Glacial geomorphology of the Gaick, Central Grampians, Scotland

The Gaick is an enigmatic glaciated landscape in the Central Grampians, Scotland, dominated by an expansive dissected plateau. Previous studies have postulated widely di ﬀ ering interpretations of the glacial landforms and current understanding of the glacial events in this area is partly restricted by the absence of detailed glacial geomorphological mapping. To address this issue, we present a comprehensive 1: 46,000-scale glacial geomorphological map, covering an area of ∼ 520 km 2 . A combination of detailed ﬁ eld mapping and interpretation of aerial photographs and Digital Surface Models (DSMs) has revealed a variety of glacial, periglacial and ﬂ uvial landforms, including an abundance of moraines and meltwater channels within valleys. We also identify a glacial sediment-landform assemblage, dissected glaciogenic material, that has not previously been reported in the Scottish Highlands. The geomorphological map provides the necessary foundation for elucidating the extent, dynamics and timing of former glaciation in the area.

This situation is exemplified by the Gaick, a dissected plateau in the Central Grampians, Scotland (Figure 1). The Gaick has proven to be an enigmatic and rather controversial area, with widely differing interpretations of the glacial sediment-landform record having been proposed. Three main conceptual models of former glaciation have previously been suggested for the Gaick: (i) an extensive Lateglacial (∼14.7-11.7 ka) re-advance which occupied much of the central and southeast Grampians (Charlesworth, 1955); (ii) advance and retreat of an extensive, locally-sourced Younger Dryas plateau icefield (Sissons, 1974); and (iii) retreat of the last British-Irish Ice Sheet across the plateau (Merritt, 2004;Merritt, Lukas, & Mitchell, 2004;Smith, Merritt, Leslie, Krabbendam, & Stephenson, 2011).
The three conflicting interpretations of glacial events in the Gaick partly reflect the absence of a detailed geomorphological map for the area. Although the Quaternary deposits and many associated landforms were mapped at a scale of 1: 25,000 in the 1990s by the British Geological Survey (BGS, 2008a(BGS, , 2008bSmith et al., 2011), the maps only depict the spatial coverage of moraine assemblages (i.e. the moraines are mapped as a 'unit'), rather than individual moraines. Similarly, the earlier geomorphological mapping of Sissons (1974) did not identify individual moraine mounds and ridges, instead identifying entire tracts of 'hummocky moraine' using a stippled pattern. Consequently, it is challenging to assess any systematic variations in the moraine patterns (e.g. upvalley changes in moraine orientation, shape and/or style) and concomitant changes in the spatial distribution of other sediment-landform assemblages, which can be important strands of evidence for differentiating distinct phases of former glaciation in upland Britain (cf. Boston et al., 2015;Lukas, 2006). Thus, the important first stage in elucidating former glacial events in the Gaick is to perform a thorough geomorphological assessment. This paper therefore presents a detailed glacial geomorphological map of the Gaick, providing the basis for establishing the glacial history of the area.

Purpose and scope
The purpose of this contribution is to present and describe a glacial geomorphological map of the Gaick (see Main Map), along with the methods used to compile it. This glacial geomorphological map was produced as part of the first phase of research to elucidate the extent, style and dynamics of former glaciation in the Gaick (see Chandler, 2018). The rationale for producing the Main Map is threefold, namely to provide the foundation for (i) examining processes of ice-marginal and proglacial sediment deposition and their implications for glaciation style and dynamics, (ii) establishing a relative chronology of glacial events in the Gaick using morphostratigraphic principles (cf. Boston et al., 2015;Lukas, 2006), and (iii) reconstructing the maximum , along with boundary of the study area (grey dashed line). Scale and orientation in (a) are provided by British National Grid (10 km intervals). Elevation data is from Ordnance Survey contours. Contains Ordnance Data © Crown Copyright and database rights [2018]. Ordnance Survey Licence no. 100021290. extent, style and (where possible) thickness of prominent phases of glaciation. These three themes will be examined in future publications, drawing on the geomorphological evidence presented here.
Our main research focus is on establishing the extent, style and dynamics of discrete, locally-sourced ice masses (i.e. corrie glacier to ice-cap scale) during the Last Glacial-Interglacial Transition (∼16-8 ka), with special reference to potential Younger Dryas ice masses. To this end, the geomorphological mapping focused on sediment-landform assemblages that could potentially be utilised for glacier reconstruction using established morphostratigraphic procedures (see Boston et al., 2015;Lukas, 2006). These sediment-landform assemblages include inter alia ice-marginal ('hummocky') moraines, ice-contact deposits (e.g. kame terraces), glaciolacustrine deposits (e.g. Gilbert-type deltas), meltwater channels, upslope limits of glaciogenic sediment cover ('drift limits'), terraces, and periglacial landforms (e.g. solifluction lobes, mountain-top detritus, talus cones). Accordingly, geomorphological features pertaining to long-term glaciated landscape evolution (e.g. glacial breaches) were not included on our map. Long-term landscape evolution has a long history of research in the Gaick and interested readers are directed to previous publications (see Hall & Jarman, 2004, and references therein).

Methods
A combination of field mapping and interpretation of remotely-sensed datasets (hereafter 'remote mapping') is typically used to map the geomorphology of alpine and plateau-style ice masses (see Chandler et al., 2018, for review). This permits a holistic approach to mapping that combines the advantages of the different methods and datasets to enhance accuracy and robustness. Following the approach and recommendations synthesised by Chandler et al. (2018), geomorphological mapping of the study area was conducted through a combination of field and remote mapping (Figures 2 and 3). Digitised example of mapping conducted in the field for part of Glen Bruar. Field mapping allowed the distinction between multiple terrace levels that were not identifiable from remotely-sensed data. Note, field mapping of moraines in this study predominantly involved establishing areas of moraines for later detailed mapping using aerial photographs. Mapping of the moraines from the aerial photographs was later verified in the field. Contours were derived from the NEXTMap Great Britain TM dataset (Intermap Technologies®).

Field mapping
Field mapping was performed on topographic base maps with a scale of 1: 10,000 (Figure 2), enlarged from an Ordnance Survey (OS) topographic map (Explorer Sheet 394). The locations of landforms were recorded with reference to 'landmarks' that were clearly identifiable both in the field and on the base maps, following standard field mapping procedures (see Chandler et al., 2018). In some cases, locations were verified using a handheld Global Positioning System (GPS; Garmin eTrex® 10), with a horizontal accuracy of ±3 m. Additional information between these known reference points was then interpolated and marked on the field maps. In addition to contour inflections (which may mark the location and boundaries of prominent ridges, for example), pacing and/or estimation of lengths, heights and widths was also employed to accurately record the size of landforms on the base maps. The detailed field mapping allowed the identification of features that are difficult or impossible to detect on remotely-sensed data, including ice-flow directional indicators (e.g. striae), low-amplitude moraines, upslope limits of glaciogenic sediment ('drift limits') and different terrace altitudes (cf. Lukas & Lukas, 2006).
During field mapping, natural sedimentary exposures were also identified and examined. Detailed analyses of selected exposures were conducted in a targeted approach, although the availability of sedimentary exposures was partly an influence. With the primary focus of our research on the extent and style of glaciation (see Section 2; Chandler, 2018), rather than a spatially extensive examination of a specific landform (e.g. Benn, 1992;Lukas, 2005), sections through landforms either demarcating or in close proximity to inferred glacial limits were mainly selected for investigation. Detailed sedimentological investigations involved section logging, lithofacies analysis, as well as clast shape and structural geological analyses, following established procedures (e.g. Evans & Benn, 2004;Gribenski et al., 2016;Lukas, 2005;Lukas et al., 2013). Detailed discussions of sedimentary processes and their implications are beyond the scope of this paper; thus, section logs, descriptions and interpretations will be presented in future publications.

Remote mapping
Three primary remotely-sensed datasets were used in the production of the Main Map, namely (i) a Digital Surface Model (DSM) from the NEXTMap Great Britain TM dataset (hereafter 'NEXTMap DSM'), (ii) digital colour aerial photographs, and (iii) hard-copy stereoscopic aerial photographs. Additionally, the primary datasets were supplemented by viewing imagery available via Microsoft® Bing™ Maps and three-dimensional visualisation of landforms in Google Earth TM Pro.
The NEXTMap DSM is a georeferenced topographic model that has a pixel size of 5 m. The horizontal accuracy of the DSM is 2.5 m root mean square error (RMSE), whilst the vertical accuracy is 1.0 m RMSE. For mapping and analytical purposes, the NEXTMap DSM data were converted into hillshaded relief models, using the spatial analyst tools in ESRI® ArcMap TM 10.3.1. An illumination angle of 30°and azimuths of 45°and 315°have been suggested as optimal settings for visualisation (e.g. Smith & Clark, 2005), and these recommendations were followed in the present study. The NEXTMap DSM was primarily used for reconnaissance purposes and to generate the topographic data for the Main Map, rather than detailed geomorphological mapping (see Section 4).
Remote mapping was largely conducted using orthorectified, seamless digital aerial photographs with a ground sampled distance (GSD) of 0.25 m per pixel (Getmapping®/UKP), as in recent studies elsewhere in the Scottish Highlands (e.g. Finlayson et al., 2011;Pearce, Rea, Bradwell, & McDougall, 2014). The aerial photographs were made available as preprocessed, orthorectified products, with the processing procedures and errors unknown. To circumvent any potential artefacts/errors in the dataset, a quality assurance check was made by cross-checking the mapping with the NEXTMap DSM and OS topographic data. Mapping from the digital aerial photographs was performed by digitally drawing geomorphological features directly in ESRI® ArcMap TM , with individual vector layers created for each feature ( Figure 3).
Analogue (hard-copy) aerial photographs (∼1: 24,000 scale) from the All Scotland Survey 1988/89 collection were also examined using a mirror stereoscope with 3x magnification. These panchromatic, vertical aerial photographs generally show a good tonal contrast that is typically superior to that of the Getmap-ping® digital aerial photographs, thus making them valuable tools in locating small glacial landforms, such as moraines (e.g. Boston, 2012aBoston, , 2012bLukas, 2003;Lukas & Lukas, 2006). Unfortunately, the aerial photographs did not provide complete coverage of the study area; thus, this dataset was used to check and enhance mapping in selected valleys/corries (where photographs were available). Mapping from the analogue aerial photographs was conducted on acetate overlays, which were scanned and then georeferenced to prominent reference points on the digital aerial photographs using the georeferencing tools in ESRI® ArcMap TM .

Map production
To produce the Main Map, an iterative process was employed, involving several consultations of the field mapping and remotely-sensed data ( Figure 4). This provided a robust approach to mapping that has broadly been employed in several previous studies in Scotland (e.g. Benn & Ballantyne, 2005;Boston, 2012aBoston, , 2012bFinlayson et al., 2011;Lukas & Lukas, 2006;Pearce et al., 2014). The process of assimilating the data from the various methods is outlined below.
The initial reconnaissance stages of mapping involved a systematic examination of the NEXTMap DSM, along with consultations of topographic, geological and existing geomorphological maps. Given the superior detectability of landforms on aerial photographs and in the field, no independent mapping sensu stricto was undertaken using the NEXTMap DSM to save duplication of effort. However, the latter dataset is useful for identifying meltwater channels (e.g. Evans, Hughes, Hansom, & Roberts, 2017;Hughes, Clark, & Jordan, 2010) and was thus used in tandem with the digital aerial photographs for this purpose.
Following the initial preparatory stage, reconnaissance-level mapping of the study area was conducted using Bing™ Maps imagery before the first field season. This field season initially focused on identifying areas for detailed mapping and route planning, before targeted 1: 10,000 scale field mapping was undertaken. Field mapping and aerial photograph interpretation were then conducted in an alternating process: the mapping was performed such that areas specifically targeted for field mapping were subsequently mapped remotely. This enabled consultation of field maps throughout the remote mapping process to ensure all features identified in the field were mapped and to minimise any errors which may arise from misinterpretation of features (cf. Chandler et al., 2018).  (4) were carried out alternately as part of an iterative process, as shown by the two-way arrows.
The geomorphological mapping was finalised through aerial photograph interpretation, in consultation with field maps and earlier versions of remote mapping. In particular, moraines, meltwater channels and dissected glaciogenic material (see Table 1) were mapped almost exclusively using the aerial photographs, since these offered the best means of accurately mapping their position and planform (cf. Boston, 2012aBoston, , 2012bLukas & Lukas, 2006). Conversely, other landforms were more easily and readily mapped in the field (e.g. river terraces, 'drift limits'). Field mapping was incorporated into the Main Map by digitising scanned field maps. Thus, the final map is the product of an iterative process that assimilated mapping from several approaches, exploiting their respective advantages ( Figure 5).
The Main Map was created using a combination of GIS and graphics software. Prior to exporting the mapping and map layout from ESRI® ArcMap TM , the position of the final geomorphological mapping was crosschecked against the NEXTMap DSM and manually adjusted to match the DSM as necessary. This is to enable direct comparison and future integration of the mapping into databases of glacial landforms in Britain, which commonly use the NEXTMap DSM as a base layer (e.g. Bickerdike et al., 2016;Clark et al., 2018). Following adjustment of the mapping, contours were generated from the NEXTMap DSM and then smoothed using the cartographic tools in ESRI® Arc-Map TM . The final map layout was initially designed in the GIS software before export to Adobe® Illustrator TM CC, along with all the individual vector layers (geomorphological features and contours). The focus at this stage was on the final map design and ensuring optimal map presentation.

Geomorphology
The 1: 46,000 scale geomorphological map of the Gaick (Main Map) records a variety of glacial sediment-landform assemblages that are defined in Table 1 and summarised below. These include moraines, dissected glaciogenic material, meltwater channels, and glaciolacustrine deposits. The Main Map also includes associated fluvial and periglacial landforms (e.g. river terraces, solifluction lobes), which are potentially useful for differentiating former distinct periods of glaciation (cf. Lukas, 2006).

Moraines
Several valleys in the Gaick, along with areas around its periphery (e.g. Glen Geldie), contain an abundance of moraines (Figures 3 and 6; Main Map). Most of the moraines display clear spatial patterns in which the crestlines of mounds and ridges can be extrapolated to form chains that trend obliquely downslope, frequently forming arcuate or chevron-shaped patterns when linked to moraines on the opposite side of the valley. Thus, most of the moraines are interpreted as ice-marginal (or latero-frontal) moraines, recording former positions of glacier margins (e.g. Bennett & Boulton, 1993a, 1993bBoston, 2012b;Lukas and Benn, 2006).
The moraines in the Gaick could all be classified under the geomorphological umbrella term 'hummocky moraine', as was previously done by Sissons (1974) and by many authors in other regions of Scotland (see Lukas, 2005Lukas, , 2006, and references therein). However, 'hummocky moraine' is not useful as a descriptive term due to potential confusion with the genetically-distinct 'hummocky moraine' deposited by ice sheet lobes through widespread in situ stagnation (cf. Eyles, Boyce, & Barendregt, 1999;Hoppe, 1952;Lukas, 2005). Moreover, ascribing the term 'hummocky moraine' would be an oversimplification that does not recognise the variations in moraine patterns across the Gaick, which are outlined below.
Overall, two broad moraine patterns can be distinguished in the Gaick. Firstly, in the lower parts of the southern valleys (e.g. Glen Edendon, Glas Choire), topographically-discordant and cross-cutting moraines are observed (Figure 6(d)). Moraines in this category descend obliquely downslope in an upvalley direction (i.e. they are topographically-discordant) and occasionally curve towards the centre of the valley. Crosscutting and overprinted moraines are also present in the northeastern Gaick, documenting eastwards (downvalley) retreat of ice in Glen Geldie. The topographically-discordant and overprinted moraines are inconsistent with outward radial flow of local ice masses from the Gaick and therefore reflect the interplay of regional ice and local, Gaick, ice.
The second, and dominant, spatial pattern observed in the Gaick is that of topographically-concordant moraines (i.e. they descend downslope in a downvalley direction) organised as inset transverse chains. These chains of moraine mounds and ridges can be connected with chains on the opposite slopes to form nested concentric arcuate loops. Moraines of this type occur in the higher parts of the catchments, sometimes continuing to the very upper parts of the catchments and to the plateau edge, providing evidence for plateau-sourced ice (cf. Boston, 2012b;Boston et al., 2015;McDougall, 2001). This spatial pattern has been observed in many areas throughout upland Britain (e.g. Benn & Ballantyne, 2005;Bennett & Boulton, 1993a, 1993bBenn, Lowe, & Walker, 1992;Boston, 2012aBoston, , 2012bFinlayson et al., 2011;Lukas and Benn, 2006;McDougall, 2001McDougall, , 2013. Within the second overarching category, further variations in the moraine pattern and landform associations are recognised. The outermost moraines in the second category may be restricted to a narrow band at the slope foot and do not trend up the valley slopes in the lower to middle parts of the Gaick valleys, with chains of moraines exhibiting very shallow average gradients (often ∼5 m per 100 m). Conversely, in the higher parts of the valleys, transverse chains of moraines often have considerably steeper average gradients (∼15-20 m per 100 m) and are found ascending the valley slopes. Additionally, there is typically a clear contrast in the slope features, with the shallower, outer moraines associated with scree-or talus-covered Figure 5. Example from part of Glen Bruar illustrating the assimilation of the field mapping (a and b) and aerial photograph interpretation (c and d) to produce the final geomorphological map (e). Key for (b) is the same as in Figure 2. Note, terrace levels were simplified from the field mapping for the final map. The aerial photograph extract in (c) is from the Getmapping® digital, orthorectified aerial photographs. Contours were derived from the NEXTMap Great Britain TM dataset (Intermap Technologies). . Deposition as Gilbert-type deltas and grounding-line fans . Indicative of former ice-damming Benn and Evans (1993), Benn (1996), Brazier, Kirkbride, and Gordon (1998) Brazier, Whittington, and Ballantyne (1988), Ballantyne (2002bBallantyne ( , 2008 Alluvial fans . Fan-shaped sediment accumulations located at stream junctions . Typically, gentler, low gradient forms than debris cones . Alluvial/fluvial deposits . Reflect lateral spreading out of alluvial/ fluvial sediment accumulations at stream junctions, broadening out from a central channel Brazier (1987), Summerfield (1991), Ballantyne (2002b) JOURNAL OF MAPS slopes. Generally, no clear distinction can be made between moraines purely based on morphological characteristics.

Dissected glaciogenic material
A distinctive feature of several valleys in the Gaick is the presence of spreads of 'dissected glaciogenic material' and intervening channels, with exemplar assemblages evident in Glen Chulaibh, Glas Choire and Glen Tarf (Figure 7; Main Map). This sedimentlandform assemblage, as defined in Table 1, has not previously been reported or mapped in the Scottish Highlands. It exhibits a jigsaw-like erosional planform, where a pre-existing, undulating glaciogenic sediment cover has been dissected into mounds and ridges by fluvial activity (i.e. the positive-relief elements are erosional, rather than constructional, products). In nearly all cases, dissection was the result of glacial meltwater activity, although pronival (snowmelt), paraglacial and/or postglacial fluvial activity is also likely to have had some influence on the final form of these sediment-landform assemblages. Paraglacial/postglacial processes (e.g. downslope movement and 'spreading') has led to final forms that are morphologically similar to moraines and ice-contact deposits. However, dissected glaciogenic material, as defined in this paper, can be clearly differentiated from moraines and icecontact deposits (i.e. 'constructional' landforms) based on their erosional, jigsaw-like planform (see also Table 1). Analyses of two sedimentary exposures reveal that the positive surface expression of the dissected surfaces (i.e. the mounds and ridges) in Glas Choire comprise facies consistent with proximal proglacial outwash (Chandler, 2018). The sedimentological evidence, together with the incision of the outwash sediments by shallow meltwater channels, therefore suggest that the sediment-landform assemblage in Glas Choire was formed in a proximal glaciofluvial environment. This contrasts with available sedimentary evidence from moraines in the Gaick that demonstrates they contain stacked debris flows and intercalated fluvial 'wash' horizons, with varying degrees of proglacial and subglacial glaciotectonism evident (see Chandler, 2018). Thus, the moraines were constructed as terrestrial ice-contact fans at active glacier margins (cf. Benn, 1992;Boston, 2012b;Lukas, 2005).
The recognition of dissected glaciogenic material in the Gaick potentially has important implications for elucidating the extent and dynamics of former glaciation in the area, since valleys in the Scottish Highlands are typically characterised by ice-marginal moraines indicative of active, oscillatory glacier retreat (e.g. Benn et al., 1992;Bennett & Boulton, 1993a, 1993bBoston, 2012aBoston, , 2012bLukas and Benn, 2006). By contrast, the dissected glaciogenic material indicates widespread erosion of pre-existing glaciogenic sediment cover, rather than construction of successive, discrete icemarginal landforms (moraines) during retreat. This is a very different sediment-landform and glaciodynamic signature, possibly akin to fluvially-dissected sediment masses evident at High Arctic glaciers on Svalbard (cf. Lovell & Boston, 2017). The potential glaciodynamic implications of dissected glaciogenic material will be examined in detail in a future publication.

Ice-contact deposits
This sediment-landform assemblage incorporates variously-shaped mounds, ridges, terraces and fans that are inferred to have been formed in intimate association with a glacier margin and deposited as positiverelief elements by glaciofluvial activity, i.e. these features are constructional (as opposed to dissected glaciogenic material: Section 5.2). A prominent area of ice-contact deposits occurs on the slopes of Fuar Mhonadh to the southwest of Coire Mhic-sith (NN 646 753). In this area, numerous flator dome-topped mounds and ridges occur (Figure 8; see also Section 5.6). These depositional landforms are ∼15-150 m long, ∼15-75 m wide and <2 m high, and chains of these mounds and ridges trend relatively uniformly across the hillside at very low angles. The ice-contact deposits are found in intimate association with several single-sided lateral meltwater channels that descend the hillside at similarly low angles, but there are numerous instances where the meltwater channels make sharp (nearly 90°) turns and plunge down the slope (see Section 5.6). Based on the arrangement and trend of the ice-contact deposits, and their association with ice-marginal meltwater channels, these landforms are interpreted as kame terrace fragments (cf. Brodzikowski & van Loon, 1991;Evans et al., 2017;Livingstone, Evans, Cofaigh Ó, & Hopkins, 2010;. A glacial origin for the features described above has previously been contested. Benn and Ballantyne (2005, p. 582) argued that they were 'produced by rock-mass creep, and the intervening "channels" are tensional crevices formed by downslope block displacement' (i.e. the assemblage is a rock slope failure). However, no hitherto-recorded rock slope failures in the Gaick and mapped in this study (see Main Map) bear any similarity to this area, all of which occur in the psammites that dominate the bedrock geology of the Gaick (cf. Jarman, 2004). There are also no clear comparators elsewhere in Scotland (cf. Jarman, 2006). Notably, there are no strong indicators of failure, with no source scar/ cavity, toe bulge or dislocation of drainage evident on the slopes of Fuar Mhonadh (cf. Holmes, 1984;Jarman, 2006). The simplest interpretation of the channels and intervening mounds is that they are meltwater channels and kame terrace fragments, as argued above.

Glaciolacustrine deposits
Evidence for the existence of former ice-dammed lakes in the Gaick is recognised in Coire Mhic-sith (NN 655 746) and at the site of the present-day Loch an Dùin (NN 723 801). In Coire Mhic-sith, evidence is in the form of glaciotectonised glaciolacustrine sediments, lake shorelines and lake overflow channels (cf. Benn & Ballantyne, 2005;Chandler, 2018;Phillips, Merritt, Auton, & Golledge, 2007). Northwards of the present-day Loch an Dùin, evidence for a former ice-dammed lake is found in the form of a Gilbert-type delta and a glaciotectonised groundingline (or subaqueous outwash) fan (Chandler, 2018).

Drift limits
Distinct upslope limits of glaciogenic sediment cover ('drift limits';cf. Lukas, 2006, and references therein) are rare in the Gaick, with only a few evident. The two most prominent examples occur in Coire Chais (NN 696 840) and Gaick Pass (NN 733 823) (see Main Map). In Coire Chais, an upslope limit of the sediment cover, which ascends from ∼550 m OD to ∼610 m OD on the southwest-facing slope, can be traced from the outer limit of small, low-amplitude (<2 m high) and indistinct moraines. This 'drift limit' coincides with the downslope limit of extensive talus deposits. A conspicuous upslope limit of glaciogenic sediment in Gaick Pass grades from the outer moraines on the northern side of the valley, ascending northwestwards into Coire an Dubh-chadha and reaching ∼580 m OD. This distinct upper limit coincides with the downslope limit of extensive screes beneath Creag an Dubh-chadha.

Meltwater channels
A large number of meltwater channels occur in the Gaick, both within the valleys and on the plateau surfaces. Many of the meltwater channels have been identified as ice-marginal (lateral) channels based on their orientation and association with other landforms: these features have downslope trending profiles and often intervene chains of moraines (see Section 5.1), occasionally curving towards the valley centre (cf. Boston, 2012aBoston, , 2012bGreenwood, Clark, & Hughes, 2007). Predominantly, the lateral meltwater channels are shallow (frequently ∼1-5 m deep), although they do attain depths of >15 m in some instances. For example, in Glen Edendon, there is a pronounced meltwater channel that is up to ∼20 m deep and ∼40 m wide (NN 715 765). Meltwater channel length is highly variable across the Gaick, ranging between ∼50 m and ∼1000 m (average ∼170 m).
Subglacial meltwater channels have also been identified on the plateau, in the vicinity of An Sligearnach (786 m OD; NN 951 782) (Figure 9). The meltwater channels are small (often ∼1 m deep and ∼2 m wide) and form part a southwestwards-directed system that has a maximum altitude of ∼830 m OD and descends to the valley bottom at ∼650 m OD. In planform, these channels are arranged as a complex network of sinuous and anastomosing channels that frequently exhibit abrupt beginnings and terminations. Occasionally, some of the channels also take sharp (∼90°) turns, with the new channel course heading directly downslope (Figure 9). These are all characteristics consistent with subglacial meltwater channels (cf. Greenwood et al., 2007;Sissons, 1961).
On the slopes of Fuar Monadh, near Coire Mhicsith, there are also meltwater channels that initially formed as ice-marginal channels before descending into the subglacial realm (NN 646 753). This is indicated by their oblique trend across the hillside before abrupt, sharp turns (∼90°) to descend directly downslope (cf. Livingstone et al., 2010;Sissons, 1961). Many of these channels also exhibit undulating long profiles, which is consistent with meltwater channels formed under hydrostatic pressure, i.e. subglacially.

Fluvial landforms
Several fluvial landforms have been recognised in the study area, namely alluvial fans and river terraces. Notable examples of alluvial fans occur in Gaick Pass, with a prominent fan emanating from Allt Bhrodainn and splaying out across the valley floor (NN 750 837) (Main Map). In some instances, the alluvial fans partially submerge and obscure moraines.
River terraces occur in a number of the glens in the Gaick, with a particularly well-developed river terrace sequence found in Glen Bruar (NN 830 764) (see Figure 5). Low river terraces (∼1-2 m above the current floodplain) extend into the upper parts of some valleys.

Periglacial landforms
Periglacial features, principally in the form of solifluction lobes and mountain-top detritus, occur on the majority of summits in the area and around the periphery of the plateau surfaces. There is a conspicuous change in the periglacial assemblages on the higher ground from west to east across the Gaick. In the west, solifluction lobes are found on the spurs surrounding the plateau, descending to elevations of ∼700 m OD. Discontinuous spreads of mountain-top detritus (frost-shattered regolith) occur around the periphery of the plateau. East of Gaick Pass, solifluction deposits and mountain-top detritus become increasingly prevalent on the higher ground ( Figure 10; Main Map). The detritus comprises blockfields on some summits in the southeastern and northeastern sectors of the Gaick, including on Beinn Dearg (1008 m OD; NN 852 777) and An Sgarsoch (1006 m OD; NN 933 836). Frost-shattered bedrock outcrops are also encountered in the eastern, but not in the western, parts of the Gaick. Thick accumulations of talus/ scree can be found in several valleys in the study area, sometimes extending to the slope foot. Extensive deposits occur in the Gaick Pass, Gleann Diridh and Gleann Mhairc.

Rock slope failures
Several rock slope failures (RSFs) have been identified in the Gaick, with most of them clustered in the Gaick Pass and in the vicinity of Loch an Dùin (Main Map). These include arrested to sub-cataclasmic slides and compressional slope deformation type RSFs, following criteria defined by Jarman (2006). An exemplar compressional slope deformation is recognised on the flank of A' Chaoirnich (NN 730 810) (see also Jarman, 2004Jarman, , 2006, which has similarities with the largest RSF in the Scottish Highlands, Beinn Fhada (see Jarman & Ballantyne, 2002). The RSFs recorded on the Main Map have been reported previously (see Holmes, 1984;Jarman, 2004).

Conclusions
The glacial geomorphological map presented here provides a detailed record of the glacial landforms in the Gaick that can be used to significantly enhance our knowledge of glacial and climatic events in the region. The spatial distribution of the glacial sediment-landform assemblages in the Gaick provides compelling evidence for at least one phase of local (valley or icefield) glaciation and the 'unzipping' of Gaicksourced ice and external (regional) ice. Strong evidence for the influence of regional ice lobes includes the presence of moraines and lateral meltwater channels that are incompatible with outward radial flow from the Gaick (e.g. topographically-discordant moraines in valley bottoms), whilst 'unzipping' events are documented by cross-cutting of moraines. Further upvalley, suites of topographically-concordant moraines and meltwater channels are present and provide a record of former valley/outlet glacier fluctuations. This geomorphological evidence, together with variations in associated fluvial and periglacial landforms, will be employed to establish a morphostratigraphic framework for the Gaick glacial events in a subsequent publication. In turn, this will provide the important foundation for establishing a glacial chronology, Figure 10. Example images of the periglacial landforms found on the higher ground in the southeastern sector of the Gaick. (a) Overview photograph of the higher ground from the slopes of Carn a' Chlamain, looking towards the northeast. Carn a' Chlamain is strewn with mountain-top detritus (M). Mountain-top detritus, along with frost-shattered bedrock outcrops (F), is evident on the summits of Conlach Bheag (middle ground) and Conlach Mhòr (background). Much of the surrounding ground is covered by peat. reconstructing the former ice masses, and assessing former glacier dynamics and palaeoclimatic conditions. Software ESRI® ArcMap TM 10.3.1 was used to (i) visualise and, where necessary, process the remotely-sensed datasets, (ii) perform on-screen geomorphological mapping, and (iii) assimilate the field and on-screen mapping. The mapping was exported to Adobe® Illustrator TM CC for final map production.

Data availability
The data that support the findings of this study are available from the corresponding author, BMPC, upon reasonable request.