection from east to west (from Clayton 1974). The near-complete population of larger RSFs (>0.25 km2) is a fair proxy for the population of all RSFs, which are widely distributed across all three zones but with great variation in clustering and sparsity. Adapted from Jarman (2006).

View this table:
Table 1.

Sample areas, populations and results

In the main RSF clusters, the Cowal–Arrochar–Luss sample area (Fig. 7) is intensely dissected by glaciated troughs and breaches. It is one of the densest clusters in Scotland, with RSF affecting 7–8% of the terrain in core areas (Jarman 2007a). The structural dip and micaceous schist interbedding are often conducive to translational sliding (Fig. 3). RSF is found at all levels and in all topographic contexts, but is least common on the flanks of mature valleys of preglacial origin. In this area, a mountain core is flanked by lower hills, both with available relief of 500–800 m. By contrast, in the Ericht–Gaick plateau sample area (Hall & Jarman 2004) RSF is clustered along the rims of two transectional breaches; one follows a major Caledonide fault, the other being sinuous and of uncertain origins (Fig. 8) (Jarman 2004a). Low-angle structures and arenaceous lithology are not conducive to sliding here, and failure tends to be by slope deformation or in deeply-weakened material. Available relief is only about 400 m. At the other extreme, the Kintail–Affric sample area has some of the highest available relief (<1000 m) and steepest terrain in Scotland. Long, well-defined ridges are separated by major transectional breaches, but only locally interrupted by cross-breaches, where RSF is most intense, affecting 6% of the terrain (Jarman 2003c). The geology of high-angle indurated metasediments dissected by faults promotes failure in both deformational and slide modes, but RSF can occur without structural assistance; it is sparse on the mature east-flowing valley sides.

Fig. 7.
Fig. 7.

The Cowal–Arrochar–Luss sample area, SW Highlands, is one of the densest RSF clusters in Britain. Smaller failures predominate in the northern (Arrochar Alps) part in higher, more glacially scoured terrain. Many RSFs are contributing to widening the complex network of glacial breaches of regional and local watersheds. Updated from Jarman (2003a).

Fig. 8.
Fig. 8.

The Gaick Pass, a narrow glacial breach of the main Grampian watershed east of Drumochter, looking south. The plateau rim is 825–850 m asl and Loch an Dùin [NN723800] is at 490 m. Despite the limited available relief of approximately 350 m, the left (east) side displays slope deformation along 1 km of trough wall with remarkably deep incipient extension (DRIM of 175 m to white dashes) into the paleic surface of A' Chaoirnich. On the right (west), the nose of An Dùn is sharpened by slippage with a DMAX of 40 m (inferred former crestline white dotted).

Away from the main clusters, the Monar–Strathfarrar sample area is in high mountains with transectional breaches and similar geology to Kintail–Affric, but displays sparse albeit bold RSF, such as Sgurr na Conbhaire (Fig. 9) and Sgurr na Ruaidhe (Fig. 10). The Dearg–Wyvis sample area (Fig. 11) has an intermediate plateau book-ended by higher massifs, with smaller-scale RSF along a narrow internal valley rim and in minor breaches, which are probably fault-directed.

Fig. 9.

Sgurr na Conbhaire RSFs, Western Highlands [NH 082429] in the Monar–Strathfarrar sample area. Vertical air photograph © RCAHMS. (a) The extant main RSF is a classic long-travel arrested translational slide, which has removed the summit and retreated the spur-end by up to 75 m (inset). The upper cavity is substantially evacuated, with a slabby west flank scarp acting as a main slide plane. (b) It is inferred from the cavity-within-a-bay form that an earlier, broader RSF occurred here. The process of wholesale glacial truncation of this spur is thus being assisted by extensive RSF with DMAX coincidentally of c. 90 m in both generations of cavity. Fenton (1991) estimated DGEOL at 150 m, reflecting the thickness of the main arrested mass. The earlier cavity may originally have been deeper, as its mid-slope arms appear to have been pared back by the valley glacier during an intervening stadial.

Fig. 10.
Fig. 10.

Sgurr na Ruaidhe RSF [NH292424] in the Monar–Strathfarrar sample area. The deep, fresh-looking short-travel arrested slide bites by 25 m (DMAX) at * and 75 m (DRIM) into paleic relief (pale tone). The adjacent cavity to the east may be the scar of an earlier RSF with all debris since evacuated, with an inferred DMAX of 120 m. It is at present a poorly developed SE-facing corrie. The extant RSF locus is separated from the main immature Strathfarrar trough by the promontory of Garbh-charn, and failure is thus inferred to have been provoked by glacial ‘parafluence’ through the col parallel to the Farrar ice flow, with an available relief of only 200 m. Note incipient encroachments 25 m into the plateau above the main RSF headscarp, with ground lowering of c. 2 m. Vertical air photograph © RCAHMS.

Fig. 11.
Fig. 11.

The Dearg–Wyvis sample area, Northern Highlands. By contrast with Figure 7, RSF is sparse and small scale, with only two larger cases. As the geological factors are not dissimilar, this may reflect its lightly dissected character at a centre of ice dispersal where the ice cap was thinning northwards. RSF incidence may here respond to local deepenings at a relatively early stage of glacial adaptation. Gleann Mór–Beag is a narrow trough incising the intermediate plateau by only 300–400 m. Alladale is a short but perhaps aggressively enlarging trough-corrie, a branch of which has transected the external plateau margin. Ben Wyvis is flanked by deep narrow glacial breaches across the NE–SW mountain barrier. Other RSF loci range from summit crests (where they are enlarging corries) to low valley sides. Nevertheless, DMAX is not significantly different from the intense clusters.

Inventories of RSF for all these sample areas have been compiled from an air photography search (Holmes 1984); from those British Geological Survey 1:50 000 maps and unpublished field slips that show landslips; and from field explorations. The 193 sites identified cover about a third of the Highland RSF population. Of these 146 had measurable breadths, but only 96 yielded map-measurable depths, including 36 cases of more than 0.25 km2, which comprise 24% of the better-verified large-RSF database (Jarman 2006).

A comparator area in north Norway offers an initial check on the possible wider relevance of the Scottish results. NE Troms (east of Tromsø) has affinities with the Highlands in its 1000 m available relief, its subhorizontal Caledonide metasedimentary geology and its dissected paleic relief. It possesses one of the densest reported RSF clusters in the Scandes (Kverndal & Sollid 1993). An area east of Lyngen fjord (Fig. 12) has been geomorphologically mapped at 1:50 000 (Tolgensbakk & Sollid 1988), identifying 50 mass movement sites that appear to span the five Highland categories of paraglacial RSF. Most are subcataclasmic debris masses sufficiently evolved to resemble rock glaciers (Kverndal & Sollid 1993); only seven are cataclasmic events filling the cirque or trough floor. None face E or NE. Seventeen deposits have measurable cavities above them, which are probable RSF sources.

Fig. 12.
Fig. 12.

The north Norway sample area, at Kåfjord east of Tromsø [69°40′N, 21°45′E]. Probable RSFs are interpreted from the 1:50 000 geomorphological map (peripheral areas omitted) with limited field verification in Olderdalen and Nordmannvikdalen. Paleic relief and DMAX contour recesses are plotted from the 1:50 000 topographical map. Balsegáisá is a paleic relief remnant undergoing demonstrable shrinkage by RSF. Dalvvesvárri is a cataclasmic slide damming a lake; its extreme cavity depth relies on reconstruction of a lost promontory (Jarman 2002) and is discounted from the average DMAX. Gavtavárri is the largest slope deformation mapped, with progression to slide lobes.

Results showing significant and consistent depth of bite

The results (Table 1) confirm that the typical paraglacial RSF makes a very substantial ‘bite’ into the trough wall. There is also consistency between the diverse sample areas, suggesting that RSF processes respond to slope scales, structures and stresses of widespread applicability. Averaged DMAX is consistently in the range 40–45 m across four of the five areas; it is coincidental with a 40 m figure reported for 30 RSFs in the modal 0.5–1.0 km breadth class in northern Iceland (Bentley & Dugmore 1998). A higher figure in north Norway (60 m) might reflect a greater degree of cavity evacuation in somewhat larger-scale relief, but may simply be sampling bias to map-recognizable cases. The Ericht–Gaick DMAX result is lower because the small sample is mainly of deformations and sags that tend not to yield pronounced contour recesses. There is greater localized variation within Cowal–Arrochar–Luss where the steep, laterally-convex Arrochar Alps favour smaller RSFs, and the open slopes of the Cowal and Luss hills accommodate broader RSFs. The mapped results for DMAX in two contrasting sample areas (Figs 7 and 11) demonstrate the consistency of paraglacial ‘bite’ for individual RSFs.

The DAVE measure is about half DMAX in the most typical area (Cowal–Arrochar–Luss); more than half in Monar–Strathfarrar with its bolder bites; and one-third in Kintail–Affric with its significantly broader failures. The numerical value of DAVE is of little intrinsic value, but confirms that DMAX is not aberrant, and that regional variations are reasonably explicable. Average bay depth (DBAY) is typically six times DMAX and ranges from 170 to 400 m; in Monar–Strathfarrar (290 m) it is about 30% of prevailing half-valley-width, suggesting substantial unexhausted potential for long-term trough widening. Greater bay depth goes with longer trough walls.

Average RSF breadth (B) ranges from 270 to 600 m. RSF affects 14% of total valley wall length in Cowal–Arrochar–Luss and 9% in Kintail–Affric, but such significant cumulative impact is limited to the dense clusters. Although a direct relationship between depth and breadth is reported by Bentley & Dugmore (1998), the results presented here suggest considerable variability. The depth : breadth ratio is lowest where structures are conducive to sliding, and where dissection limits RSF width. It is markedly higher where slope deformation is prevalent.

These averaged results meld the full range of failure sizes, types and topographic contexts; they do not consider variations in lithology, gradient, aspect or valley-side roughness. Some individual sites are very large (Table 2), in areas of both dense and sparse RSF. Six such RSFs bite into wall or rim by 100 m or more, notably in Cowal–Arrochar–Luss (Fig. 7), with Streap (Fig. 1) attaining 155 m. Twelve sites or adjacent groups exceed 1 km in breadth, with Beinn Fhada at 3 km being the largest RSF in the British mountains.

View this table:
Table 2.

Significant individual cases of RSF trough-widening

Concentrated RSF trough widening effects

Extant paraglacial RSF is clustered both regionally and locally. Its quantitative geomorphic impact over a whole mountain range is relatively small: the interest here lies in where and why it becomes intense. This may be in specific valleys, in over-enlarging corries (cirques) or on particular plateau rims.

Whole-valley impacts: two case studies

While overall valley-wall RSF incidence does not exceed 14% even in dense clusters, locally it attains 18% in the Gaick Pass (Fig. 8) and 25% in the north Loch Ericht breach.

In Knoydart (Fig. 13), RSF affects 28% of the valley sides of Gleann an Dubh–Lochain and Gleann Meadail, while it is almost absent in adjacent Gleann na Guiserein which is of similar character. One possible difference lies in the inputs of transfluent ice over breaching cols to the first two valleys. The north flank of Gleann an Dubh–Lochain is affected by RSF for at least 47% of its length, with Aonach Sgoilte displaying the most dramatic split ridge in Britain, comparable to an alpine ‘doppelgrat’ (see fig. 2.12 in Jarman 2007a). Since most of the RSFs are broad slope deformations, depth measures have not been taken.

Fig. 13.
Fig. 13.

High RSF incidence in central Knoydart valleys (for location see Fig. 6; NW part of Cluster 2). Available relief ranges from 600 to 1000 m. Deformational RSF is intense in two valleys with glaciated cols at their heads, but sparse elsewhere except locally on Luinne Bheinn below a breach of the main watershed.

At Glen Luss (Fig. 14), a remarkable sequence of RSFs ranging from near-in situ slope deformations to long-travel slides occupies 52% of the north valley side (whole-valley impact c. 26%). Taking DAVE and weighting it by RSF breadth, extant RSF cavities alone are widening about half of one side of the trough by 25 m, along the site-by-site contour of greatest indentation. There is no obvious explanation for such concentrated activity here, nor similarly on the north side of adjacent Glen Douglas. Failure is predominantly on south aspects partly in response to a favourable structural dip (e.g. Fig. 3); north aspects seem generally less susceptible to RSF. Asymmetric trough profiles are common, and may evolve by such preferential widening.

Fig. 14.
Fig. 14.

High RSF incidence in Glen Luss (for location see Fig. 7). Despite available relief not exceeding 500 m, paraglacial trough widening is pronounced. RSF is favoured on the SW-facing valley side by a consistent regional dip in fallible lithologies with extensive through-going discontinuities. On other aspects in the Luss area, RSF tends to be limited to progressive deformation or small-scale collapses.

Other individual valleys with high RSF incidence include Glen Ample, which follows the major Caledonide Loch Tay Fault and is one of the better candidates for a neotectonic association (located in Cluster 7 of Fig. 6) (see Jarman 2007d). Glen Roy (located in Cluster 3) is noted for its proglacial lake jökulhlaups, but the RSFs are of varied dates and characters (Fenton 1991), and the valley is a breach of the main pre-glacial watershed (Jarman 2008). RSF mini-clusters are also associated with breach systems such as Tyndrum–Orchy–Lyon, Tay–Almond–Amulree and Cluanie–Affric.

Corrie widening

While RSF commonly contributes to widening main valleys and troughs (in about 70% of cases in the two largest clusters; Jarman 2003a, c), it also promotes side bay and corrie enlargement (15% of cases). RSF is uncommon in classic cirque bowls, most cases being in elongated side-troughs and open embayments. It occurs mainly on corrie flanks rather than headwalls: even at Ben Hee, where the head of Gorm-choire appears to have failed, this is inferred to be a former spur within a compound corrie (Jarman & Lukas 2007). This suggests that headwalls retreat by incremental attrition, with stresses continually relieved by rockfalls that rarely reach the RSF-scale seen in the Garbh Choire Mòr (Fannaich) slabslide (Holmes 1984). Headwalls may be buttressed against larger-scale failure by their arch-form in plan. Corrie flanks behave more like valley sides, with RSF promoting lateral rather than headward enlargement (e.g. Coire Gabhail; Ballantyne 2007b). This may help to account for the evolution of the broad compound cirques common in parts of the Highlands (Gordon 1977).

In Glen Clova, a sequence of disproportionately large cirques facing SW has attracted attention (Holmes 1984). Here, a cluster of RSFs has developed, including two on either flank of Corrie Brandy (Fig. 15), the larger biting into the flank rim by 130 m. This cirque is undergoing significant lateral widening by RSF; adjacent cirque flanks display angular shapes and scars suggesting post-late Devensian failures removed by the last corrie glaciers.

Fig. 15.
Fig. 15.

Corrie Brandy [NO343756] in Glen Clova, a glacial trough incising an intermediate plateau in the SE Highlands (for location see Fig. 6). (a) Lateral widening of the atypically SW-facing corrie by the extant main RSF slippage (DRIM of 130 m), by incipient extension above and adjacent to it (dotted line), and by a thin slice partially detached from the west rim; vertical air photograph © RCAHMS. (b) The main RSF on the east flank, with incipient scarplets top left; note the paleic relief in Linton Zone II of limited dissection (see Fig. 6).

Incipient trough widening

Incipient RSF is identified where a rock mass has clearly failed along fissures or fractures but has only displaced by a few metres or decimetres; no contour cavity is revealed, but the potential bite into plateau or ridge measured by DRIM can exceed 100 m (Table 3). It often develops above manifest RSFs (in 19 out of 72 RSFs in Cowal–Arrochar–Luss), where it is additional to valley-wall DPLAN measures, and attests to upward migration of failure. Few of these incipient RSFs appear active, a rare case (evidenced by torn vegetation) being on Meall a' Chleirich (Fig. 16). More common are apparently dormant step-scarps (Fig. 17b) or false antiscarps (uphill-facing source scarps where the incipient failure daylights behind the crest, e.g. Fig. 3).

Fig. 16.
Fig. 16.

Meall a' Chleirich, Reay Forest, Sutherland [NC412364] (for location see Fig. 6). (a) The prominent lobe of coarse shattered debris is only subcataclasmic, barely reaching the lower slope. It protrudes by c. 30 m and is ripe for removal by the next glacier. It leaves a DMAX cavity above it of similar depth. On its right is a precarious failed mass that has only just parted company from the rim; there are large fresh rockfalls and extensive incipient fracturing behind the rim. On its left is a debris-free cavity with stepped source scarp inferred to date from earlier event(s). View NW across breach of main watershed. (b) The broad 0.5 km2 RSF complex encroaches into the flat summit (a paleic relief residual) and a lower etch-surface for 1 km along the breach rim. The dashed line suggests the loss of paleic relief to RSF in recent cycles. Platy slippage NE of point 500 has a DRIM of c. 20 m. Vertical air photograph © RCAHMS.

Fig. 17.

Garbh RSF, Ardgoil [NS243998] (for location see Fig. 7). (a) The broad shoulder of Cnoc Coinnich (761 m asl) has Caledonoid lineaments exploited successively as A–A collapse scar along the trough rim; B–B source scarp to subsidence zone with extensive visible antiscarping; C–C1 step scarp (Fig. 17c) continuing down the north flank to C2 as head of the zone of antiscarped creep; D–D ground rupture of uncertain relationship to the main failed area; and E–E fluvial erosion gully. C–C marks by far the greatest known depth of incipient scarp retreat (DRIM) in Britain at c. 550 m. The extent of failure is uncertain; C.T. Clough mapped failure down to the shore, now obscured by forestry. The photograph has the sun angle from SE, and shows distinctly darker vegetation within the main failed area up to D–D, indicative of freer-draining (fractured) terrain, endorsed by extensive springs below the northern boundary. Vertical air photograph © RCAHMS. (b) Long section showing that only a very low angle of failure is possible; the effect of RSF is as much trough-deepening as trough-widening in such cases. (c) The fresh-looking (unglaciated) step-scarp at C–C, view at C1 looking north.

View this table:
Table 3.

Large incipient rim retreats in the sample areas

Beinn Fhada and A' Chaoirnich are the largest reported incipient RSFs, both affecting kilometric lengths of valley wall to hectometric depths, clearly bounded by a low (1–3 m) scarp or antiscarp fracture, and with extensive disturbance of the intervening preglacial land surface. Both are located above troughs transecting main watersheds, but in some of the greatest and least relief to bear RSF in Britain (850 and 330 m floor to rim, respectively).

The case at Garbh, Ardgoil [NN 241002] is exceptional and has not been discussed previously (Fig. 17). This RSF complex is on a broad shoulder breaking at the trough rim into overhanging crags with slipped and broken masses below. The trough is occupied by Loch Long, a fjord on the Caledonian trend. RSF extent is uncertain: Holmes (1984) identifies two sites totalling 0.79 km2, but Geological Survey field slips by C. T. Clough (c. 1897) delineate 1.75 km2. His failure boundary extends well up the heavily dislocated shoulder above the crag collapse rim to a linear feature (shown as C–C1 on Fig. 17a) noted as ‘small cracks and slips’. This is, in fact, a sharp step-break of headscarp-and-furrow character 1–2 m high (Fig. 17b), which is distinct for approximately 500 m on the Caledonoid NNE–SSW trend. It is clearly post-glacial and not erosional. It resembles the Nordmannvikdalen ‘neotectonic fault scarp’ in north Norway (Dehls et al. 2000). Of two parallel features upslope on the air photograph, the farther E–E is simply erosional, but the nearer D–D is locally a submetric scarplet and furrow, and marks the extent of dryer-ground vegetation. The trough rim is degraded by scouring, but is interpolated along the truncated spurs at 500–550 m asl (A–A). Deep-seated incipience up the 1 km-broad shoulder could attain (DRIM): 400 m to B–B, a bold 30 m high scarp at 660 m asl [NN 23850045]; 550 m to C–C, the sharp step at 670 m asl [NN 23750055]; and 675 m to D–D, the lineament at 690 m asl [NN 23630062].

Such large-scale sequential slicing back of this truncated spur along successive Caledonoid lineaments is similar to Mullach Coire a' Chuir nearby (Jarman 2003a). However, the long-section (Fig. 17c) shows that the basal discontinuity cannot be steeper than c. 18°, which is below the usual threshold permitting creep to progress to sliding in schists (Watters 1972). This gives a depth of incipient RSF of up to 75 m (DGEOL), which would increase if an even gentler rupture zone existed, as with the 14° slope obtained for the very large Ben Our RSF (Jarman 2007d). The cause of such pervasive slope instability in relatively subdued relief on both sides of Loch Long (Fig. 7) may be glacial overdeepening along a Caledonian structural weakness transecting former interfluves. These may have been breached by transfluent ice, or a pre-glacial Clyde-river capture of the Forth headwaters may have been enlarged by diffluent Loch Lomond ice (Linton & Moisley 1960; Jarman 2003a).

Comparative rockwall retreat rates

Previous studies in the Scottish Highlands have not taken map-based measures of RSF depth. Watters (1972) inferred long profiles through 13 large cases, from which DGEOL can be scaled ranging 15–90 m and averaging 50 m. Fenton (1991) suggested that the bolder RSFs in the NW Highlands have DGEOL typically of 30–100 m, with Sgurr na Conbhaire (Fig. 9) reaching 150 m. These figures suggest the moderate extent to which DMAX contour bites may understate true cavity depths. But caution is needed with large slope deformations, where deep arcuate failure surfaces (Jarvis 1985) are less likely than shallower stepped surfaces or transitional creep zones as inferred at Beinn Fhada (Jarman 2006).

Holmes (1984) obtained DSLOPE values in six small (sub)cataclasmic cases in the 0.1−1 million m3 range with quantifiable debris volumes. His retreat rates average 25 m, which is consistent with the DMAX result for Arrochar/Ardgoil where such RSFs predominate. Scarp retreats of 6 m and 14.3 m have also been calculated from debris volumes at Baosbheinn, Gairloch (Sissons 1976) and Beinn Shiantaidh, Jura (Dawson 1977). These deposits had been assumed to be rock glaciers, but are now regarded as more likely to be RSFs (Ballantyne & Harris 1994; Ballantyne 1997). Ballantyne (2007a) found a DSLOPE of 25 m by contour extrapolation across the exceptional evacuated cavity of Beinn Alligin. At Ben Hee, where on a convexity the RSF does not display contour recesses, Jarman & Lukas (2007) applied DVERT to a terrain reconstruction with balanced failure volume and pre-failure relief. This gives a maximum vertical depth of surface lowering of 60 m, with ‘scarp retreat’ of the crestal position of 80–120 m.

It is difficult to compare singular RSF depths with incremental para/periglacial rates of rockwall retreat. Fifteen studies of talus slopes, debris flows and pronival ramparts yield retreat rates of 0.01–3.3 m ka−1 (Ballantyne & Harris 1994). Over an interglacial these are orders of magnitude less than RSF in Scotland, but upper-end rates could be comparable under prolonged periglacial conditions. In a rare study of contemporary RSF as a continuing process from rockslides onto an Alaskan glacier, Arsenault & Meigs (2005) calculated a mean valley-wall retreat rate of 6.7 m ka−1, which scarcely increases if all rockfall debris is included. Even these modest process rates greatly exceed surface-dating results of less than 2 m of glacial erosion on Scandian trough walls in the entire last glacial cycle (Li et al. 2005).

Discussion

However vivid these illustrations of paraglacial RSF ‘bite’, they are only a response to one deglaciation amongst many, while the quantitative measures are of limited value in themselves. Their significance has to be assessed within the context of the glacial–paraglacial cycle as it has evolved over the Quaternary, and must bear in mind problems associated with the Loch Lomond Stadial. The efficacy of glacial exploitation of RSF bite must then be critically evaluated. A further problem arises with the sporadic incidence of RSF along trough walls, making the ‘paraglacial relaxation’ process stochastic rather than deterministic. Finally, the implications of RSF trough widening for landscape evolution merit discussion.

Glacial–paraglacial cycling: an evolutionary model

In a simple glacial–paraglacial sequence, glaciers erode trough walls, then retreat and debuttress them; the walls fail paraglacially; the next glaciers remove the failed material, and then repeat the cycle. The earliest known perception of the scale and cyclical significance of paraglacial RSF is by Holmes (1984, p. 232), who concluded: ‘Millions of cubic metres of failed debris, mainly located in the Western Highlands, presently await glacial transportation. If similar volumes to those found today were created repeatedly in the past, then rock slope failure combined with glacial transportation must have been a major component of the denudation of the areas of the Highlands where RSF is important’.

Subsequent recognition of many glacial–paraglacial cycles over the Quaternary makes cumulative impact of RSF even more substantial, especially in the most geologically susceptible areas (Evans 1997). This impact may be greater than it might appear from the present sporadic incidence of RSF, if it was considerably more prevalent at early stages of fluvial valley adaptation to ice discharge (Jarman 2003a).

A schematic model (Fig. 18) demonstrating this evolving contribution over successive glacial–paraglacial cycles is necessarily speculative as past RSF incidence cannot be extrapolated from extant evidence, and there can be no comparator regions at immature ‘early Quaternary' stages of development. An analogy might be found in young mountain ranges, where landsliding is now being recognized as the largest contributor to their erosion (Hovius & Stark 2006). There, RSF rates tail off as tectonic uplift wanes; here, they decline as glacial trough profiles mature and, conjecturally, slopes become ‘stress-hardened’ over repeated cycles, with exhaustion of the main fallible weaknesses. Rejuvenation will invigorate trough-floor incision and promote renewed RSF pulses, if system perturbations are vigorous enough. These might be regional, such as glacioisostatic rebound, or localized, such as glacial breaching. Present RSF clusters are often associated with glacial breaches (Jarman 2002, 2003a).

Fig. 18.
Fig. 18.

Model of evolving paraglacial trough-widening effects over time. RSF occurs in ‘zones of paraglacial relaxation’ of varying scale. Each stage may well persist through a number of glacial–paraglacial cycles. The two processes alternately undo each other's efforts in terms of preferred slope angle. Stage 1: fluvial V to glacial U profile conversion. During early glacial stages, as original fluvial valleys adapt to ice discharge, RSF is likely to be intense, especially where glacial erosion is concentrated. In fallible lithologies erosion will be endemic, with all valley sides affected. In massive lithologies such as granite, where the unit of rock-mass failure seldom achieves the RSF threshold size of 0.01 km2/100 000 m3, hyperabundant minor failure is likely. Stage 2: preferential trough enlargement. As trough shape and size approach peak efficiency for catchment ice discharge, concentrated glacial erosion will diminish, as will the scope for paraglacial RSF. In structurally controlled terrain, RSF will occur preferentially on failure planes dipping valleyward, promoting trough asymmetry and lateral displacement of divides. Stage 3: maturity. Where valleys have adjusted to ice discharge and ‘stress-hardened’ slopes have regained quasi-stability, RSF will become sparse. This fits Scottish evidence for mature glaciated valleys with relaxed profiles such as Dochart–Tay (Jarman 2003a). In valleys long adapted to ice discharge, bulk erosion over a glacial cycle has become insufficient to daylight new failure planes, or to provoke stresses during isostatic rebound in excess of shear strengths. Some main valleys are essentially pre-glacial forms with rather limited modification by ice, notably where they follow ancient lineaments rather than dendritic or zig-zag courses. Stage 4: rejuvenation. Late Quaternary RSF activity may be a response to: (a) widespread glacial reincision where wholesale glacial erosion and ensuing glacio-isostatic rebound maintain available relief sufficiently that glaciers are constantly reincising their floors, perhaps creating ‘alps’ with a fresh zone of paraglacial relaxation (e.g. the Randa RSF locus, Eberhardt et al. 2004); and (b) locally concentrated erosion, where a glacier is exploiting a linear weakness, such as fault crush, or where shifting icesheds/dispersal routes and ice streaming promote glacial transfluence and breaching. Augmentation of local catchment ice by several-fold is possible: this will render the existing trough underfit and promote its rapid enlargement.

Interstadials, interglacials and the Loch Lomond Stadial

RSF efficacy over the Quaternary in this model is clearly dependent on the number and erosional severity of glacial–paraglacial cycles. In ranges such as northern Iceland (Bentley & Dugmore 1998) where the main troughs remain occupied by ice during interstadials, cycle frequency is much reduced. Conversely, shorter stadials with less glacial bedrock erosion will leave lower rock-mass stresses upon deglaciation, as with the Loch Lomond Stadial which, in the Highlands, merely reworked valley debris (Godard 1965). However, this takes a static view of slope stress fields: if rebound stresses (Hutchinson 1988) are dynamic, it might be that the abrupt loading and unloading of Loch Lomond ice exercised a ‘trampoline effect’, triggering a spate of RSFs. The Loch Lomond Stadial glacial–paraglacial cycle could either be a freak event overstating the typical mature-stage incidence of RSF (or understating it if its glaciers removed the evidence of some lateglacial RSFs) or it could suggest ‘trampolining’ as a significant factor during earlier rapid climatic fluctuations.

How effectively can glaciers exploit RSF cavities?

The glacial–paraglacial cycle model also implies that glaciers will substantially exploit the failed slopes during each stadial, in two ways: (1) by excavating the weakened material from the cavities; and (2) by gaining purchase on the cavity angles to increase erosion rates on intervening segments of trough wall. The latter process is inferred from Bentley & Dugmore (1998, p. 14) who noted that RSFs ‘create irregularities in the trough side [which] could enhance trough widening as the glacier re-establishes a smoother channel form’. Figure 19 tests these critical assumptions. If all its obstacles or alternatives to cavity exploitation apply, it could take several glacial cycles to exhaust the deeper RSF bites.

Fig. 19.
Fig. 19.

Four possible reasons why the ‘zone of paraglacial relaxation’ may not be fully exploited for trough widening during the next glaciation. In (1) cavities are occupied by dead ice, whether endo- or exogenic (the classic glacier image is of uniform width). Likewise in (2) where failed masses remain largely in their cavities, they might simply be bypassed by the next glaciers; trimming of the failed toe may only promote limited further creep in the next interstadial, thus requiring many cycles for complete evacuation. It is not clear why slope segments between failure bays are eroded more rapidly than ordinary open slopes: channelling of ice through a constriction (3) might promote acceleration or overdeepening rather than enhanced widening. And where noses between RSF cavities are broad (4), or in resistant rock, the ‘channel-smoothing’ contribution to valley-widening must be restricted to rubbing off their angular mouths.

The ‘zone of paraglacial relaxation’

In Scotland and Scandinavia, as against higher ranges, extant RSF spacing is rarely close enough for intervening trough-wall segments to be significantly exposed to future glacial attack, especially in resistant bedrock. Wholesale trough widening by this means requires that, over multiple glacial cycles, cumulative RSF incidence must affect almost the entire valley side. If structural strength/fallibility is consistent along a valley side, then this may be possible: geotechnical studies of RSFs generally find that slopes were close to critical thresholds, and only required small triggering forces to fail (e.g. Bjerrum & Jørstad 1968; Holmes 1984). This would imply that adjacent sectors of valley wall are ripe for failure in the next cycle. However, it is equally likely that variations in rock strength and discontinuity configuration will promote survival of intercavity bluffs. The evidence of incipient failures (Table 3) suggests that RSF propagates from itself rather than in fresh locations. Indeed, RSF may promote initiation of corries and side bays as much as widening of troughs (e.g. Fig. 10), if not to the extent suggested by Turnbull & Davies (2006) that most corries originate from (seismically-triggered) RSF cavities.

Any measures derived from extant RSF bites are thus no more than a broad indication of their actual contribution to valley widening during recent glacial cycles, and a tentative guide to their overall contribution during the Quaternary. The combined DBAY and incipient DRIM measures (typically 200–500 m) suggest the depth of a ‘zone of paraglacial relaxation’ along trough walls (Fig. 18) within which RSF may occur over a long period, and which will migrate upslope at a widely varying pace.

Overwidening, slope reduction, scarp retreat and paleic relief elimination

It is conventionally assumed that glacial trough size correlates with ice discharge, and that troughs cease to widen once equilibrium is attained, as attested by declining gross glacial erosion over the Quaternary (Sugden & John 1976). However, some glacial troughs display unusually high width : depth ratios, and Bentley & Dugmore (1998) proposed RSF cyclicity as a possible factor. This would be a classic positive feedback loop, with alternating glacial and paraglacial erosional processes widening valleys beyond the standard parabola. A further positive feedback arises if the overwidened valleys promote ice streaming, which draws in transfluent ice via glacial breaches (ice piracy).

In reality, troughs that appear overwidened in relation to ice catchment can be found in weak and massive lithologies alike, and with RSF both sparse (e.g. Rondane, Norway) and prolific (e.g. northern Iceland). Conversely, narrow profiles can occur in Scottish metasediments whether or not RSF is abundant. The three valleys in north Norway (Fig. 12) are variously parabolic with a few RSFs, shovel-shaped with abundant RSF, and maturely overwide with negligible RSF. With complex preglacial landscape and glaciological interactions, the role of RSF is hard to detect and quantify. What this feedback loop does highlight is that RSF is in itself a slope reduction process (e.g. Fig. 17). Only in tandem with glacial erosion does it become a scarp-retreat process, which is what ‘overwidening’ amounts to.

The corollary of trough widening is encroachment into the preglacial upland surface or ‘paleic relief’ (Gjessing 1967). Many RSFs bite significantly, whether by actual cavity or incipient failure, into paleic surface rims. Paleic relief is extensive in the selectively eroded eastern Highlands, and although many extant RSFs encroach into it (e.g. Figs 8 and 15) their percentage impact is small. Conversely, in Linton Zones III–IV (Fig. 6), where paleic remnants become vestigial (Godard 1965), RSF is a major contributor to their elimination (e.g. Figs 10 and 16). In north Norway several paleic surface blocks are undergoing significant attrition by RSF (Fig. 12). Identifying evolving RSF impacts in these migrating zones might assist in reconstructing the pre-Quaternary extent and character of upland relief.

Conclusions

  • Some quantification of RSF contribution to trough widening is possible, but deriving a process rate is presently unrealistic. It is easy and instructive to measure RSF breadth and the proportion of valley sides affected. However, measuring ‘depth of bite’ is compromised by problems of planes of reference, irregular configurations and non-evacuated cavities. Map-based results obtained here show order-of-magnitude RSF impacts. Geotechnical evidence of failure cavity shape and depth would identify failure surface/zone behaviour, and help calibrate a slope-stress model for glacial troughs. Analysing glacier–failed mass interactions might clarify the probable ‘yield’ (or conversion rate into future widening) for RSFs of different types over glacial–paraglacial cycles.

  • RSF is a high-magnitude–low-frequency contributor to trough widening of locally considerable impact. In the Scottish Highlands, the average RSF makes a significantly deeper ‘bite’ into trough walls than any other glacial or para/periglacial process. Average RSF breadths across the five sample areas range 270–600 m. In two of the main clusters RSF affects 9% and 14% of total valley wall, rising to 26–28% in three valleys in Knoydart and the Luss Hills. Average maximum contour recess (DMAX) ranges from 40 to 45 m across the four largest areas. This suggests the scale on which metasedimentary structures in older ranges respond to deglaciation stresses. The DMAX value must not be misrepresented as a real measure of RSF depth or as a widening rate per glacial cycle. It is merely a spot measure of the scale of visible RSF cavities. It may exaggerate the ‘degree of purchase’ offered to the next valley glacier in exploiting failed trough walls, if the average value across the whole failure is low. It may understate the impact where bedrock cavities are concealed by residual failed masses and subsequent infill.

  • The ‘zone of paraglacial relaxation’ is best approximated by the DBAY measure that ranges from 170 to 400 m, in conjunction with cases of rim bite and incipient failure, with DRIM values of over 100 m (maximally 550 m).

  • A model of RSF is developed for glacial–paraglacial cycles over the Quaternary (Fig. 18). The absence of such a model in the literature reflects: a historic focus on glacial trough deepening rather than widening; a lack of awareness of paraglacial RSF as a process; and the lack of evidence from ranges at earlier stages of glaciation. The model shows that RSF is a key process within a glacial–paraglacial cycle with powerful feedbacks. Glacial troughs may be widened by RSF to more than the normal parabolic profile, enabling ice piracy with consequences for ice-sheet profiles and elevations, ice dispersal patterns, and landscape evolution. RSF intensity would have been greatest as fluvial valleys underwent adaptation to ice discharge, and diminished as trough walls became stress-hardened, except where undergoing rejuvenated incision, for example in response to glacial breaching.

  • The zone of paraglacial relaxation will vary over space and time, and will not be fully exploited in any one cycle. Unknown factors include: the ability of glaciers to evacuate failed debris masses or exploit valley-wall cavities; the effects of long glacials as against short stadials; and how mountain slopes respond to glacio-isostatic compression and rebound. RSF in the Scottish Highlands may be primarily a paraglacial response to the last main (late Devensian) deglaciation, with the Loch Lomond Stadial as a possibly anomalous event complicating interpretation of extant RSF incidence.

  • Within the zone of paraglacial relaxation, RSF incidence is localized and unlikely to affect entire valley sides. The failure of some valley walls and not others may be due to intact slopes persisting as strong points. The extent to which the rate of trough widening by glaciers can be accelerated by prior RSF pock-marking the walls is thus debatable. Present RSF sparsity may indicate maturity where it was previously prevalent; or that structure or lithology are unconducive to RSF. Alternatively, troughs may have widened too slowly for RSF ever to have been significant, notably in areas with capacious preglacial valleys. Wholesale trough widening is therefore a complex and little-understood process to which RSF makes a significant contribution. The model suggests why this contribution has been greater in the past, and plan-depth data from extant Scottish RSFs can offer a pointer to its efficacy.

Acknowledgments

I would like to thank G. Holmes for the RSF database in his unpublished thesis; J. Gordon for access to air photographs at Scottish Natural Heritage; the British Geological Survey (Edinburgh) for access to unpublished field mapping; J.L. Sollid for providing the geomorphological map of Kåfjord; and the editors and referees for their advice and encouragement.

References

  1. ↵
    1. Arsenault A. M.,
    2. Meigs A. J.
    (2005) Contribution of deep-seated bedrock landslides to erosion of a glaciated basin in southern Alaska. Earth Surface Processes and Landforms 30:1111–1125.
    OpenUrlCrossRefWeb of Science
  2. ↵
    1. Augustinus P. C.
    (1995) Glacial valley cross-profile development: the influence of in situ rock stress and rock mass strength, with examples from the Southern Alps, New Zealand. Geomorphology 14:87–97.
    OpenUrlCrossRefWeb of Science
  3. ↵
    1. Ballantyne C. K.
    (1997) Report on field meeting. The Quaternary of Islay and Jura. Quaternary Newsletter 83:31–34.
    OpenUrl
  4. ↵
    1. Ballantyne C. K.
    (2002) Paraglacial geomorphology. Quaternary Science Reviews 21:1935–2017.
    OpenUrlCrossRefWeb of Science
  5. ↵
    1. Cooper R.
    1. Ballantyne C. K.
    (2007a) in Mass Movement in Great Britain, Beinn Alligin, ed Cooper R. (Joint Nature Conservation Committee, Peterborough) Geological Conservation Review, 33, pp 111–116.
    OpenUrl
  6. ↵
    1. Cooper R.
    1. Ballantyne C. K.
    (2007b) in Mass Movement in Great Britain, Coire Gabhail, ed Cooper R. (Joint Nature Conservation Committee, Peterborough) Geological Conservation Review, 33, pp 132–136.
    OpenUrl
  7. ↵
    1. Ballantyne C. K.,
    2. Harris C.
    (1994) The Periglaciation of Great Britain (Cambridge University Press, Cambridge).
  8. ↵
    1. Benn D. I.,
    2. Evans D. J. A.
    (1998) Glaciers and Glaciation (Wiley, Chichester).
  9. ↵
    1. Bentley M.,
    2. Dugmore A. J.
    (1998) Landslides and the rate of glacial trough formation in Iceland. Quaternary Proceedings 6:11–15.
    OpenUrl
  10. ↵
    1. Bjerrum L.,
    2. Jørstad F. A.
    (1968) Stability of Rock Slopes in Norway, Norwegian Technical Institute, Publication, 69.
  11. ↵
    1. Clayton K. M.
    (1974) Zones of glacial erosion. Institute of British Geographers Special Publication 7:163–176, (includes unpublished material by D. L. Linton).
    OpenUrl
  12. ↵
    1. Cruden D. M.,
    2. Hu X. Q.
    (1993) Exhaustion and steady-state models for predicting landslide hazards in the Canadian Rocky Mountains. Geomorphology 8:279–285.
    OpenUrlCrossRefWeb of Science
  13. ↵
    1. Dawson A. G.
    (1977) A fossil lobate rock glacier in Jura. Scottish Journal of Geology 13:37–42.
    OpenUrlAbstract/FREE Full Text
  14. ↵
    1. Dehls J. F.,
    2. Olesen O.,
    3. Olsen L.,
    4. Blikra L. H.
    (2000) Neotectonic faulting in northern Norway; the Stuoragurra and Nordmannvikdalen postglacial faults. Quaternary Science Reviews 19:1447–1460.
    OpenUrlCrossRefWeb of Science
  15. ↵
    1. Eberhardt E.,
    2. Stead D.,
    3. Coggan J. S.
    (2004) Numerical analysis of initiation and progressive failure in natural rock slopes – the Randa rockslide. International Journal of Rock Mechanics and Mining Sciences 41:69–87.
    OpenUrlCrossRef
  16. ↵
    1. Stoddart D. R.
    1. Evans I. S.
    (1997) in Process and Form in Geomorphology, Process and form in the erosion of glaciated mountains, ed Stoddart D. R. (Routledge, London), pp 145–174.
  17. ↵
    1. Evans S. G.,
    2. Mugnozza G. S.,
    3. Strom A.,
    4. Hermanns R. L.
    (2006) Landslides From Massive Rock Slope Failure (Springer, Dordrecht).
  18. ↵
    1. Fenton C. H.
    (1991) Neotectonics and Paleoseismicity in NW Scotland. PhD thesis (University of Glasgow).
  19. ↵
    1. Gjessing J.
    (1967) Norway's paleic surface. Norsk Geografisk Tidsskrift 21:69–132.
    OpenUrlCrossRef
  20. ↵
    1. Glasser N. F.,
    2. Hall A. M.
    (1997) Calculating glacial erosion rates in NE Scotland. Geomorphology 20:29–48.
    OpenUrlCrossRefWeb of Science
  21. ↵
    1. Godard A.
    (1965) Recherches de Géomorphologie en Écosse du Nord-Ouest (Université de Strasbourg, Publications de la Faculté des Lettres, Fondation Baulig).
  22. ↵
    1. Gordon J. E.
    (1977) Morphometry of cirques in the Kintail–Affric–Cannich area of northwest Scotland. Geografiska Annaler 59A:177–194.
    OpenUrlCrossRef
  23. ↵
    1. Gordon J. E.,
    2. Birnie R. V.,
    3. Timmis R.
    (1978) A major rockfall and debris slide on the Lyell Glacier, South Georgia. Arctic and Alpine Research 10:49–60.
    OpenUrl
  24. ↵
    1. Lukas S.,
    2. Merritt J.,
    3. Mitchell W.
    1. Hall A. M.,
    2. Jarman D.
    (2004) in Central Grampian Highlands – Field Guide, Quaternary landscape evolution – plateau dissection by transfluent glacial breaching, eds Lukas S., Merritt J., Mitchell W. (Quaternary Research Association, London), pp 26–40.
  25. ↵
    1. Harbor J. M.
    (1992) Numerical modelling of the development of U-shaped valleys by glacial erosion. Bulletin of the Geological Society of America 104:1364–1375.
    OpenUrlAbstract/FREE Full Text
  26. ↵
    1. Hebdon N. J.,
    2. Atkinson T. C.,
    3. Lawson T. J.,
    4. Young I. R.
    (1997) Rate of glacial valley deepening during the late Quaternary in Assynt, Scotland. Earth Surface Processes and Landforms 22:307–315.
    OpenUrlCrossRefWeb of Science
  27. ↵
    1. Hewitt K.
    (1988) Science, Catastrophic landslide deposits in the Karakoram Himalaya, 242, pp 64–67.
    OpenUrlAbstract/FREE Full Text
  28. ↵
    1. Holmes G.
    (1984) Rock-slope Failure in Parts of the Scottish Highlands. PhD thesis (University of Edinburgh).
  29. ↵
    1. Evans S. G.,
    2. Mugnozza G. S.,
    3. Strom A.,
    4. Hermanns R. L.
    1. Hovius N.,
    2. Stark C. P.
    (2006) in Landslides From Massive Rock Slope Failure, Landslide-driven erosion and topographic evolution of active mountain belts, eds Evans S. G., Mugnozza G. S., Strom A., Hermanns R. L. (Springer, Dordrecht), pp 573–590.
  30. ↵
    1. Hutchinson J. N.
    (1988) Proceedings of the fifth Internationl Symposium on Landslides, Lausanne, Volume 1 General report: morphological and geotechnical parameters of landslides in relation to geology and hydrogeology (Balkema, Rotterdam), pp 3–35.
  31. ↵
    1. Jarman D.
    (2002) Rock slope failure and landscape evolution in the Caledonian Mountains, as exemplified in the Abisko area, northern Sweden. Geografiska Annaler 84A:213–224.
    OpenUrlCrossRef
  32. ↵
    1. Evans D. J. A.
    1. Jarman D.
    (2003a) in Western Highland Boundary – Field Guide, Paraglacial landscape evolution – the significance of rock slope failure, ed Evans D. J. A. (Quaternary Research Association, London), pp 50–68.
  33. ↵
    1. Evans D. J. A.
    1. Jarman D.
    (2003b) in Western Highland Boundary – Field Guide, Tullich Hill rock slope failures, Glen Douglas, ed Evans D. J. A. (Quaternary Research Association, London), pp 200–208.
  34. ↵
    1. Tipping R.
    1. Jarman D.
    (2003c) in Glen Affric and Glen Shiel: Field Guide, The Glen Shiel rock slope failure cluster, ed Tipping R. (Quaternary Research Association, London), pp 165–183.
  35. ↵
    1. Lukas S.,
    2. Merritt J.,
    3. Mitchell W.
    1. Jarman D.
    (2004a) in Central Grampian Highlands – Field Guide, Rock slope failures of the Gaick Pass, eds Lukas S., Merritt J., Mitchell W. (Quaternary Research Association, London), pp 103–111.
  36. ↵
    1. Jarman D.
    (2004b) The Cobbler – a mountain shaped by rock slope failure. Scottish Geographical Journal 120:227–240.
    OpenUrlCrossRefWeb of Science
  37. ↵
    1. Jarman D.
    (2006) Large rock slope failures in the Scottish Highlands: characterisation, causes and spatial distribution. Engineering Geology 83:161–182.
    OpenUrlCrossRefWeb of Science
  38. ↵
    1. Cooper R.
    1. Jarman D.
    (2007a) in Mass Movements in Great Britain, Introduction to the mass movements in the older mountain areas of Britain, ed Cooper R. (Joint Nature Conservation Committee, Peterborough) Geological Conservation Review, 33, pp 33–55.
    OpenUrl
  39. ↵
    1. Cooper R.
    1. Jarman D.
    (2007b) in Mass Movements in Great Britain, Sgurr na Ciste Duibhe, ed Cooper R. (Joint Nature Conservation Committee, Peterborough) Geological Conservation Review, 33, pp 62–69.
    OpenUrl
    1. Cooper R.
    1. Jarman D.
    (2007c) in Mass Movements in Great Britain, Druim Shionnach, ed Cooper R. (Joint Nature Conservation Committee, Peterborough) Geological Conservation Review, 33, pp 70–74.
    OpenUrl
  41. ↵
    1. Cooper R.
    1. Jarman D.
    (2007d) in Mass Movements in Great Britain, Glen Ample, ed Cooper R. (Joint Nature Conservation Committee, Peterborough) Geological Conservation Review, 33, pp 82–91.
    OpenUrl
  42. ↵
    1. Palmer A. P.,
    2. Lowe J. J.,
    3. Rose J.
    1. Jarman D.
    (2008) in The Quaternary of Glen Roy and Vicinity – Field Guide, The Roy–Lochy rock slope failure cluster: implications for glacial breaching, ice movements, and Parallel Road dislocations, eds Palmer A. P., Lowe J. J., Rose J. (Quaternary Research Association, London), pp 98–104.
  43. ↵
    1. Jarman D.,
    2. Ballantyne C. K.
    (2002) Beinn Fhada, Kintail: a classic example of paraglacial rock slope deformation. Scottish Geographical Journal 118:59–68.
    OpenUrlWeb of Science
  44. ↵
    1. Cooper R.
    1. Jarman D.,
    2. Lukas S.
    (2007) in Mass Movements in Great Britain, Ben Hee, ed Cooper R. (Joint Nature Conservation Committee, Peterborough) Geological Conservation Review, 33, pp 99–107.
    OpenUrl
  45. ↵
    1. Jarvis J. J.
    (1985) Large-scale Toppling Failure in Metamorphic Rock Slopes. PhD thesis (University of London).
  46. ↵
    1. Korup O.,
    2. Clague J. L.,
    3. Hermanns R. L.,
    4. Hewitt K.,
    5. Strom A. L.,
    6. Weidinger J. T.
    (2007) Giant landslides, topography, and erosion. Earth and Planetary Science Letters 261:578–589.
    OpenUrlCrossRefWeb of Science
  47. ↵
    1. Kverndal A. I.,
    2. Sollid J. L.
    (1993) Late Weichselian glaciation and deglaciation in NE Troms, north Norway. Norsk Geografisk Tidsskrift 47:163–177.
    OpenUrlCrossRef
  48. ↵
    1. Li Y.,
    2. Harbor J.,
    3. et al.
    (2005) Ice sheet erosion patterns in valley systems in northern Sweden investigated using cosmogenic nuclides. Earth Surface Processes and Landforms 30:1039–1049.
    OpenUrlCrossRefWeb of Science
  49. ↵
    1. Linton D. L.,
    2. Moisley H. A.
    (1960) The origin of Loch Lomond. Scottish Geographical Magazine 76:26–37.
    OpenUrlCrossRef
  50. ↵
    1. SigurĐsson O.,
    2. Williams R. S. Jr..
    (1991) Rockslides on the terminus of Jökulsárgilsjökull, southern Iceland. Geografiska Annaler 73A:129–140.
    OpenUrlCrossRef
  51. ↵
    1. Sissons J. B.
    (1976) A remarkable protalus rampart in Wester Ross. Scottish Geographical Magazine 92:182–190.
    OpenUrlCrossRef
  52. ↵
    1. Sugden D. E.,
    2. John B. S.
    (1976) Glaciers and Landscape (Arnold, London).
  53. ↵
    1. Tolgensbakk J.,
    2. Sollid J. L.
    (1988) Kåfjord, Kvartaergeologi og geomorfologi, 1:50 000, 1634 II (Geografisk Institutt, Universitet i Oslo).
  54. ↵
    1. Turnbull J. M.,
    2. Davies T. R. H.
    (2006) A mass movement origin for cirques. Earth Surface Processes and Landforms 31:1129–1148.
    OpenUrlCrossRefWeb of Science
  55. ↵
    1. Watters R. J.
    (1972) Slope Stability in the Metamorphic Rock of the Scottish Highlands. PhD thesis (University of London).
Back to top
Alerts
Citation tools
Permissions
View PDF
Email to
Print
Download PPT

Related Articles

Similar Articles

Cited By...