Chris D Clark Sarah L Greenwood and David J A Evansf

*Department of Geography, University of Sheffield, Sheffield S10 2TN, UK

fDepartment of Geography, Durham University, South Road, Durham DH1 3LE, UK

* m









' &



r- i
















_ JST*


« -

' A








Reconstructing the extent, flow geometry and topography of former ice sheets has recently become more than an academic exercise because of the increasing perception of the importance of the cryosphere in climate change in the earth-ocean-atmosphere system. Of particular note is the discovery that punctuated delivery of freshwater from ice sheets into the oceans (via icebergs and ice-dammed lakes) has the ability to significantly perturb the thermohaline circulation, and as a consequence force extremely abrupt climate flips (e.g. Broecker, 1994). The abruptness of change raises concerns about the robustness of current (global warming) climate predictions and severely challenges our process knowledge of how the climate system works. In addition, a better understanding of palaeo-ice sheets will allow us to assess: (i) the effect of their vertical extent on atmospheric circulation and climate variation (e.g. Shin & Barron, 1989); (ii) the influence of land-ice volumes on global sea level and glacial rebound, leading also to an improved knowledge of mantle viscosity (e.g. Lambeck, 1993a,b); (iii) the effect of ice as an agent of long-term landscape evolution, specifically its impacts on uplift, tectonics and climate (e.g. Sugden et al., 1999); (iv) ice-sheet dynamics (e.g. Boulton et al., 2001); and (v) more national issues related to Quaternary icesheet deposits, such as the explanation of landscapes (e.g. Meehan & Warren, 1999), land-use planning, construction engineering, waste disposal and mineral exploitation (e.g. Eyles & Dearman, 1981; DiLabio & Coker, 1989; Merritt, 1992; Gray, 1993; McMillan, 2002).

In order to address the above, we need to work towards palaeo-geographicalal reconstructions of ice sheets that include: centres of initial nucleation and subsequent growth; the main centres of mass (ice divides) and margin configuration; migrations of ice divides and readjustments of the overall flow geometry and margins; major ice streams and calving bays; and retreat patterns to the final places of disappearance. Ideally we would be able to constrain the ice thickness (and hence volume) through the growth and decay phases, and know the timing of major events (i.e. maximum extent, ice-divide shifts, successive margin positions, activation of ice streams and major calving events, and chronology of ice retreat). Figures 50.1 & 50.2 schematically illustrate the type of palaeoglaciological reconstructions that we need to work towards.

If we possessed a wide knowledge of the palaeogeography of the last British-Irish Ice Sheet (BIIS) it would contribute considerably to the wider earth system science objectives outlined earlier in which ice-sheet volumes, configurations, discharge events and their timing are prioritized. However, we are far from being able to provide such complete information, even though more than 150yr of research and publication have uncovered many key fragments of proxy data and the British and Irish landscapes are rich with relevant data. Further field investigations, mapping programmes, more geochronometric dating and developments in the techniques used to map glacial geomorphology from satellite images and digital elevation models (DEMs) will undoubtedly help us to advance our palaeogeographical reconstructions. We argue, however, that the key weakness at present relates to a limited effort at assembling pieces of the jigsaw puzzle. This is not intended as a criticism of existing work, but derives from comparison of what is known about the BIIS compared against reconstructions of other palaeo-ice sheets such as the Laurentide and

Figure 50.1 A schematic reconstruction of the advance and retreat pattern, and surface topography of a hypothetical ice sheet. (a) If climatic conditions are appropriate ice may first nucleate in the highlands as separate ice caps, which eventually merge to form a large ice sheet. (b) The surface topography of the resulting maximal ice sheet is likely to reflect its growth history, but also might be conditioned by changes in atmospheric circulation and zones of moisture delivery arising from presence of the ice mass. (c) Climatic feed-back effects (moisture starvation in certain sectors), sea-level controls on calving flux, changes in thermal regime or the activation of ice streams might drive the ice sheet's morphology into a different configuration. For example, as illustrated, major ice streams might evacuate ice in sufficient quantities to drive the ice divide away from its original position. (d) The retreat pattern would now reflect such major configuration changes, might not be back towards the high ground, and may be asymmetric and differ considerably from the growth pattern. Retreat might not be to a single ice mass, but may fragment into component ice caps (e.g. consider the incomplete deglaciation of Iceland into its ice caps). We emphasize the hypothetical nature of the above, and use it merely to illustrate the kind of dynamism, migration of ice divides and asymmetry that it is reasonable to expect of an ice sheet's evolution. For convenience the topography of the Kamchatka Peninsula is used as a background, but in no way is the above based on any evidence, indeed glaciation of this region is highly controversial and poorly known.

Figure 50.2 An example of the kind of palaeoglaciological reconstruction it should be possible to work towards. Time-space diagram illustrating reconstructed margin positions, flow patterns and ice divide locations of the Quebec-Labrador sector of the Laurentide Ice Sheet. Note how growth and decay is asymmetric and that divide positions migrate. This synthesis is based on an inversion approach using glacial landforms to derive successive flow patterns, combined with information on available dating constraints and moraine positions. Data based on the reconstruction in Clark et al (2000). (See for colour version.)

Figure 50.2 An example of the kind of palaeoglaciological reconstruction it should be possible to work towards. Time-space diagram illustrating reconstructed margin positions, flow patterns and ice divide locations of the Quebec-Labrador sector of the Laurentide Ice Sheet. Note how growth and decay is asymmetric and that divide positions migrate. This synthesis is based on an inversion approach using glacial landforms to derive successive flow patterns, combined with information on available dating constraints and moraine positions. Data based on the reconstruction in Clark et al (2000). (See for colour version.)

Fennoscandian for which reconstructions of ice-divide locations, flow patterns, ice margins, ice streams and retreat patterns and timing are available (e.g. Dyke & Prest, 1987a; Dyke et al., 2003; Clark et al., 2000; Kleman et al., 1997; Boulton et al., 2001).

It is perhaps unrealistic to expect full reconstructions of the ice sheet and its dynamics through time, based solely on the geo-morphological and geological evidence that it left behind. This is because ice sheets do not always inscribe a record of their activity in the landscape, and because we still have incomplete knowledge of, and gaps in, what likely exists. The most profitable route of enquiry is to advance knowledge through combined use of numerical ice-sheet modelling (e.g. Boulton et al., 1985), isosta-tic inversion modelling (e.g. Lambeck, 1993a,b) and palaeoglaciological reconstruction driven by geomorphological and geological evidence. Intercomparisons between these approaches and hybrids combining them should ultimately yield the palaeoglacio-logical information required.

The motivation for this chapter is not to review existing knowledge about the BIIS (cf. Evans et al., 2005), but to clarify challenges, point to where efforts could be most usefully directed and highlight some recent developments that might take us towards an empirically realistic BIIS palaeoglaciology.

50.2 The British-Irish Ice Sheet

Presently we have neither a palaeogeographical overview of the regional landform and sedimentary evidence of the BIIS, nor an empirically driven palaeoglaciological reconstruction. This underlines the point made earlier about a lack of synthesis of the existing pieces of the jigsaw puzzle. For such overviews and reconstructions, we need to appeal to the outputs of numerical models, which to varying degrees have utilized field evidence as inputs or as (limited) validation. Such models are useful, because they aim to use the physics of, for example, ice motion or isostatic rebound, to extend our knowledge beyond what is known from the fragmentary geological record. Figure 50.3 illustrates four such reconstructions for the BIIS at its maximum extent in the Late Devensian. It is not sensible to make detailed comparisons between them because they have been derived by different methods of modelling and with different degrees and types of constraining field information. At the broad scale, however, similarities include: centres of mass over Ireland and the Highlands of Scotland and to varying degrees over the Southern Uplands and the Pennines and Wales. The southern extent of ice is similar and all portray ice cover over the Irish Sea. Ice limits elsewhere vary markedly, especially on the west coast of Ireland and Scotland and over the North Sea. Considerable debate has ensued as to whether the BIIS and Fennoscandian Ice Sheet were confluent over the North Sea, with most recent interpretations based on geological evidence confirming confluence at the LGM (Carr et al., 2000; Sejrup et al., 2003). Large differences exist between the models regarding ice thickness and volume. Over Scotland, for example, outputs suggest a range between 1000 and 2000 m for surface elevation. It is difficult to assess which is the most appropriate reconstruction of the ice sheet, and the best way of doing so requires careful comparison with the observational record of geomorphology and geology. For example, Ballantyne et al. (1998a,b) reconstructed the surface profile for northwest Scotland based on observed trimlines, and Gray (1995) has constrained isostatic rebound for the Southern Uplands based on former sea-level evidence and comments on the validity of the various models. Examples such as these, where field evidence is brought together to test the robustness of model outputs, are rare. It is clear that advances in knowledge of the palaeoglaciology of the BIIS critically depend on our ability to use field observations to guide or test numerical models. The onus is on modellers to make greater use of the observational record to either drive their models or test them, and the onus is on field researchers to make available their findings at a scale appropriate for this.

Many questions arise from mismatches between modelled simulations and field observations. Is the physics in the model flawed or missing a key ingredient (i.e. ice streaming)? Are the climate drivers inappropriate? Has the observational data been interpreted incorrectly or is the geochronology inaccurate? Advances depend on greater interaction between modelling and empirical approaches, because alone, it is unlikely that either approach can yield the answers. At present, the small number of published numerical models that exist for the BIIS remain largely unrelated to the wider observational record. Moreover, the field evidence remains fragmented into too many small parcels of information, thereby limiting model intercomparisons and coherent syntheses of the regional glacial geomorphology.

50.3 Observational-based palaeoglaciology: the main hurdles

Most investigations of the BIIS have been on a local to regional basis, which makes ice-sheet-wide synthesis difficult, especially where differences in interpretation between areas remain unresolved. Reconstructions of the whole ice-sheet geometry, based on available evidence, have rarely been attempted. The lack of synthesis, or reconstruction, may be attributed to the complexity and scale of the task. Barriers to the production of a coherent description of the BIIS are considered to be:

1 The fragmented nature of the evidence, i.e. many spatially separate studies with few links or even gaps between them, and many unresolved contradictions between areas.

2 The volume of information. Paradoxically, it might be argued, there is too much evidence. There has been so much written and mapped that it is daunting to attempt a synthesis.

3 Much of the data may be what Rhoads & Thorn (1993) call theory-laden evidence, i.e. as information has been collected over a long period of time, and during which glaciological ideas have changed considerably, it is likely that 'evidence' has been tainted by interpretations, some of which may no longer be valid. Some of the theory-laden evidence has likely propagated through the literature to add to the confusion.

4 Contemporaneity of evidence. In seeking to reconstruct icesheet geometry and extent based on the evidence available it is easiest and most convenient to assume that most evidence was formed penecontemporaneously as this provides maximum information about the ice sheet at a snapshot in time. However, this approach encourages contrived or unrealistic reconstructions that can be falsified in places by evidence that does not match or by implausible ice dynamics. We presume that most evidence is likely to relate to the pattern of deglaciation with underlying palimpsests of maximal or even build-up phases of ice-sheet configuration. Recognition of these multi-temporal aspects and an 'inversion' methodology for making sense of it (Kleman & Borgstrom, 1996, Clark, 1997; Kleman, this volume, Chapter 38) has led to advances in our understanding of other ice sheets, but has yet to be widely or systematically applied to the British Ice Sheet (cf. Salt, 2001).

5 Dating control. Much of the landform and stratigraphical evidence remains undated, and is thus difficult to fix in time and use in dynamic reconstructions.

6 Incomplete mapping. Some key parcels of information that could unlock important parts of the glacial history are likely unidentified and unmapped at present.

Figure 50.3 Surface topography of British-Irish Ice Sheet at its maximum (Late Devensian) extent, as reconstructed by four numerical models. The first three are ice sheet models partly driven by geomorphological data and the last is produced by iso-static inversion modelling. (a) from Boulton et al. (1977), (b) from Boulton et al. (1985), (c) from Boulton et al. (1991) and (d) Lambeck (1993b). Information redrawn from Gray (1995).

Figure 50.3 Surface topography of British-Irish Ice Sheet at its maximum (Late Devensian) extent, as reconstructed by four numerical models. The first three are ice sheet models partly driven by geomorphological data and the last is produced by iso-static inversion modelling. (a) from Boulton et al. (1977), (b) from Boulton et al. (1985), (c) from Boulton et al. (1991) and (d) Lambeck (1993b). Information redrawn from Gray (1995).

7 Glaciological plausibility. It is likely that many evidence-based reconstructions of ice-sheet flow and geometry may in fact be glaciologically implausible. Greater use of modelling may help to eliminate these cases and provide alternatives. In addition to modelling we can use contemporary analogues such as the Antarctic Ice Sheet, but this may be misleading because of its predominantly cold-based thermal regime and its climatic setting.

8 An objective methodology. There appears to be little consensus regarding the method by which we utilize observations to build a reconstruction. There are many different approaches taken. At the worst we could (perhaps justifiably) be accused of

'story-building' by attempting to fit patterns of retreat or flow geometry to best explain our own observations and data. To remain objective it is essential that the basis of the methodology is well described, with all assumptions clearly stated, and actual evidence kept distinct from interpretation and interpolation.

50.4 Improved utilization of existing observations

From points 1-3 above and our earlier comments it is clear that it is difficult to make use of published knowledge because there is so much written (at least 2000 papers), and which is also difficult to assess because of propagation of theory-laden evidence, and because of fragmentation with numerous small and separate studies. Critical literature reviews can synthesize and overcome some of these problems, and when performed have significantly helped to advance knowledge (e.g. Sutherland, 1984; Ehlers et al., 1991; Ehlers & Gibbard, 2004). However, for reasons of brevity they are likely to be selective, and focus on the generalized picture rather than on localized details. Also, it is questionable whether the richness in data generated over the past 100 yr can be adequately assessed and presented in a journal article or book chapter when the more appropriate format is a map or map series. Ideally, we require a map of Britain and Ireland and the surrounding continental shelf on which all geomorphological, geological and geochronometric information (i.e. moraines, striae, till limits, dates, etc) relating to the last ice sheet is plotted. This would be the jigsaw puzzle assembled, from which ice-sheet-wide reconstructions could be attempted. Notably, for the Laurentide Ice Sheet such a 'glacial map' has existed since 1968 (Prest et al., 1968; Fulton, 1995) and numerous reconstructions have been derived from it. We argue that a similar approach is required for Britain and Ireland, and would likely yield similar and significant advances in our knowledge.

A recent project, called BRITICE, has partly addressed this issue by reviewing all the relevant literature from academic journals and PhD theses (Evans et al., 2005) and 100yr of BGS maps of onshore and offshore geology (Clark et al., 2004). Information was extracted and entered into a geographical information system (GIS), from which a 'glacial map' (1: 625,000 scale, 1 x 1.6 m) has been produced (Clark et al., 2004). This project focused primarily on geomorphological features that inform us about the last ice sheet, and includes the following information: moraines, eskers, drumlins, meltwater channels, tunnel valleys, trimlines, limit of key glacigenic deposits, glaciolacustrine deposits, ice-dammed lakes, erratic dispersal paths and shelf-edge fans. The GIS contains over 20,000 individual features split into thematic layers (as above). Figure 50.4 provides an overview of the database and Fig. 50.5 is an enlarged extract illustrating some of the types of data and their distribution. The task is incomplete in that some information was excluded, such as striae and geochronometric dates, but it does represent the most complete assembly of the BIIS jigsaw puzzle to date. The purpose of the work was to bring together valuable information from the literature in the hope that it can help motivate the following.

1 Evidence-based reconstruction of the ice sheet. Meltwater channels for example could be combined with the moraines, ice-dammed lakes and eskers to build a sequential pattern of glacier retreat, to infer palaeo-ice dynamics and inform palaeoclimate reconstructions (cf. landsystems approach; Evans, 2003a). The drumlin and erratic-pathway data could be analysed and enhanced by mapping from digital elevation models (DEMs) and satellite images to derive the ice-flow patterns. These could be used to reconstruct changes in flow geometry and ice-divide positions through time. Ideally a full inversion approach utilizing all the data could be attempted and constrained by available stratigraphical and dating evidence.

2 Numerical ice-sheet modelling. Modelling has become increasingly important as an aid to reconstructing ice sheets and for assessing their relationship to other factors such as sea-surface temperatures, climate and sediment discharge (cf. Siegert, 2001). Flow-pattern or ice-extent information can be used to drive the modelled reconstructions or as validation of the modelled result, or in some combination. An example of the former was presented by Boulton et al. (1977), who used flow-patterns to constrain British Ice Sheet geometry (Fig. 50.3a) and from this derived a modelled estimate of the surface topography. The alternative is to 'grow' an ice sheet over the topography using climate drivers (i.e. derived from ice-core records) and then assess the plausibility of the modelled ice sheet by comparing ('testing') it with empirically derived evidence of ice-flow configuration. It is frequent to hear field investigators criticize the work of ice-sheet modellers because they sometimes fail to use the wealth of geological information available and because of a lack of testing of their results, against what is known. Such criticism is perhaps unfair given that evidence is so rarely compiled in a consistent format so as to make it useable. It is hoped that the BRITICE GIS compilation, made freely available via the world-wide web, will facilitate increased use of geomorphological data in modelling experiments.

3 Directing fieldwork. It is evident from the Glacial Map (Fig. 50.4) that although there is a reasonable spread of information across the ice-sheet bed, there are notable gaps and great variability in data density. This compilation may assist field workers in choosing areas for future investigation, mitigating the tendency within the academic literature to keep reinvestigating the same area.

Conflicts and discrepancies exist in the data recorded on the Glacial Map and in the BRITICE GIS database. For example, some confusion is likely to exist with regard to the true age of some of the features. The inclusion of some pre-Devensian and Loch Lomond Stadial features is likely in some locations where dating control is poor. It is hoped that, in addition to the points above, compilations such as BRITICE may encourage greater scrutiny of valuable published work.

50.5 Case study: using meltwater channels to constrain ice-sheet retreat patterns

Here we illustrate how some of the data layers in the BRITICE GIS can be used to build a preliminary reconstruction of the

0 0

Post a comment