The glacial origins of relict ‘pingos’, Wales, UK

Ramparted depressions (doughnut–shaped debris-cored ridges 13 surrounding peat-filled basins) are commonly perceived to represent the relict 14 collapsed forms of permafrost ground-ice mounds (i.e. pingos or lithalsas). In 15 Wales, UK, ramparted depressions of Late Pleistocene age have been widely 16 attributed to permafrost-related processes. However, a variety of alternative 17 glacial origins for these enigmatic landforms are also consistent with the avail18 able geological and geomorphological evidence, although previous studies have 19 barely considered such alternative processes of formation. From detailed geo20 physical, sedimentological and remote sensing studies at multiple field sites, 21 we present and assess the hypothesis that glacial processes, associated with 22 the wastage of stagnating glacier ice were responsible for the formation of ram23 parted depressions in Wales. Our findings demonstrate that: (i) glacial, not 24 periglacial, processes are the most likely cause for many ramparted depressions 25 in Wales; (ii) ramparted depressions have significant potential for character26 ising the nature of deglaciation around the margins of the Irish Sea during 27 the last glacial cycle; and (iii) future interpretation of ramparted depressions 28 within formerly glaciated terrains must carefully evaluate all possible (glacial 29 and periglacial) mechanisms of formation. 30


31
The collapsed pingos (hereafter 'ramparted depressions') (Watson, 1971; Watson and Watson, 1974; Gur-32 question is yes, then these landforms are a significant overlooked resource that can provide insight into the 91 properties, behaviour, dynamics, limits and potential interactions of Welsh and Irish Sea ice masses during 92 the late Quaternary.

94
Electrical resistivity tomography 95 Electrical resistivity measurements were acquired using an IRIS Instruments SYSCAL Junior Switch 72. 96 This system uses a multi-channel switching unit to control up to 72 steel electrodes to induce current 97 into the subsurface and record the voltage response. Electrodes were inserted into the ground at regular  Seismic refraction 105 Seismic refraction data were acquired using a Bison 9000 24-channel seismograph. A spread of 24 Geosource 106 100 Hz geophones were used to record the data, at a spacing of 2 m, resulting in refraction lines 46 m in 107 length. The seismic source was generated by between five to eight stacked sledgehammer blows on a metal 108 plate. Shot points were located at various distances on and off the geophone spread. To determine the depth 109 and dip of the refractors, multiple offset shot points were positioned in both forward and reverse directions 110 (reversed profile refraction technique) so that multiple shots were recorded for each spread. Processing 111 of the seismic data was performed in Reflexw, with the first arrival times (first negative deflection on 112 the seismogram) for each geophone picked. Direct and refracted events were identified from travel-time 113 graphs and arrivals were assigned to the direct wave and specific refracting horizons. Best-fit lines were 114 produced for each velocity segment by linear regression. Processing used the Common Receiver Method 115 (Hagedoorn, 1959). The aim of the seismic data acquisition was to determine: (i) sub-surface sediment 116 (acoustic) properties; and (ii) depth-to-rockhead. In addition, the seismic data were used to calibrate 117 the ERT measurements of rockhead, which were easier and quicker to acquire, particularly over peaty waterlogged basins. Details on the processing of both the seismic and resistivity data are available in Ross 119 (2006).

120
Coring and sedimentology 121 An Atlas Copco Cobra vibro-coring system was utilised to drill boreholes for sedimentological analysis. 122 Cores were logged in the field through the open windows of the core barrels, using a Munsell chart to 123 describe the colour of the sediments. Representative disturbed samples were collected for laboratory grain-124 size analysis. The aim of the sedimentological observations was to constrain geophysical measurements and 125 to determine near-surface sediment properties directly. Trial pits of depth 1-2 m were excavated using a 126 mini-digger.  144 The upper 4-4.5 m of Borehole 1 and Borehole 3 ( Figure 3) were dominated by poorly sorted, well-graded, 145 compact, non-calcareous gravelly silt to silty gravel diamictons. The gravels comprised subangular to 146 rounded mudstone and sandstone clasts, with long axes up to 7 cm in length. Underlying these upper 147 sediments, more than 0.5-1 m of non-calcareous clayey silt was found beneath 4.75 m in Borehole 1 and 148 3.91 m in Borehole 3. This lower unit was characterised, in parts, by very fine, faint laminations (e.g. BH3   170 Three distinct P-wave first break velocity segments (i.e. direct wave and two refracted waves) were recog-171 nised in both the forward and reverse directions of the travel-time graph (Figure 5a and 5b). The average 172 P-wave velocity of the direct wave was 1000 msec -1 . The P-wave velocity of the first refracted wave had 173 an average of 1830 msec -1 , whilst the second refracted wave had a velocity of 3150 msec -1 ( Figure 5).

174
The observation of two refracted waves from the first break data at Rhos Llawr Cwrt indicates a three-

223
Our sedimentological and geophysical data demonstrate that the Cletwr and Cledyn valleys are charac-224 terised by thick (>20 m) sequences of superficial deposits dominated by a mixture of fine-grained silts and 225 compact and overconsolidated diamict. The compact nature of the latter, and the common presence of 226 heavily striated clasts within it, indicates that this deposit is a glacial till, with components that have 227 undergone transport at the base of an ice mass. We do acknowledge that under permafrost conditions 228 mass movement by slope processes is enhanced, often leading to the production of sediments similar in 229 nature to glacial till. However, the gently-sloping hills that characterise the interfluves of these valleys, 230 and the physical properties of the material, suggest that emplacement of this sediment was directly by 231 glaciation, and not by landscape-scale slope modification of older glacial deposits followed by deformation 232 by massive ground-ice (e.g. Watson and Watson, 1974). Our interpretation of the fine-grained silts is that  (Figures 1 and 9).
Our sedimentological investigations were unable to identify significant thicknesses of laterally exten-250 sive glacio-lacustrine silts in the Cledyn valley necessitated by the lithalsa model (e.g. Gurney and Worsley 251 1996). Although our resistivity data, and that of Harris (2001), suggest highly conductive materials be-252 neath the Cledlyn landforms, it is not possible from the geophysical data alone to define lithology. Though 253 they could be consistent with a thick sequence of glacio-lacustrine silts, the resistivity values are equally 254 interpretable as water-saturated glacial tills with a silty-clayey matrix (Reynolds, 2011), a material which 255 we directly evidenced extensively in our trial pits and in boreholes from the Cledlyn valley (Ross, 2006). 256 We did observe fine-grained glacio-lacustrine material at depth beneath the ramparted feature investigated The geomorphological data extracted from the LiDAR-derived DEMs (Figures 1, 2, 6  During deglaciation it is highly unlikely that any ice that had flowed onto the higher elevation area would  Figure 9) would also be 316 likely due to marginal and subglacial processes such as sediment squeezing into basal crevasses.

317
Though existing observations and data do not allow us to confirm whether Irish Sea or Welsh ice was 318 responsible for landform formation, the orientation of most of the ridges (e.g. Figure 9 and in the Cletwr  Gurney (1994,1995) and Gurney and Worsley (1996) believe that these sediments were deposited in an 331 extensive proglacial lake prior to landform (lithalsa) formation, there is no evidence for a thick sequence  cycle, invoking periglacial processes for these landforms seems overly complex. We argue that the simplest 384 explanation for these landforms is that they are glacial in origin, and formed by processes associated with 385 ice margin stagnation, and the meltout of debris-rich buried ice. These findings have broader implications 386 for the palaeoglaciology of the broader Irish Sea basin, as these landforms provide an extensively overlooked 387 resource of glacial conditions relating to both the Irish Sea and Welsh ice masses during the last glacial cycle. 388 We acknowledge that our interpretation presents a bold hypothesis, that requires future field investigations.

389
However, for the first time such field investigations can proceed with an ice-stagnation hypothesis to test.

390
What is unequivocal is that there is much remaining to be discovered about the glacial history of west Wales