Little Ice Age glaciers in Britain: Glacier–climate modelling in the Cairngorm Mountains

Abstract

It is widely believed that the last glaciers in the British Isles disappeared at the end of the Younger Dryas stadial (12.9–11.7 cal. kyr BP). Here, we use a glacier–climate model driven by data from local weather stations to show for the first time that glaciers developed during the Little Ice Age (LIA) in the Cairngorm Mountains. Our model is forced from contemporary conditions by a realistic difference in mean annual air temperature of −1.5°C and an increase in annual precipitation of 10%, and confirmed by sensitivity analyses. These results are supported by the presence of small boulder moraines well within Younger Dryas ice limits, and by a dating programme on a moraine in one cirque. As a result, we argue that the last glaciers in the Cairngorm Mountains (and perhaps elsewhere in upland Britain) existed in the LIA within the last few hundred years, rather than during the Younger Dryas.

Keywords

British Isles Cairngorms Little Ice Age glaciers modelling

Introduction

There has long been a consensus that the last glaciers in the British Isles disappeared at the end of the Younger Dryas stadial (12.9–11.7 cal. kyr BP; e.g. Bennett and Boulton, 1993; Golledge and Hubbard, 2005; Sissons, 1979, 1980; Sissons et al., 1973). This paradigm holds that all moraines in the British Isles relate to this period of cooling or earlier and that ice-free conditions have existed throughout the Holocene. However, up until the 1970s, there was speculation that sheltered cirques in some mountain massifs in northern and eastern Scotland may have developed glaciers more recently than Younger Dryas stadial times, although the evidence for this has hitherto been circumstantial. Much of this speculation derived from the presence of long-lasting snow patches in sheltered locations (e.g. Bonacina, 1938; Dansey, 1905; Hudson, 1977; Manley, 1949, 1971a, 1971b) and the suggestion from travellers’ accounts that significant snow cover existed in the Cairngorm Mountains, eastern Scotland, during the Little Ice Age (LIA) between the 16th and 19th centuries (Lamb, 1977). This was a period of pronounced climatic cooling in parts of the Northern Hemisphere (especially northern Europe) characterised by glacier expansion, increased flooding and reduced agricultural productivity (e.g. Anderson et al., 2000; Macklin et al., 2005; Matthews and Briffa, 2005; Mann et al., 2009).

Sugden (1977) made the first assessment of the likelihood of small glaciers existing in the Cairngorms during the late Holocene and argued that a difference in temperature relative to contemporary conditions of −1.6°C to −2.0°C would have been sufficient to generate glaciers in the highest cirques above 900 m a.s.l., a figure sustained at various times during the late Holocene and in particular the LIA. Sugden further argued that small boulder moraines in several of the high Cairngorm cirques exist within mapped Younger Dryas ice limits and may therefore represent the downvalley extent of later glaciers (see Rapson, 1990 and Figure 1a and b).

Figure 1.

(a) Location maps showing the Cairngorm Mountains, British Isles and the location of major cirques. Mapped Loch Lomond Stadial (Younger Dryas) glacier limits are from Sissons (1979). Water bodies are shaded grey. The location of the automatic weather station at Cairngorm is shown, and the Braemar weather station is 21 km to the southeast of the study area. (b) Photographs of boulder moraine ridge in Garbh Choire Mòr (arrowed). Left photograph is taken looking south (photographs taken by E Boyle). The limit and location of the boulder moraine ridge shown in Figure 1(b) at the head of Garbh Choire Mòr is arrowed in Figure 1(a).

We test the hypothesis that the cirques of the northern Cairngorm Mountains supported LIA glaciers by using a glacier–climate model driven by plausible differences in LIA temperature and precipitation from contemporary values. We demonstrate the likelihood that small cirque glaciers did exist during the LIA in the Cairngorms. A companion paper (Kirkbride et al., 2013) further supports these assertions by demonstrating the presence of a late-Holocene moraine ridge in one of our modelled cirques, at Coire an Lochain (Figure 1a).

Study sites

The Cairngorm Mountains reach an altitude of 1309 m a.s.l. at Ben Macdui. Mean annual air temperature (MAAT) at Cairngorm (1245 m a.s.l.) is 0.5°C, and temperatures average −1.7°C in the winter months (October–March). Precipitation in summit regions exceeds 2000 mm annually, and at present, ~30–40% of the winter precipitation falls as snow (Dunn and Colohan, 2009). These contemporary conditions allow semi-permanent snow patches to exist in several of the high-altitude cirques on the northern and northeastern flanks of the Cairngorm massif, and we focused our research on these cirques because they are climatically the most likely to have nurtured small glaciers during the LIA (e.g. Manley, 1971a, 1971b; Watson et al., 2009). The east- and northeast-facing cirques around Garbh Choire Mòr, south of Braeriach, at Loch Coire an Lochain to the north of Braeriach, and to the west of Cairngorm in Coire an Lochain and Coire an t-Sneachda (Figure 1a), have developed in the lee of prevailing westerly winds, and under present conditions, these contain the longest lying snow patches in the British Isles (Figure 1b). The most persistent snow is found in Garbh Choire Mòr where the snow patches (named Michelmas Fare, Sphinx and Pinnacle after the rock climbs above them) appear to have only melted six times since the end of the 19th century (Cameron et al., 2012; Watson et al., 2004). Some 130 m in front of these snow patches is a well-defined ridge composed of angular boulders crossing the cirque floor (Figure 1b). This ridge is almost 2 km within mapped Younger Dryas ice limits (Sissons, 1979; Figure 1a). Similar boulder ridges are found in other Cairngorm cirques within Younger Dryas limits (Rapson, 1985).

Glacier–climate modelling

Glacier reconstructions were carried out using a two-dimensional (2D) glaciological model (Plummer and Phillips, 2003). The glaciological model considers atmospheric and topographic controls on surface energy balance, including radiative fluxes based on contemporary solar position and topographic shading, and avalanching of snow from oversteepened hillslopes, and as such is appropriate for reconstruction of small cirque glaciers (Plummer and Phillips, 2003). The model domain was defined using the ASTER GDEM v.2 digital elevation model (DEM) with a 25-m grid spacing for each of the cirques under investigation (Garbh Choire Mòr, Loch Coire an Lochain, Coire an Lochain and Coire an t-Sneachda; Figure 1a). A description of the model parameterisation is given in Table 1. We forced the model with monthly mean meteorological data from two local automatic weather stations. Precipitation data were collected from the Meteorological Office Station at Braemar (operational from 1959 to present and 21 km to the southeast of our study area), and temperature and wind speed data were collected from the Heriot-Watt University Cairngorm station (operational from 1977 to present). We simulated glaciers resulting from differences in MAAT from contemporary conditions (hereafter referred to as ΔT) of up to −3°C and an increase in annual precipitation amount of either 10% or 20% from contemporary conditions to investigate plausible LIA palaeoclimate scenarios. The key results for the glacier–climate sensitivity analyses are presented in Figure 2.

Table 1.

Parameterisation of the glacier–climate model used in this study. Climatological parameters are given as mean annual values. Where monthly values were used, these are also given as seasonal means for summer (April–September) and winter (October–March). For a description of the application of the model, the reader is referred to Plummer and Phillips (2003).

Model domain description		Climatological parameters	Annual	Summer	Winter
Native horizontal grid spacing of ASTER GDEM (m)	30	Monthly sea-level air temperature (°C)	8.0	10.3	5.8
Vertical grid spacing of ASTER GDEM (m)	1	Standard deviation of temperature (°C)	1.5	1.5	1.4
Grid spacing of the model domain (m)	25	Atmospheric lapse rate (°C/km)	−7	–	–
Model domain grid size (number of cells)	571 × 528	Critical temperature for snowfall (°C)	2	–	–
		Precipitation maximum (mm)	3228	1548	1680
		Precipitation minimum (mm)	1860	864	996
Glaciological parameters		Precipitation mean (mm)	2522	1195	1327
High albedo	0.8	Precipitation standard deviation (mm)	296	148	148
Low albedo	0.4	Snowfall mean as SWE (mm)	1509	364	1145
Maximum slope that can hold snow (°)	30	Snowfall standard deviation as SWE (mm)	390	134	260
Slope increment for avalanching routine (°)	12	Wind speed (m/s)	15.4	13.7	17.1
Minimum new snow for avalanching to occur (m)	0.1	Cloudiness	0.75	–	–
Deformation constant (yr/kPa³)	2.20E−08	Relative humidity	0.80	0.78	0.82
Integrated balance threshold for steady state (%)	1	Emissivity of snow	0.99	–	–
		Emissivity of the surrounding terrain	0.94	–	–
		Dimensionless transfer coefficient for snow	0.0015	–	–
		Ground heat flux (W/m²)	0.1	–	–

SWE: snow water equivalent.

Figure 2.

Glacier model sensitivity tests. (a) Wind redistribution of snowfall calculated using the model of Purves et al. (1999), and simulated glaciers using the results from Purves’ snow redistribution model for differences in mean annual air temperature (ΔT) from contemporary values of (b) 0°C and (c) −1.5°C. Glaciers simulated for ΔT of (d) −0.5°C, (e) −1.0°C and (f) −2.0°C. Glaciers simulated with increased precipitation (P = 110%) to account for slightly wetter conditions during the LIA, for ΔT of (g) −0.5°C, (h) −1.0°C and (i) −2.0°C. Catchment and model domain boundaries (solid black lines) and contours with 50-m spacing derived from the ASTER GDEM (dotted brown lines) are shown in each case.

We tested how wind transport and redistribution of snow may modify glacier extent by replicating the snow drift model developed for the Cairngorms by Purves et al. (1998, 1999). The Purves model is a simple qualitative model used to derive the relative accumulation and distribution of snow that uses the DEM to derive wind flow over the terrain. Arbitrary units for wind speed and the threshold wind speed for snow transport were set to 15 and 5, respectively, and implemented for a westerly wind direction allowing snow transport over distances of 150 m (we note the distances used by Purves et al. (1998) were 50 m). Simulated snowfall was calibrated using values for snow erosion and deposition from Purves’ model (Figure 2a). The redistribution of snow by wind transport increased ice cover at the ridgelines under both ΔT = 0°C and −1.5°C conditions and reduced the equilibrium line altitude (ELA) by 6 and 30 m, but did not significantly modify the development of cirque glaciers (Figure 2b and c). Glacier sensitivity to snow redistribution was low compared with sensitivity to difference in temperature and topographic controls on mass balance. Therefore, to avoid introducing further uncertainties to our simulations, we did not include the wind distribution of snow in the glacier reconstructions presented in this paper.

Glacier modelling results

Forcing the glacier–climate model with contemporary climate conditions produced simulated minor perennial snow patches in the high-elevation cirques consistent with those observed in Garbh Choire Mòr, and this resulted in a glaciated area of 0.076 km². Minor contemporary snow patches were also simulated in Loch Coire an Lochain and Coire an Lochain (Figure 3a). The eastern Coire an t-Sneachda catchment contains no perennial snow or ice. Glacier sensitivity to difference in temperature was high. ΔT of −1°C reduced the ELA by 60 m from the simulated contemporary value and increased the glaciated area to 2.57 km² (Figure 2e). ΔT of −2°C reduced the ELA by 103 m and increased the glaciated area to 11.42 km² (Figure 2f). Glacier sensitivity to precipitation was lower than to differences in temperature. An increase in precipitation of 10% from contemporary values (from a mean of 2552–2807 mm) increased the glaciated area by 1.03 km² when ΔT = −1°C (Figure 2h), and 3.59 km² when ΔT = −2°C (Figure 2i).

Figure 3.

Simulated glaciers for Garbh Choire Mòr, Loch Coire an Lochain, Coire an Lochain and Coire an t-Sneachda under (a) contemporary (ΔT = 0°C and P = 100%) and (b) LIA (ΔT = −1.5°C and P = 110%) conditions. Catchment and model domain boundaries (solid black lines) and contours with 50 m spacing derived from the ASTER GDEM (dotted brown lines) are shown in each case.

LIA glaciers were simulated with a difference in temperature of −1.5°C and an increase in precipitation amount of 10% from contemporary values to produce a glaciated area of 7.54 km² (Figure 3b). The LIA simulations produced small glaciers within each of the cirques, and perennial snow and ice cover on many of the ridges and headwalls within the model domain. Ice thickness was greatest in the southern cirques, reaching 82 m in Garbh Choire Mòr and 100 m in Loch Coire an Lochain. In Coire an t-Sneachda, glaciers were slightly smaller than in those to the south and maximum ice thickness was 69 m. We attribute this difference in simulated ice to the slightly higher mean cirque floor elevations in the southern cirques.

Discussion

We tested a range of palaeoclimate envelopes to simulate LIA glaciers in the Cairngorms by forcing our glacier–climate model with temperature differences from contemporary values of up to −3°C and increased precipitation of 20%. Glacier sensitivity to temperature was high, indicated by the significant changes in glacier extent with small differences in temperature (Figure 2). Our glacier–climate model indicates that a difference in temperature of −1.5°C and an increase in precipitation of 10% from contemporary conditions are sufficient to develop glaciers in the Cairngorm Mountains. The last time such conditions prevailed was during the LIA. The application of a glacier–climate model will have some uncertainties associated with the representation of palaeoclimate used to drive the model, and the glaciological parameters used. We consider that the ΔT = 0.5°C increments in temperature used here represent a realistic level of precision in the application of this model, and do not attempt to resolve the LIA palaeoclimate at a finer resolution. Topographic shading would have influenced radiative fluxes to the glaciers, and avalanching of snow from steep slopes would have affected mass balance, both of which are likely to modify the extents of small glaciers (e.g. Coleman et al., 2009). However, our energy balance calculation is topographically driven, and as the contemporary topography of the Cairngorms is unlikely to have changed since the LIA, we consider that the use of a DEM representing the current terrain with a description of the contemporary solar position provides a sufficiently precise indication of the topoclimatic influences on these glaciers during the LIA.

We dismiss arguments that the small boulder ridges in several high cirques are protalus, pronival or avalanche landforms (Watson, 2011) based on clast angularity and ridge position. The clearest boulder ridge (in Garbh Choire Mòr; Figure 1b) is too far from the foot of the talus slope at the backwall of the cirque for it to have formed by pronival processes (Ballantyne and Benn, 1994), and small LIA glaciers were probably unable to generate subglacial stresses sufficient to produce rounded clasts, and merely reworked pre-existing rockfall material to form moraine ridges. In addition, the presence of presumed early-Holocene detrital material in several cirque lakes demonstrated by Rapson (1985) also does not argue against the presence of late-Holocene glaciers in such locations as such glaciers were unlikely to have been dynamic enough to excavate pre-existing material from subglacial positions.

The results from a dating programme on one of these boulder ridges in Coire an Lochain support our hypothesis (Kirkbride et al., 2013). They estimated exposure ages on five boulders using cosmogenic radionuclide dating and show that the boulders were emplaced by glacier ice during the late Holocene.

Conclusion

In conjunction with the companion paper (Kirkbride et al., 2013), we argue that the glacier–climate model simulations indicate that small glaciers existed in the Cairngorms during the late Holocene and most likely as recently as the LIA. We have shown that glaciers in the Cairngorms were sensitive to late-Holocene cooling, and this introduces the intriguing possibility that similar glaciers existed elsewhere in the British Isles during the LIA and perhaps also at earlier cold periods in the Holocene (e.g. during the 8.2 kyr event). Small glaciers may also have existed during the Holocene in sites where avalanches delivered snow and ice to glacier surfaces at relatively low altitudes (e.g. below the NE cliffs of Ben Nevis and on Aonach Mòr in western Scotland).

Footnotes

Acknowledgements

ASTER GDEM is a product of METI and NASA and was downloaded from . We thank David Smith, Ed Anderson, Tom Bradwell, Kate Smith and Tim Mighall for discussing this research. Edward Boyle took the photographs, and the maps were drawn by Antony Smith. We also thank Martin Kirkbride for many detailed conversations about this research and for providing us with information on his dating project in Coire an Lochain. We are very grateful for the detailed comments made by Simon Carr and an anonymous referee on an earlier version of this paper.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Anderson

Harrison

Passmore

. (2000) Holocene alluvial fan development in the Macgillycuddy’s Reeks, southwest Ireland. Geological Society of America Bulletin 112: 1834–1849.

Ballantyne

Benn

(1994) Glaciological constraints on protalus rampart development. Permafrost and Periglacial Processes 5: 145–153.

Bennett

Boulton

(1993) Deglaciation of the Younger Dryas or Loch Lomond Stadial ice-field in the Northern Highlands, Scotland. Journal of Quaternary Science 8: 133–145.

Bonacina

LCW

(1938) Drift problems suggested by severe snowstorms in the British Isles with special reference to the permanent Scottish snowbeds. Bulletin of the International Association of Hydrologists 23: 91–110.

Cameron

Watson

Duncan

(2012) Two Scottish snow patches survive until winter 2011/2012. Weather 67(6): 162–164.

Coleman

Carr

Parker

(2009) Modelling topoclimatic controls on palaeoglaciers: Implications for inferring palaeoclimate from geomorphic evidence. Quaternary Science Reviews 28: 249–259.

Dansey

(1905) The glacial snow of Ben Nevis. Symons’s Meteorological Magazine, vol. 40, pp. 29–32.

Dunn

Colohan

RJE

(1999) Developing the snow component of a distributed hydrological model: A step-wise approach based on multi-objective analysis. Journal of Hydrology 223: 1–16.

Golledge

Hubbard

(2005) Evaluating Younger Dryas glacier reconstructions in part of the western Scottish Highlands: A combined empirical and theoretical approach. Boreas 34: 274–286.

10.

Hudson

(1977) Cairngorm snowfield report 1976. Journal of Meteorology 2: 163–166.

11.

Kirkbride

Everest

Benn

. (2013) Late Holocene and Younger Dryas glaciers in the northern Cairngorm Mountains, Scotland. The Holocene. DOI: 10.1177/0959683613516171.

12.

Lamb

(1977) Climate: Present, Past and Future: Climate History and the Future, vol. 2. London: Methuen, 837 pp.

13.

Macklin

Johnstone

Lewin

(2005) Pervasive and long-term forcing of Holocene river instability and flooding by centennial-scale climate change. The Holocene 15(7): 937–943.

14.

Manley

(1949) The snowline in Britain. Geografiska Annaler 31: 179–193.

15.

Manley

(1971a) Climate and the British Scene. London: Collins.

16.

Manley

(1971b) Scotland’s semipermanent snows. Weather 26: 458–471.

17.

Mann

Zhang

Rutherford

. (2009) Global signatures and dynamical origins of the Little Ice Age and Medieval Climate Anomaly. Science 326(5957): 1256–1260.

18.

Matthews

Briffa

(2005) The ‘Little Ice Age’: Re-evaluation of an evolving concept. Geografiska Annaler, Series A, Physical Geography 87(A): 17–36.

19.

Plummer

Phillips

(2003) A 2-D numerical model of snow/ice energy balance and ice flow for paleoclimatic interpretation of glacial geomorphic features. Quaternary Science Reviews 22: 1389–1406.

20.

Purves

Barton

Mackaness

. (1998) The development of a rule-based spatial model of wind transport and deposition of snow. Annals of Glaciology 26: 197–202.

21.

Purves

Barton

Mackaness

. (1999) An approach to modelling the impact of snow drift on glaciation in the Cairngorm Mountains, Scotland. Journal of Quaternary Science 14: 313–321.

22.

Rapson

(1985) Minimum age of corrie moraine ridges in the Cairngorm Mountains, Scotland. Boreas 14: 155–159.

23.

Rapson

(1990) The age of the Cairngorm corrie moraines. Scottish Mountaineering Club Journal 34(181): 457–463.

24.

Sissons

(1979) The Loch Lomond Advance in the Cairngorm Mountains. Scottish Geographical Magazine 95(2): 66–82.

25.

Sissons

(1980) Palaeoclimatic inferences from Loch Lomond Advance glaciers. In: Lowe

Gray

Robinson

(eds) Studies in the Lateglacial of North-West Europe. Oxford: Pergamon Press, pp. 31–43.

26.

Sissons

Lowe

Thompson

KSR

. (1973) Loch Lomond Readvance in the Grampian Highlands of Scotland. Nature 244: 75–77.

27.

Sugden

(1977) Did glaciers form in the Cairngorms in the 17–19th centuries? Cairngorm Club Journal 97: 189–201.

28.

Watson

(2011) A Snow Book, Northern Scotland. Northants: Paragon Publishing.

29.

Watson

Pottie

Duncan

(2004) No Scottish snow patches survive through the summer of 2003. Weather 59(5): 125–126.

30.

Watson

Duncan

Cameron

. (2009) Twelve Scottish snow patches survive until winter 2008/2009. Weather 64: 184–186.