Department of Geography University of Edinburgh Edinburgh EH8 9XP UK

Examining changes in past ice cover across Iceland is enlightening owing to the many feedbacks that are apparent between ice, atmosphere, ocean and the lithosphere. Unravelling this interaction presents a tough job but is important from three related perspectives:

1 Iceland is influenced by an extreme maritime climate which yields ice-caps with large turnover and outlet glaciers that are both sensitive and responsive. Hence, changes in long-term North Atlantic circulation will have an immediate and significant impact on glacier dynamics and extent.

2 Its location astride the mid-Atlantic ridge means that Iceland experiences extreme geothermal conditions, in certain cases leading to intense subglacial volcanism and concomitant meltwater flooding or Jokulhlaups. This is not only a contemporary concern but has been a key control on past ice dynamics (Jull & McKenzie, 1996).

3 Sediments originating from Iceland have been found in the lower (i.e. earlier) sequences of Heinrich IRD layers, suggesting that large and critically located pulses associated with the collapse of the ice sheet during the Late-glacial may have affected North Atlantic Deep Water (NADW) formation (Grousset et al., 2000). Rapid isostatic rebound, which promotes enhanced mantle upwelling, means that volcanic activity may have been a hundredfold greater during glacial unloading compared with the present (Maclennan et al., 2002) and therefore has the potential to trigger widespread ice-sheet collapse.

Investigating these interactions is a complex and multidisci-plined task involving empirical data from a variety of sources, but is one in which numerical modelling potentially plays a pivotal role. One significant obstacle to a comprehensive Late-glacial reconstruction is chronological uncertainty in the sedimentary and geomorphological records. It is generally believed though that the Weichselian glaciation most likely culminated in its maximum extent at ca. 21kyrBP (the Last Glacial Maximum— LGM) and terminated ca. 10kyrBP (Norddahl, 1991). Strong evidence for the extent, thickness and flow of the former ice sheet certainly persists but apart from a handful of key ocean cores, the pertinent evidence (in particular the terrestrial geomorphology) is invariably undated. Furthermore, the piecemeal offshore record (summarized in Fig. 22.1) remains equivocal. Limits are identified either from bathymetric or seismic surveys or are inferred from radiocarbon dated basal diamicton from recovered cores. Dates from Marine Oxygen Isotope Stages (MIS) 2 and 3 have been obtained from sediments recovered from Hunafloaall (north Iceland), Djupall (northwest Iceland) and Latra Bank (west Iceland) (Andrews et al., 2000; Andrews, in press). Furthermore, one core at Djupall, 50km off the northwest, indicates that the site has been unglaciated for 30.914CkyrBP and thus provides a definitive constraint on LGM extent in this sector. Elsewhere, the absolute extent of the LGM ice sheet remains ambiguous but is placed at the shelf edge in the south and west. Grimsey Island, off northern Iceland shows extensive evidence of ice erosion, hence the LGM limit is inferred to be more extensive in this sector also.

Here, a three-dimensional coupled ice-flow-degree-day model is used to provide an experimental framework by which the interactions between ice-atmosphere-lithosphere can be investigated, validated and refined against available observation data. Such modelling provides the classic link between form (the erosional and depositional record) and process (climate change, glacier-meteorology, geothermal activity and ice-dynamics). Specifically,

Figure 22.1 Map of Iceland showing key locations referred to in the text. Also shown are critical basal marine core dates, evidence of former ice limits and the main ice caps in existence today. The 200 m bathymetric contour is marked for reference and all dates are in k14CyrBP. (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.1 Map of Iceland showing key locations referred to in the text. Also shown are critical basal marine core dates, evidence of former ice limits and the main ice caps in existence today. The 200 m bathymetric contour is marked for reference and all dates are in k14CyrBP. (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.2 Modelled present-day ice surface using a two-stepped cooling of 2°C for 1000yr followed by 200yr at —1°C. Also shown are the modelled ice-sheet extents associated with 2000 yr of cooling of 3,4 and 5°C under the present precipitation regime and no sea-level change. (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.2 Modelled present-day ice surface using a two-stepped cooling of 2°C for 1000yr followed by 200yr at —1°C. Also shown are the modelled ice-sheet extents associated with 2000 yr of cooling of 3,4 and 5°C under the present precipitation regime and no sea-level change. (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.3 The optimum LGM modelled ice-sheet surface and its flow regime most compatible with the available offshore evidence corresponding to a cooling of 12.5°C and a 35% decrease in precipitation with an additional 30% suppression applied north of the 65th parallel. (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.3 The optimum LGM modelled ice-sheet surface and its flow regime most compatible with the available offshore evidence corresponding to a cooling of 12.5°C and a 35% decrease in precipitation with an additional 30% suppression applied north of the 65th parallel. (See www.blackwellpublishing.com/knight for colour version.)

the time-dependent, thermomechanical model is applied to a 2km grid and requires distributions of topography, geothermal heat flux, temperature and precipitation. It enables the variables of ice thickness, bed adjustment, flow and temperature to interact freely and caters for the dynamics of ice-shelves, calving and thermally triggered basal sliding through longitudinal coupling. It is forced through perturbations from present by changes in sea level, precipitation and temperature and yields space-time distributions of ice thickness, isostatic response, stress, velocity, temperature, melt and iceberg flux. As an initial test, Iceland's contemporary ice cover is well replicated from ice-free initial conditions given 1000 years of 2°C cooling followed by 200 yr at 1°C (Fig. 22.2). Experiments also reveal that given its present rainfall regime, Iceland is highly susceptible to widespread glaciation, with an ice sheet advancing to the southern coastline given just 3°C annual cooling and beyond it in many quarters with 4 to 5°C cooling (Fig. 22.2).

Modelling the LGM presents a challenge though, given that virtually no local palaeoclimatic data exist. Assuming though, that the form, if not the magnitude, of the forcing climate signal took that of the GRIP oxygen isotope curve, then a 'shot-gun' approach may be adopted. Fifteen experiments were initiated from 24kyr BP for 3kyr using the GRIP isotope curve scaled for maximum cooling at 21kyrBP ranging from 5 to 15°C in 2.5°C increments, each with a corresponding 20, 35 and 50% suppression of today's precipitation. Two of these experiments generally matched the offshore limits but all overran appreciably to the north. The optimal LGM configuration required 10 to 12.5°C of cooling with 35% overall precipitation reduction, with a further 30% enhanced aridity across the north. This results in a large, offshore ice sheet with an area of 3.29 x 105km2 and a volume of 2.38 x 105km3. Sensitivity experiments highlight the central role played by geothermal activity in activating extensive zones of basal melting, yielding a highly dynamic and low-aspect ice sheet with a mean thickness of ca. 800 m drained by numerous ice streams extending far into the interior and breached by numerous nunataks (Fig. 22.3).

Norddahl (1983) uses trimlines and the upper limit for evidence of glacial erosion and deposition on nunataks to reconstruct the LGM ice-sheet surface in north Iceland across the Trollaskagi and Flateyjarskargi mountains. The high level of correspondence observed between the modelled profiles down both Eyjafjordur and Bárdardalur ice streams and those reconstructed by Norddahl (1983) indicates not only that the modelled LGM is coherent with this evidence but that the model is also replicating the dynamics of these outlet ice streams satisfactorily (Fig. 22.4). The low-lying aspect of the modelled LGM is also consistent with the profile of the 'Pleistocene ice sheet' reconstructed by Walker (1965) from the summit altitudes of palagonite tuff-breccia (Móberg Mountains). These 'table-mountain' volcanoes form subglacially, but summit capping of subaerial lavas indicates that they topped-out at the level of the palaeo-ice-sheet surface.

Comparison between the orientation of ice-contact features indicating past ice flow with basal velocity vectors from the modelled LGM is also encouraging, even if the result is not entirely unambiguous (Fig. 22.5a). The main issue is whether the features observed can be considered chronosynchronous with the LGM. However, assuming that ice thickness, volume, extent and hence

Flowline distance [km]

Figure 22.4 The long profile of Bardardalur fjord showing the basal topography, its mountain relief, the modelled optimum LGM ice stream and that reconstructed by Norddahl (1991) from trimline evidence. Superimposed are the two long-profiles modelled using the standard value for geothermal heat flux (G = 54.2 mWm-2) applied across the model domain and zero basal sliding (As = 0). (See www.blackwellpublishing.com/knight for colour version.)

Flowline distance [km]

Figure 22.4 The long profile of Bardardalur fjord showing the basal topography, its mountain relief, the modelled optimum LGM ice stream and that reconstructed by Norddahl (1991) from trimline evidence. Superimposed are the two long-profiles modelled using the standard value for geothermal heat flux (G = 54.2 mWm-2) applied across the model domain and zero basal sliding (As = 0). (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.5 (a) Modelled basal velocity from the optimum LGM experiment overlain with the orientation of observed ice directional features (adapted from Bourgeois et al., 2000) superimposed with the direction of the corresponding modelled basal vector for the same location. (b & c) Rose diagrams of modelled basal velocity vectors and observed orientations of ice directional features for the two locations indicated. (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.5 (a) Modelled basal velocity from the optimum LGM experiment overlain with the orientation of observed ice directional features (adapted from Bourgeois et al., 2000) superimposed with the direction of the corresponding modelled basal vector for the same location. (b & c) Rose diagrams of modelled basal velocity vectors and observed orientations of ice directional features for the two locations indicated. (See www.blackwellpublishing.com/knight for colour version.)

10 1000

10 1000

20,000

1 8,000

22,000

20,000

18,000

16,000

12,000

16,000 14,000

Cal years BP

12,000

10,000

8,000 H

20,000

10 1000

10 1000

20,000

16,000 14,000

Cal years BP

Figure 22.6 Modelled time-series from LGM through to early Holocene of: (a) predicted mean ELA, (b) ice sheet area, (c) ice volume and (d) bulk freshwater runoff.

8,000

Figure 22.6 Modelled time-series from LGM through to early Holocene of: (a) predicted mean ELA, (b) ice sheet area, (c) ice volume and (d) bulk freshwater runoff.

a b c d potential 'work-done' on the substrate were maximized during the LGM, then it may be argued that a significant proportion of these features should have formed during this time. A broad comparison reveals numerous areas displaying a high level of consistency, but also some specific areas under the ice divide that are divergent, or where there is no basal sliding component modelled. Detailed comparison of model and observed orientations in two specific localities, within the northern fjords and the southern central lowlands, are, though, encouraging (Fig. 22.5b & c). In both cases modelled vectors not only correspond but are 'tighter' than observed orientations, reflecting the alternative flow regimes associated with a subsequent, less extensive ice sheet during deglaciation.

From this optimum LGM configuration the model is integrated forward in time using the scaled GRIP record from 21 kyr BP to the early Holocene. Output time-series (Fig. 22.6b, c & d) reveal a highly dynamic ice sheet characterized by large and rapid fluctuations in area, volume and bulk runoff (melt and calving flux). A most notable pulse of up to 300km3yr-1 occurs at ca. 14 kyr BP, for a period of ca. 100 yr during a period of ice-sheet retrenchment onshore which is synonymous with the Heinrich 1 event. Such an event will have been amplified with the onset of volcan-

ism due to magma upwelling in the crust after the initial deglaciation episode up to 16kyrBP. By 12kyrBP though, the ice sheet readvances in the Younger Dryas episode during which ice expands to the present-day coastline (Fig. 22.7a), although the ice cap over the northwest remains independent of the main ice and many areas, especially the main peninsulas and northern highlands, escape inundation.

Comparison of modelled ice-extent with mapped shorelines and end moraines in the north, northeast and southeast (Fig. 22.7b, c & d) demonstrates that the model accurately creates a coherent three-dimensional reconstruction of the Younger Dryas, which matches available observation data and which, given the freely determined nature of the simulation, corroborates its output and robustness. To summarize it is hoped that this modelling effort demonstrates that the unique climatic, volcanic and oceanic setting of Iceland had a major impact on the Late-glacial ice sheet rendering it highly dynamic with significant potential for rapid collapse, which through the resulting salinity changes of the Nordic Sea could have had substantial impacts extending well beyond Iceland's immediate coastline.

Coupled, time-dependent modelling is one means of unifying the reconstruction of the late Weichselian Icelandic ice sheet,

Figure 22.7 (a) The modelled Younger Dryas ice-sheet geometry and associated flowlines compared with empirical reconstructions for: (b) the northern fjords, (c) the northeast and (d) the southeast based on geomorphological mapping and dating of end moraines, trimlines and raised shorelines (Norddahl & Petursson, in press). (See www.blackwellpublishing.com/knight for colour version.)

Figure 22.7 (a) The modelled Younger Dryas ice-sheet geometry and associated flowlines compared with empirical reconstructions for: (b) the northern fjords, (c) the northeast and (d) the southeast based on geomorphological mapping and dating of end moraines, trimlines and raised shorelines (Norddahl & Petursson, in press). (See www.blackwellpublishing.com/knight for colour version.)

providing a coherent framework by which the underlying link between form and process can be explored and understood. Specifically, it allows for the controlled and systematic perturbation of an envelope of external environmental forcing variables along with those internal parameters governing ice dynamics, to determine the sensitivity and response of the climate-ice-sheet-landscape system. Significantly, the approach enables the trajectory of multiple virtual ice sheets to be investigated and, ultimately, guided to be optimally compatible with the on- and offshore glacial records. Here, such a framework has been specifically used to: (i) investigate the palaeoclimatic change required to yield an LGM ice sheet which is optimal with the empirical record; (ii) elucidate the mechanisms that control its response; and (iii) provide insight into the limits and geometry of the palaeo-ice sheet in areas where field evidence is deficient.

Was this article helpful?

0 0

Post a comment