The Southern Boundary of the Northern Hemisphere Periglacial Zone at the Faroe Islands

The southern boundary of the Northern Hemisphere periglacial zone at the Faroe Islands

H.H. Christiansen Institute of Geography, University of Oslo, Blindern, Norway O. Humlum The University Courses on Svalbard, Norway

ABSTRACT: The Faroe Islands are used as the observational basis for determining meteorological and periglacial conditions at the southern boundary of the Northern Hemisphere periglacial zone in the North Atlantic. MAAT for the period spanning the Little Ice Age to present are reconstructed for altitudes 0–850 m a.s.l., using the long term (AD 1867–2002) Tórshavn meteorological station record and recent (1995–2002) records from four stations located from sea level up to 850 m a.s.l. Late 20th century temperatures and the altitudinal FDD, TDD and GDD distributions do not suggest modern permafrost at the Faroe Islands, but the highest mountains presumably are close to permafrost conditions. Since about 1930, a cooling trend has prevailed, cul- minating around 1980, followed by a slight warming trend. Widespread shallow sorting producing small-scale sorted circles and stripes occur in the highlands were shallow seasonal freezing reach 10–20 cm. The mountains have a continuous winter snow cover from December to April.

1 INTRODUCTION of the global thermohaline circulation, and is con- sidered of global importance (Broecker, 1991). In com- Northern Hemisphere periglacial terrain near the paratively warm North Atlantic periods, when generally southern (warm) limit is at risk of losing its periglacial strong, or northward-displaced, circulation occurs in character over the next 50–100 years, as indicated by the atmosphere and ocean, the Faroe Islands lie con- projections of climate change from Global Circulation tinually in the main arm of the North Atlantic Drift Models (GCMs) (Houghton et al. 2001). The Faroe (the Gulf Stream). In colder periods, when the North Islands (62°N) in the North Atlantic lie within this Atlantic Drift weakens or its main branch takes a more potentially zone of change. Motivated by the impor- southerly position, a tongue of polar water from the tance of estimating the impact of projected climate East Iceland branch of the East Greenland Current change on periglacial environments, we have recon- approaches the Faroe Islands from the north. As a structed Faroese air temperatures from the years consequence, the Faroe Islands are well placed to regis- 1867–2002, over elevations of 0–850 m a.s.l., to ter periglacial imprints of any large amplitude shifts examine climate change in the Faroes within the in North Atlantic oceanic variations, both past and observational period. We also present short-term present. (1995–2002) measurements of freezing degree days The Faroe Islands are characterised by alpine (FDD), thawing degree days (TDD), growing degree topography due to Quaternary glaciations. There is no days (GDD), snow cover, and ground temperatures tree vegetation, except where planted in few sheltered to characterize Faroese periglacial conditions and to locations below 100 m a.s.l. At sea level the warmest provide baseline measurements for comparison with month is near 10°C and sorted ground phenomena are projected future climate change. widespread above 200–300 m a.s.l. (Humlum and Christiansen, 1998a). Most of the Faroese landscape is therefore within the periglacial zone and is exposed 2 STUDY AREA to Arctic conditions. By representing a series of modern periglacial envi- Warm and saline Atlantic surface water presently flows ronments ranging from marginal close to sea level to around the Faroe Islands into the Norwegian and full periglacial in the highlands, the modern Faroese Greenland Seas, where evaporation and cooling during landscape may provide useful information regarding winter produces a gradually higher water density. This the climatic constraints on the North Atlantic periglacial dense water then overturns, resulting in deep convec- environment. tion (Bigg, 1996). The sinking cold water represents a Meteorological observations were initiated in major constituent of North Atlantic Deep Water, part Tórshavn in AD 1867 (Brandt, 1994). The data series

139 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 are further used to reconstruct MAAT for different TORSHAVN,FAROE ISLANDS altitudes back to the final parts of the LIA. 2000 annual values and 11-yr running mean 2000

1600 1600

1200 Precipitation (mm w.e) 1200 3 AIR TEMPERATURES

800 JJA temp (deg.C) 800 11 11 In the highest mountain massif, Slættaratindur (50 km 10 10 NNE of Tórshavn), air temperatures at different alti- 9 9

8 8 tudes have been recorded since 1995. Temperatures MAAT (deg.C) 7 7 were measured using miniature single channel 6 6 Tinytalk (1995–1997) and TinyTag (1997–2002) data 5 5 loggers with external sensors located 10 cm above the 4 4

3 3 terrain surface in small stone cairns, protecting the

2 DJF temp (deg.C) 2 sensors from sheep, people and direct solar radiation 1 1 and exposing them to efficient ventilation. Presumably 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 the recorded temperatures are slightly higher than if Figure 1. Variations of air temperatures (°C) and precipi- recorded at standard height, 2 m above ground. Data tation (mm w.e.) at Tórshavn since 1867. Solid lines show recording intervals were 5 hours May 1995–1996, running 11 yr mean values. 2 hours May 1996–1999 and 1 hour since May 1999. The accuracy of the sensors is 0.1°C. To describe the altitudinal variation data from 303 m, 634 m and 850 m a.s.l. are used. All these three document the final part of the Little Ice Age (LIA) stations were located in wind-exposed parts of the and its termination shortly after 1920, as indicated by landscape, in passes or on mountain summits, without a temperature increase (Fig. 1). The transition was any significant snow coverage. Data were down- mainly signalled by higher winter temperatures and loaded annually. They showed no signs of snow cover only to a lesser degree by other seasonal temperatures. preventing quick temperature variations. The data After 1940 the mean annual air temperature (MAAT) series are almost continuous, as only very short periods gradually decreased to typical LIA values until of maximum 34 hours of failure occurred. However, around 1980, again mainly caused by changes in win- for the 850 m station, the sensor was damaged after a ter temperatures. Since then, a slight warming has few months of operation in 1995 and was not replaced occurred. From a temperature point of view, the LIA before May 1996. thus more or less still continues on the Faroe Islands, The mean monthly air temperature (MMAT) (Fig. 2 and was only shortly interrupted by a relatively warm and table 1) shows the same general annual variation period 1925–1940. There is no clear evidence for an at the different altitudes, but with the largest ampli- association between MAAT and annual precipitation tude (13°C) registered for the higher parts of the land- in the Faroe Islands during the observational period scape, while the amplitude is somewhat smaller (r-squared: 0.12), and precipitation is presumably also (10°C) close to sea level. controlled by local factors such as wind direction and The coldest month varies from November to March. orographic effects. In contrast, the warmest month is always either July or Until recently all Faroese meteorological stations August, but it varies at each station for each indi- were located close to sea level with only a few stations vidual year. The MMAT difference between stations is operating at higher altitudes (up to 282 m a.s.l.). The largest during winter when the lapse rate generally is establishment of a mountain meteorological station in at maximum (0.7 to 0.9°C/100 m), while in sum- the year 2000, however, suggested a position of the mer it is generally smaller (0.4 to 0.5°C/100 m) low arctic boundary at only 200 m a.s.l. (Christiansen presumably due to increased insolation and lack of and Mortensen, 2002). This altitude corresponds to a topographic induced shading on the high ground. Dur- mean annual air temperature (MAAT) of 5.0 to 3.5°C, ing summer, the highest ground may penetrate above depending upon exposure to the prevailing wind the cloud cover, and therefore receive more direct (Humlum and Christiansen, 1998). radiation than the valleys below. In the present paper new 1995–2002 data series of The fact that the monthly variations show an identi- air- and ground temperatures, supplemented by auto- cal pattern for all stations, including the official station matic digital snow cover observations are used to in Tórshavn, suggests that the distribution of stations improve the description of the modern Faroese used in the present analysis ensures well-ventilated periglacial environment and to analyse the magnitude temperature sensors due to a combination of an open and significance of interannual variations. The data landscape without trees and frequent high wind speeds.

140 MMAT 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Torshavn 303 m asl 14 634 m asl 14 8 8 13 850 m asl 13 12 12 7 7 11 11 10 10 6 Torshavn 6 9 9 8 8 7 7 5 5 6 6 303 m 5 5 4 4 4 4 3 3 3 3 degrees C 2 2 634 m 1 1 2 2 0 0 -1 -1 850 m -2 -2 1 1 -3 -3 -4 -4 0 0 -5 -5 Jun-1995 Jun-1996 Jun-1997 Jun-1998 Jun-1999 Jun-2000 Jun-2001 Jun-2002 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Figure 2. Mean monthly air temperature (MMAT; °C) Figure 3. Mean annual air temperature (MAAT) °C AD from the meteorological station in Tórshavn (50 m a.s.l.) 1867–2002 at various altitudes, reconstructed from Tórshavn and from the miniature data loggers at different altitudes in measurements. the Slættaratindur area from June 1995 to June 2002. 3000 2500 2000 1500 1000 500 0

Table 1. Summary data June 1995 to June 2002 of 800 800 MMAT (°C). Data from Tórshavn and the measurement 700 700 sites in the Slættaratindur massif. 600 600 MMAT Tórshavn 500 500

m asl 400 400 m asl

1995–2002 50 m a.s.l. 303 m a.s.l 634 m a.s.l. 850 m a.s.l. 300 300 FDD Obs. No. 84 85 85 72 200 200 100 100 Minimum 1.9 1.4 3.3 4.5 TDD GDD Maximum 12 13.2 10.2 8.2 0 0 Mean 6.7 4.7 2.6 1.3 -3000 -2500 -2000 -1500 -1000 -500 0 Median 6.5 4.5 2.0 0.5 Figure 4. Cumulated annual freezing degree days (FDD; Variance 8.1 12.5 13.9 12.3 solid lines, lower x-axis), thawing degree days (TDD; solid St. dev. 2.8 3.5 3.7 3.5 lines, upper x-axis) and growing degree days (GDD; stippled lines, upper x-axis), calculated from daily mean values June 1995 to June 2002 for 33 m, 303 m, 634 m and 3.1 Mountain air temperature reconstruction 850 m a.s.l. in the Slættaratindur area. Each line represents back to AD 1867 one measurement year from 1 June to 31 May the following year. Thick stippled lines are FFD and TDD for the warmest Using the correlation between MMAT values for the year (1872) and the coldest year (1979) within the 1995–2002 Slættaratindur series area and the official 1867–2002 observational period, calculated from meteoro- meteorological station in Tórshavn, a MAAT recon- logical measurements in Tórshavn. struction back to 1867 for different altitudes was car- ried out for the Slættaratindur area (Fig. 3). The 3.2 Freezing, thawing and growing degree days correlation between MMAT at altitudes of 303, 634 and 850 m a.s.l. and the Tórshavn data was 0.92, 0.95, Freezing degree days (FDD), thawing degree days 0.94 (r-squared values), respectively, for the period (TDD) and growing degree days (GDD) (Molau and 1995–2002. As mentioned above, this rather high Mølgård, 1996) have been calculated for the different correlation is presumably derived from efficient wind altitudes in the Slættaratindur area, based on daily ventilation at all altitudes. mean air temperature values 1995–2002 (Fig. 4). For the reconstructed temperature series since AD There are less FDD than TDD recorded at all alti- 1867 the warmest period falls in the early part of the tudes, indicating permafrost to be absent. The number measurement period, actually within the LIA, whereas of FDD exceeds the number of GDD for altitudes the coldest period is registered after the official end of above 600–700 m a.s.l. Significant plant growth, gener- the LIA, from 1950 to 1985. Figure 3 shows that MAAT ating a closed plant cover, ceases above 300 m a.s.l., in the highest mountains presumably approached 0°C suggesting a GDD value of about 500 to represent a during certain cold periods in the 20th century, such critical value for modern vegetation on the Faroe as 1917–1920 and 1979–1981. Islands. TDD has a much larger interannual variation,

141 particularly at lower altitudes, than both FDD and 3.7°C and the mean annual ground temperature GDD. FDD has the largest interannual variation at (MAGT) was 3.8°C at both 5 and 10 cm depth. At 1 m high altitudes and show only little interannual varia- depth, the MAGT was almost identical, 3.7°C. The tion close to sea level, reflecting the importance of maximum temperature range at the terrain surface nearby sea surface temperatures. was 39.1°C, decreasing to 22.0°C at 5 cm depth, 18.9°C Figure 4 also shows reconstructed FDD and TDD at 10 cm, 13.6°C at 20 cm, while at 1 m depth the annual for the warmest year on the Faroese meteorological temperature range was only 7.2°C. record (1872) and the coldest year (1979), respectively, thereby defining limits for the annual variation during the observational period (since AD 1867). The 5 PATTERNED GROUND period 1995–2002 clearly plots within the colder part of the empirical FDD envelope, but are still close to As the annual ground freezing is shallow, reaching average for TDD. Interestingly, the maximum FDD depths of 10–40 cm (Fig. 5), this presumably explains and minimum TDD do overlap above 820 m a.s.l., indi- why the vertical sorting of the widespread small scale cating that in the highest parts of the Faroese land- sorted circles (Fig. 6) generally extends only to depths scape there might well be sites with a negative of 5–10 cm (Humlum & Christiansen, 1998b). TDD-FDD balance during cold years. The surface geomorphic activity is, however, considerable. Recurrent annual photography 1995–2002 of sorted circles and –stripes, at 634 m a.s.l. in the 4 GROUND TEMPERATURES Slættaratindur massif, demonstrates that particles up to 3–4 cm are involved in the annual surface movement. Ground temperatures have been measured at 3 sites in Surface displacements of up to 1 cm/yr have been the Slættaratindur area. These data series are not con- observed on horizontal ground, while the annual move- tinuous because of thermistor problems. However, a ment may be considerably larger on sloping ground. continuous data series on soil temperatures has been Large particles (5 cm) are usually stable in the pre- measured at the meteorological station on the Sornfelli vailing periglacial environment, as is suggested by mountain summit plateau (Christiansen and Mortensen, their lichen cover. 2002), 26 km S of the Slættaratindur area. Here an 11 m borehole was drilled at an altitude of 722 m a.s.l., in the summer of 1999, instrumented for ground tempera- 6 PRECIPITATION AND SNOW COVER ture monitoring in November 1999 and has operated satisfactorily since June 2000 (Fig. 5). Precipitation on the Faroe Islands is considerable, rang- Figure 5 shows seasonal freezing to start in late ing from less than 800 mm (w.e.) on peripheral islands December, continuing to late April, but only from mid to more than 3000 mm in certain central highlands February to late April does the 0°C isotherm reach (Cappelen and Laursen, 1998). Much of the precipita- depths of 30–40 cm. Due to high soil water content, tion is solid, and snow may fall in the mountains in only the upper 10–15 cm of the soil cools below 2°C any month of the year. In Tórshavn, close to sea level, during winter. The seasonal ground freezing is thus snow covers the ground for 44 days annually (Cappelen characterised by near 0°C conditions prevailing during much of the winter, suggesting that all available soil water never freezes. The mean annual ground surface temperature (MAGST) (June 2000–June 2002) was

2000/2001 2001/2002 0 cm 2 2 2 25 cm 2

50 cm 2 2 2 75 cm 2

100 cm 2000/2001 2001/2002 Figure 5. Ground temperature profile to 1 m depth, measured at the Sornfelli summit plateau at 722 m a.s.l., from 1 June 2000 to 19 September 2002. Temperatures (°C) are recorded hourly at 0, 5, 10, 20, 40, 80 100 cm depth. Shading indicates temperatures below 0°C. Contour Figure 6. Small-scale sorted stripes shortly W of Tórshavn, equidistance 2°C. at 260 m a.s.l. The sorting extends to 5–10 cm depth.

142 and Laursen, 1998). To obtain information on the 30 April and 6 May (Fig. 7), as well as during and snow cover duration and timing in the highlands, auto- following a heavy rainstorm on 18 September 2000. matic digital photography (Christiansen, 2001) has been applied from June 1999 to July 2002. A daily photograph was taken showing an east- 7 DISCUSSION facing cirque, Givrabotnur, below the summit of the highest mountain in the Faroe Islands, Slættaratindur At the Faroe Islands low temperatures and high pre- (882 m a.s.l.). The automatic camera was located cipitation presumably enhances ongoing periglacial approximately 700 m from the headwall of the cirque. activity such as weathering and sorting processes. From the photographs the daily snow cover was From this point of view, the Faroe Islands have expe- mapped. rienced improved climatic conditions for increased Based on analysis of the daily photos obtained from periglacial activity towards the end of the 20th century. June 1999 to July 2002, the seasonal, continuous Today, significant sorting phenomena occur down to snow cover was established by mid November in about 150–200 m a.s.l. MAAT is only slightly above 1999, by mid December in 2000 and by late January 0°C at the highest mountains on the Faroe Islands and in 2002, respectively. A continuous, although thin, presumably the present climate is relatively close to snow cover also occurred outside the main winter sea- providing background for glaciation but particularly for son, e.g. from late September to mid November 1999, permafrost establishment in the highest mountains. from late October 2000 to mid December 2000, and in In contrast to this, the relatively warm period June 2001. The thickest snow cover existed from late 1925–1940 was characterised by less precipitation March to early April in 1999, 2000 and 2001, but and was therefore, presumably, a period of relatively occurred in February– March in 2002. Melting of the low periglacial activity (only activities such as needle continuous snow cover started late April and occurred ice action, sorting and soil erosion by wind). The mainly in May, but snowpatches existed long into the lower limit for periglacial activity at that time was summer (Fig. 7). The last snow disappeared 29 July in probably located 100–150 m above the present position 1999, 30 August 2000, between 17 and 20 July in (Fig. 8). 2001, and between 30 June and 9 July in 2002. Due to very frequent wind activity the most exposed parts of Mean annual the landscape, like the Slættaratindur mountain top air temperature (ºC) and ridges are very seldom snow covered. -15 mean equilibrium line on glaciers Geomorphic activity such as snow avalanches and NE Greenland (74N) rock falls was recorded during late winter all observa- ELA tion years. In 2000 snow avalanche activity was -10 recorded on 29 April and 1 May (Fig. 7). In 2002 rock continuous cold firm fall activity and debris slides were recorded between -7 permafrost -6 -5 Discontinuous temperare

permafrost firm Zone of potential glaciation W Greenland (66N) ELA SE Greenland (66N) -2 Sporadic permafrost Lower limit for permafrost 0 Faeroe Islands lale 20th century Periglacial zone calm 3.5 Periglacial boundary zone 5 windy Reykjavik (64N) Bergen (60N)

Aberdeen (57N) Torshavn (62N) Faeroe Islands 1925-1940

500 10001500 2000 Mean annual precipitation (mm w.e.)

Figure 8. Approximate climatic constraints for glaciation, Figure 7. Automatic digital photographs of Givrabotnur periglaciation and permafrost (Humlum and Christiansen, cirque (600 m a.s.l.) and summit of Slættaratindur 1998a). The shaded envelope indicates the 1950–1985 (882 m a.s.l.) showing avalanche activity 1 May 2000, rock climatic position of the Faroese landscape, while the dotted fall 8 May 2002, and large interannual variation between 29 envelope indicate the position during the warm period August 2000 and 30 June 2002 in the date of final existence 1925–1940. The highest Faroese mountains are apparently of snowpatches. Arrows indicate position of activity close to fulfilling climatic requirements for glaciation but referred to. particularly for permafrost establishment.

143 At the onset of the 21st century there is no indication ACKNOWLEDGEMENTS of any exceptional climatic warming in the Faroe Islands, and the small MAAT increase since 1980 is Funding 1999–2002 was provided by the Danish presumably a normal recovery following the relatively Science Research Councils programme on the North cold period 1950–1980. However, should a significant Atlantic to the ‘Linking Land and Sea at the Faroe warming commence sometime in the future, as sug- Islands: Mapping and Understanding North Atlantic gested by GCMs (Houghton et al. 2001), the unique Changes’ LINK research project. Likewise the Danish meteorological record from Tórshavn together with Natural Science Research Council provided financial continued climatic monitoring in the Faroese moun- support for the 1995–1998 project ‘Climatic controls tains could provide an important means for estimating on the Faroese landscape’. The Arctic Environmental environmental effects on the Faroese landscape, using a Programme of the Danish Environmental Ministry is graphic approach as indicated in Fig. 8. thanked for funding the ongoing data collections at the Sornfelli meteorological station. Lis Mortensen, Faroese Geological Survey, has collected the data from 8 CONCLUSIONS the borehole at Sornfelli. Ulf P. Thomas, Institute of Geography, University of Copenhagen, instrumented The Faroe Islands represent an extreme maritime mod- the borehole on Sornfelli and produced the automatic ern southern limit of the Northern Hemisphere digital camera. periglacial zone. According to the natural tree line, the modern periglacial environment extends almost to sea REFERENCES level. The lower occurrence of active periglacial features such as patterned ground and sorted stripes, however, Bigg, G.R. 1996. The Oceans and Climate. Cambridge suggest the periglacial boundary to be located within a University Press, 266 pp. range from 150 to 300 m a.s.l. Temperatures as well as Brandt, M.L. 1994. The North Atlantic Climatological exposure to wind and insolation control this altitudinal Dataset. Danish Meteorological Institute Technical range. Above the modern periglacial boundary the plant Report 94–18, 67 pp. cover rapidly becomes patchy and periglacial sorting Broecker, W.S. 1991. The great ocean conveyor. phenomena widespread. Oceanography, 4, 79–89. The seasonal snow cover duration is about 44 days Cappelen, J. and Laursen, E.V. 1998. The Climate of the Faroe Islands – with Climatological Standard Normals, near sea level. The snow distribution is mainly 1961–1990. controlled by the large wind activity. In the highest Christiansen, H.H. 2001. Snow-cover depth, distribution mountains the snow cover may locally last for more and duration data from northeast Greenland obtained than 300 days in lee sites where snowpatches accumu- by continuous automatic digital photography. Annals late, while the exposed parts of the landscape, such as of Glaciology, 32, 102–108. mountaintops and ridges, do not have any significant Christiansen, H.H. and Mortensen, L.E. (2002). Arctic snow cover. The maximum snow cover thickness is Mountain Meteorology at the Sornfelli Mountain in usually attained in the period late March to early April. Year 2000 in the Faroe Islands. Fróðskaparrit, 50, The modern limit of periglacial activity on the 93–110. Faroe Islands corresponds to a MAAT of about French, H.M. 1996. The Periglacial Environment. 2nd edi- tion. Longman, Essex, 341 pp. 3.5–5°C. This is somewhat warmer than what is usu- Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der ally suggested (French, 1996), and the difference most Linden, P.J., Dai, X., Maskell, K. and Johnson, C.A. likely reflects the windy character of the Faroese cli- (eds) 2001. Climate Change 2001: The Scientific mate, leading to overall difficult growing conditions Basis. Cambridge University Press, 879 pp. for plants. Humlum, O. and Christiansen, H.H. 1998a. Mountain The land areas above the periglacial boundary Climate and Periglacial Phenomena in the Faeroe (presently about 50% of the total land area) represent Islands. Permafrost and Periglacial Processes 9: a typical arctic environment from a geomorphologi- 189–211. cal point of view. MAAT presently is only 1–2°C Humlum, O. and Christiansen, H.H. 1998b. Late Holocene Climatic Forcing of Geomorphic Activity in the Faroe above 0°C, and the TDD/FDD balance is around Islands, North Atlantic Ocean. Fróðskaparrit, 46, 1.3–3.4 at the highest mountains. Presumably the 169–189. highest mountains are close to fulfilling climatic con- LINK 1999–2002. http://www.geogr.ku.dk/projects/link. ditions for glaciation but particularly for permafrost Molau, U. and Mølgaard, P. 1996. ITEX Manual, Inter- establishment. national Tundra Experiment. Danish Polar Centre, 53 pp.

144