Quaternary International, Vol. 27, pp. 53-58, 1995. Copyright © 1995 INQUA/Elsevier Science Ltd Printed in Great Britain. All rights reserved. 1040-6182/95 $29.00
Pergamon 1040-6182(94)00060-3
A TIME D E P E N D E N T G L A C I O L O G I C A L M O D E L OF THE WEICHSELIAN ICE SHEET P. Holmlund* and J. F a s t o o k t
*Department of Physical Geography, University of Stockholm, S-10691 Stockholm, Sweden tDepartment of Computer Science, University of Maine, Orono, ME 04469, U.S.A. A time dependent model is used to study the evolution of the Weichselian Ice Sheet. The results indicate three stages of the glaciation. A mountain-centered ice sheet between 110,000 and 80,000 BP followed by a phase with an extension similar to or slightly larger than the Younger Dryas zone. The last and maximum extent was reached about 30,000-20,000 BP and the main deglaciation began about 15,000 BP. The model is based on a finite-element method solution of the continuity equation. The climatic records from the Camp Century and Summit ice cores are used for climate control and the model is calibrated using geological deglaciation evidence in Scandinavia. The assumption of a general sliding condition for elevations below 100 m, with an enhanced sliding zone through the center of the Baltic Sea and Gulf of Bothnia yield broad agreement for marginal positions in Sweden and Finland during a simulated termination.
INTRODUCTION
the deglaciation indicate movement in several directions. The glacial morphology has been described by Hoppe (1959, 1963, 1967) and Fromm (1965). Later studies by Lagerback indicate several deglaciation patterns superimposed on each other (Lagerb~ick, 1988). A last cold-based advance may have preserved eskers and other landforms from destruction. Kleman came to similar conclusions concerning the deglaciation pattern in the Swedish mountains. Using relative datings he distinguished three or four phases of deglaciation. Thus, there is a need of a new time dependent model for the Weichselian glaciation in order to accommodate the complicated pattern of new data and to see whether these data fit into the old, accepted view of the last glaciation, or if we need to revise our model. The model and the physics used in this model are described by Fastook and Holmlund (1994) and Fastook and Prentice (1994). The calibration and handling of geological data is described by Holmlund and Fastook (1993). A more extensive description of geological implications of the glaciation is published by Holmlund (1993). This paper summarizes the results of the entire project.
After the ice age concept was established in the mid-19th century, a model of the Scandinavian Ice Sheet was gradually formed. At the end of this century the ice age was subdivided into three or four phases (Geikie, 1894), named after the rivers near the ice margin. The last large extent was named Weichsel. Until 1950-1960 only relative datings were possible to obtain but with radiocarbon dating and other new methods an accurate chronology was introduced. In 1949 Ljungner made a careful description of the development of the Weichselian Ice Sheet, comparing it to the existing mountain glaciation in Patagonia, where he found similarities in climate and topography with northern Scandinavia (Ljungner, 1949). According to Ljungner as the Weichselian ice thickened the ice divide migrated eastwards, finally reaching a position extending from southern Norway to the Gulf of Bothnia. Almost 80 years before Ljungner .published his model Torell (1873) introduced the concept of the 'Baltic Ice Stream', draining the eastern part of the Scandinavian Ice Sheet. The modified concept of a Baltic Ice Stream was recently discussed by Holmlund and Fastook (1993). Boulton and Jones (1979) and Boulton et al. (1985) postulated two models for the Weichselian Ice Sheet. The first addressed the deglaciation forms in the landscape, requiring an ice divide situated further to the east than Ljungner (1949) suggested. A second model addressed the maximum extent of the Weichselian Ice Sheet, it allowed for a deformable bed, sliding, and the presence of a Baltic Ice Stream, and hence produced a flatter surface profile than other models. Based on studies of striaes, Kleman (1990) suggested a two-domed ice divide with the main dome situated in the Gulf of Bothnia and a minor dome in southern Norway, linked by a saddle. Kleman's view agrees well with Ljungner's model (Ljungner 1949). The geological traces of the last glaciation in northern Sweden are quite complicated. Landforms originating from
THE MODEL The model consists of a solution of the two-dimensional (map-plane) time dependent mass-continuity equation. The finite-element method assumes linear spatial interpolants for the solution of differential equations, turning the problem into a set of algebraic equations that can be solved by ordinary matrix methods (Becker et al., 1981). The model is described in detail elsewhere (Fastook, 1990; Fastook and Holmlund, 1994; Fastook and Prentice, 1994; Holmlund 1993; Holmlund and Fastook, 1993). The primary input, besides the bedrock topography, is the mass balance at each of the 1492 nodes in a grid covering Scandinavia. We developed a mass-balance relationship based on the empirical fit to present Antarctic accumulation rates. The relationship depends on surface elevation, surface slope and latitude. Complementary ablation rates are based 53
54
p. Holmlund and J. Fastook
on south Greenland mass-balance data, and are appropriate for a modest warming of the Antarctic climate. The climate is adjusted by varying the mean annual sea-level air temperature, which provides a starting point lot all temperature calculations at present sea level. We are aware of the limitations of this very simplified model of the mass balance, but feel that it is an appropriate approximation of the actual situation. The above described mass-balance scheme implies a net accumulation pattern dominated by elevation, surface slope and latitude. An enhancement factor, proportional to the magnitude of the prevailing wind, is defined, and the symmetric mass-balance distribution is modified by this factor times the directional surface gradient along the wind direction. This implies enhanced precipitation on the windward sides of topographic barriers, and reduced precipitation within their downwind precipitation shadows. This allows for a more accurate representation of the massbalance distribution in maritime or orographicallycontrolled climate regimes. Assumptions on the basal conditions are based on geologic evidences and hypsographic information from the Baltic Sea and the Gulf of Bothnia. The general assumption is that the bed was frozen on all bed elevations above today's 100 m level. Between this level and -100 m the glacier is assumed to slide in accordance with the relation suggested by Weertman (1957, 1964). Where the bed is below -100 m we used a softer sliding constant which was half the value used in the -100 to 100 m zone (Fig. 1). This enhanced rate was added to simulate deformable sediments. One exception from this general scheme is that the ice was assumed to be frozen to its bed on the island Alan& This was done in order to provide an obstacle for the flow from the Gulf of Bothnia out into the Baltic Sea. In order to define the best relation between frozen and thawed bed under the ice sheet we made some test runs on extreme conditions. These test runs began with a configuration close to the maximum extent of the Weichselian Ice Sheet. In all of these experiments the climatic signal is a sudden change of the equilibrium line from 300 m to 1500 m. In the first experiment the whole ice sheet suddenly becomes wet-based with no frozen spots, the ice sheet collapses and is gone after only 2400 years. The last remnants of the ice would be found in the Gulf of Bothnia. In the second experiment the ice is frozen everywhere to its bed, and after about 1000 years the glacier attains a new state of equilibrium. After 4000 years its areal extent is smaller than at maximum, but it is almost 100 m thicker in the center, despite the warmer climate. This ice sheet would probably not disappear even during the warmest parts of the Holocene. It is from these considerations that we obtained the simple scheme of allowing sliding below 100 m bedrock elevation. Places where the bed was lowest would generally coincide with areas where the ice was thickest, and hence most likely to be at the pressure melting point. For higher elevations we expect that the thinner ice overburden would allow more of the heat to escape so that the bed is frozen and no sliding can occur.
The Younger Dryas stillstand was difficult to simulate. A
Basal conditions during the deglaclation
FIG. 1. Basal conditions used in the model. The assumptions concerning the basal conditions are based on g e o l o g i c a l e v i d e n c e and h y p s o g r a p h i c information from the Baltic Sea and the G u l f of Bothnia. The general assumption is that the bed was frozen on all bed elevations above today's 100 m level. Between this level and - 1 0 0 m sliding was present. Where the bed was below - 1 0 0 m we used a softer sliding constant, which was half the value used in the - 1 0 0 to 100 m zone. This enhanced rate was added to simulate deformable sediments.
cold period will most probably cause an advance of the ice front after a significant time-lag. In order to keep the climate event and the dynamic response within the 1000 years chronozone suggested for Younger Dryas (Mangerud et al., 1974), we must assume that the cold climate signal barely exceeded 500 years (Holmlund and Fastook, 1993). This is a consequence of the long response time of an ice sheet with low basal stresses as well as low longitudinal strain rates (Nye, 1960). In order to run the model we needed a reliable climatic curve for the last 120,000 years. We chose to use Greenland ice cores as they may represent the Scandinavian climate fairly well. Though the Dye 3 ice core was taken from southeastern Greenland it was not used for this experiment. Letrrguilly et al. (1991 ) and Reeh (1991) have demonstrated that this location probably was deglaciated during the Eemian and thus has undergone large dynamic changes during the Weichselian glaciation. Instead we used the Summit data (Johnsen et al., 1992) from 38,000 BP to the present and Camp Century data for the period prior to 38,000 BP (Reeh, 1991). The oxygen isotope relation was translated into temperature using the relation 1% ~(~80) = 0.67°C (Johnsen et al., 1992). However, for some reason the temperatures interpreted from the Greenland ice cores show too little warming during the phase of deglaciation. In this
A Time Dependent Glaciological Model of the Weichselian Ice Sheet 5
5
O-
-0
-5
-5
lO
lO
F,-'15
130000
10400/8" 78000 52000 Time BP
26000
0
Oi
2
The temperatures used are shown in Figs 2 and 3. Figure 3 also shows the areal extension of the ice sheet during the three stages of the Weichselian glaciation. Small changes in the calibration of the temperature curve will affect the curve of areal distribution of the ice sheet.
,0
M O D E L L I N G RESULTS
i;
-5
-5-
-10
30000
model we had to raise the temperature by 3°C in order to retrieve an acceptable scenario and chronology for the deglaciation (Fig. 2). During glacial time the Greenland Ice Sheet expands and grows thicker. This change in ice surface elevation requires a correction to the climatic signal. This was accommodated by using the calculation on thickness changes of the ice sheet during the last 100,000 years, carried out by Latr6guilly et al. (1991). We translated the changes in elevation to temperature by using an adiabatic lapse rate of 1 °C/100 m.
!
~lJ
55
-10
24000
18000 12000 6000 0 Time BP FIG. 2. Air temperature related to the present during the Weichselian as calculated from the Camp Century and the Summit ice cores. The upper figure shows the entire glaciation with the different marine time stages. The lower figure shows the last phase of the glaciation and the deglaciation. Line (a) shows the direct interpreted tsO-curve and (b) shows the 3oc raised temperature between 14,000 BP and 9100 BP used in this model in order to fit the geological evidence.
The model is forced by climate signals and thus by the chronology of the Greenland ice cores (Reeh, 1991). This fact should be considered while comparing ages of events, for example with 14C-ages. As a result of the data from the Greenland ice cores the model gave the following result. Weichsel I initiates by some small ice caps in southern Norway about 112,000 BP. Around 110,000 BP an ice sheet forms, covering most of the Scandinavian mountain range. Shortly after that the ice sheet divides into one northern and one southern ice cap (Fig. 4a). This configuration lasts for 10,000 years. At 99,000 BP the two ice caps join and a single, elongated dome forms. The ice begins to advance in the northeast. At 97,000 BP it reaches the Kola Peninsula and at 91,000 BP, the Gulf of Bothnia (Fig. 4b). After a maximum at 90,000 BP a considerable recession begins, reaching to a
/1.
3.0 _T
~
A
r
N -4 m
F-
zm
2.0-
~" v
I,,x w
-10
1-~r-r~-,
'13 m :0 )'-4 r. =O Ili o
.1
E < 1.0--
13 0
--20
104
718
5'2
2'6
,I I 0
TIME BP FIG. 3. Areal extension of the ice sheet (A) and air temperature at sea level (T) as calculated from Greenland ice cores, as a function of time.
56
P. Holmlund and J. Fastook
FIG. 4. Six images of the extent of the Weichselian Ice Sheet according to this model.
A Time DependentGlaciologicalModelof the Weichselian Ice Sheet minimum at 82,000 BP (Fig. 4c). The ice sheet is now only covering the mountain area. The next phase of the glaciation begins at 78,000 BP and is characterized by a substantial growth. After 1000 years of growth it reaches the Kola Peninsula and the Gulf of Bothnia. Aland is ice covered at 72,000 BP and all of Finland is ice covered at 65,000 BP. By 61,000 BP the margin reaches a stillstand in Lithuania. Large parts of southern Sweden is ice covered (Fig. 4d). The stillstand is followed by a slow recession which precedes until 53,000 BP especially in southern Sweden. By 47,000-43,000 BP the ice advances in the southern Baltic Sea and the margin moves into Scania (Fig. 4e). After a slight retreat Scania is again covered (40,000-39,000 BP) followed by another retreat. The last stage of the Weichselian begins at 37,000 BP with a major advance. About 30,000 BP the ice front reaches northern Germany where it stands until about 15,000 BP. The maximum extent occurs about 20,000 BP (Fig. 4f). The deglaciation begins around 14,500 BP and all ice is gone at 8000 BP. We assume that the ice at the bottom of the Baltic Sea and the Gulf of Bothnia was at the pressure melting point all through the glaciation. In the Baltic Sea the ice may have been frozen to its bed in the initial stage of the glaciation but became thawed when the ice sheet was thick enough. Accepting these initial conditions leads to the conclusion that the ice cap probably was drained by a 'Baltic Ice Stream' (Boulton et al., 1985; Holmlund and Fastook, 1993; Lundquist, 1987; Torell, 1873). The fact that the ice stream turned west in the southern part of the Baltic is probably due to more favorable basal conditions and higher melt rates towards the west. When the warmer post-glacial climate begins the low profile of the Baltic Ice Stream results in high ablation rates. It melts-off quickly up to A~land island. The islands Gotland and Oland do not seem to have been major obstacles for the ice flow. Alan& on the other hand, must have been an important obstacle. If not, the ice sheet would have been drained too quickly, leaving one major dome in Sweden and a minor one in Finland. In addition, the climatic Younger Dryas event does not have much influence on the deglaciation of the interior of the ice sheet as the interior has already been drained by downdraw effects (Hughes, 1987). As the island/,land becomes ice free the ice stream from the Gulf of Bothnia is reactivated and the draining of the interior of the remaining ice sheet proceeds quickly. Table 1 summarizes the modelling results and Fig. 5 shows the time of ice sheet coverage during the Weichselian glaciation. It shows that the mountain range was covered for
FIG. 5. Timeof ice sheetcoverageduringthe Weichselianaccordingto this model. The numberof years are expressedin thousands of years. 80,000-100,000 years. The area occupied during Weichsel II was covered for 40,000-60,000 years. The maximum phase, Weichselian III, lasted for only about 20,000 years. CONCLUSIONS In the Baltic Sea the ice flow was probably governed by high rates of basal sliding in the deeper parts. The island/;,land must have been a severe obstacle for the drainage of ice from the interior of the ice cap. If not, the interior of the ice sheet drains too rapidly and leaves a small dome in Finland in addition to the main dome in northern Sweden. This is not in accordance with geological evidence. The delay of the dynamic response of the ice sheet to the climate signal during Younger Dryas must have been in the order of hundreds of years. In order to fit the climate signal and the dynamic response into the chronozone Younger Dryas, which lasted for 1000 years, the climatic signal must have lasted only during the first half of this period. Weichsel I was a mountain centered glaciation, while Weichsel II had an extent similar to the Younger Dryas stillstand. The changes in the extent of the inland ice were controlled by the average temperature and short term anomalies. The expansion of Weichsel III from the Weichsel
TABLE 1. Result of the modellingproject 110,000-81,000 BP 75,000 BP 65,000 BP 65,0(0)-34,000 BP 34,000 BP 20,000 BP 15,000 BP 8000 BP
57
mountain centered glaciation expansion over northern Swedenand the Gulf of Bothnia close to YoungerDryas zone minor fluctuations expansion Weichselian maximum deglaciation began all ice was gone
58
P. Holmlund and J. Fastook
II extent was primarily due to frequent short term cold events. One conclusion we found important that could be drawn from this experiment was that all new data on the basal conditions and hypotheses on multiple deglaciation phases in the mountains of Sweden seem to fit into a slightly refined version of the commonly used model of the last glaciation.
ACKNOWLEDGEMENTS We are very grateful for the financial support given by the Swedish authority for spent nuclear fuel (SKN) and the Swedish Nuclear Power Inspectorate (SKI). Special thanks to NSF grant RII-8922105 which provided the computer equipment that made this modelling possible. We would also thank the following persons for providing us with data and/or giving us valuable comments to the manuscript: Professor B. Andersen, Dr S. Bjtrck, Professor G. Denton, Professor G. Hoppe, Professor T. Hughes, Dr J. Kleman and Professor J. Lundquist. Thanks also to Karin Weilow and Hans Drake who drew the illustrations.
REFERENCES Becker, E.B., Carey, C.F. and Oden, J.T. (1981). Finite Elements, An Introduction. Prentice-Hall, Englewood Cliffs. Bjtrck, S. and Digerfeldt, G. (1986). Late Weichselina-Early Holocene shore displacement west of Mt. BiUingen, within the Middle Swedish end-moraine zone. Boreas, 15, 1-18. Bjtrck, S. and Digerfelt, G. (1989). Lake Mullsjfn - - A key site for understanding the final stage of the Baltic lee lake east of Mt. Billingen. Boreas, 18, 209-219. Boulton, G.S. and Jones, A.S. (1979). Stability of temperate ice caps and ice sheets resting on beds of deformable sediments. Journal of Glaciology 24(90), 29-43. Boulton, G.S., Smith, G.D., Jones, A.S. and Newsome, J. (1985). Glacial geology and glaciology of the last mid-latitude ice sheets. Journal of Geological Society, London, 142, 447-474. Fastook, J.L. (1990). A map-plane finite-element program for ice sheet reconstruction: A steady-state calibration with Antarctica and a reconstruction of the Laurentide Ice Sheet for 18,000 BP. In: Hilton, U. Brown (ed.), ComputerAssisted Analysis and Modeling on the IBM 3090. IBM Scientific and Technical Computing Department, White Plains, New York. Fastook, J. and Holmlund, P. (1994). A glaciological model of the Younger Dryas event in Scandinavia. Journal of Glaciology, 40(134), 125-131. Fastook J.L. and Prentice, M. (1994). A finie-element model of Antarctica. Sensitivity test for meteorological mass balance relationship. Journalof Glaciology, 40(134), 167-175. Fromm, E. (1965). Beskrivning till jordartskartan 6ver Norrbottens l/in nedanftr lappmarksgr'~insen. Sveriges geologiska undersi~kning, C, 257 pp. Geikie, J. (1894). The Great Ice Age, and its Relation to the Antiquity of Man. Third edn, Edward Stanford, London 1894, 850 pp. Holmlund, P. (1993). Den senaste istiden i skandinavien. En modellering av
Weichselisen. A report to the Swedish Nuclear Power Inspectorate. Swedish text with an English summary, SKI Teknisk rapport, 93:44, 54 pp. Holmlund, P. and Fastook, J. (1993). Numerical modelling provides evidence of a Baltic Ice Stream during the Younger Dryas in Scandinavia. Boreas, 22(2), 77-86. Hoppe, G. (1959). Glacial morphology and inland ice recession in northern Sweden. GeografiskaAnnaler, 41(4), 193-212. Hoppe, G. (1963). Subglacial sedimentation with examples from northern Sweden. GeografiskaAnnaler, 45(1), 41-49. Hoppe, G. (1967). Case studies of deglaciation patterns. Geografiska Annaler, 49A(2-4), 204-212. Hughes, T.J. (1987). Ice dynamics and deglaciation models when ice sheets collapsed. In: Ruddiman, W.F. and Wright, H.E., Jr (eds), North American and Adjacent Oceansduring the Last Deglaciation. Boulder, Colorado, Geological Society of America, The Geology of North America, K-3:183-220. Johnsen, S.J., Clausen, H.B., Dansgaard, W., Fuhrer, K., Gundestrnp, N., Hammer, C.U., Iversen, P., Jouzel, J., Stauffer, B. and Steffensen, J.P. (1992). Irregular glacial interstadials recorded in a new Greenland ice core. Nature, 359, 311-313. Kleman, J. (1990). On the use of glacial striae for reconstruction of paleoice sheet flow patterns - - With application to the Scandinavian ice sheet. GeografiskaAnnaler, 72A(3-4), 217-236. Kleman, J. and Borgstrtm, I. (1990). The boulder fields of Mt. Fulufjiillet, west-central Sweden. GeografiskaAnnaler, 72A(1), 63-78. Lagerb~ick, R. (1988). Periglacial phenomenon in the wooded areas of northern Sweden - - Relicts from the T~,end6 interstadial. Boreas, 17, 487-500. Letrtguilly, A., Reeh, N. and Huybrechts, P. (1991). The Greenland ice sheet through the last glacial-interglacial cycle. Paleography,
Paleoclimatology, Paleoecology (Global Planetary Change Section), 90, 385-394. Ljungner, E. (1949). East-west balance of the Quaternary ice caps in Patagonia and Scandinavia. Bulletin of the Geological Institution, University of Uppsala, 33, 11-96. Lundqvist, J. (1958). Beskrivning till jordartskarta 6ver Viirrnlands l~in. Sveriges Geologiska Understkning, Ca38, 229 pp. Lundqvist, J. (1987). Glaciodynamics of the Younger Dryas marginal zone in Scandinavia. Implications of a revised glaciation model. Geografiska Annaler, 69A(2), 305-320. Mangernd, J., Andersen, S.T., Berglund, B.E. and Donner, J.J. (1974). Quaternary stratigraphy of Norden, a proposal for terminology and classification. Boreas, 3, 109-I 27. Nye, J. (1960). The response of glaciers and ice sheets to seasonal and climatic changes. Proceedings of the Royal Society, 256A(1287), 559-584. Reeh, N. (1991). The last interglacial as recorded in the Greenland Ice Sheet and Canadian Arctic ice caps. Quaternary International, 10--12, 123-142. Torell, O. (1873). Understkningar 6fver istiden. II. Skandinaviska inlandsisens utstr~ickning under isperioden. Ofversigt af Kungliga Vetenskaps-akademiensfi~rhandlingar, 30(1), 47--64. Weertman, J. (1957). On the sliding of glaciers. Journal of Glaciology, 3(21), 33-38. Weertman, J. (1964). The theory of glacier sliding. Journal of Glaciology, 5(39), 287-303.