ArticlePDF Available

Abstract and Figures

We present results from a suite of forward transient numerical modelling experiments of the British and Irish Ice Sheet (BIIS), consisting of Scottish, Welsh and Irish accumulation centres, spanning the last Glacial period from 38 to 10 ka BP. The 3D thermomechanical model employed uses higher-order physics to solve longitudinal (membrane) stresses and to reproduce grounding-line dynamics. Surface mass balance is derived using a distributed degree-day calculation based on a reference climatology from mean (1961–1990) precipitation and temperature patterns. The model is perturbed from this reference state by a scaled NGRIP oxygen isotope curve and the SPECMAP sea-level reconstruction. Isostatic response to ice loading is computed using an elastic lithosphere/relaxed asthenosphere scheme. A suite of 350 simulations were designed to explore the parameter space of model uncertainties and sensitivities, to yield a subset of experiments that showed close correspondence to offshore and onshore ice-directional indicators, broad BIIS chronology, and the relative sea-level record. Three of these simulations are described in further detail and indicate that the separate ice centres of the modelled BIIS complex are dynamically interdependent during the build up to maximum conditions, but remain largely independent throughout much of the simulation. The modelled BIIS is extremely dynamic, drained mainly by a number of transient but recurrent ice streams which dynamically switch and fluctuate in extent and intensity on a centennial time-scale. A series of binge/purge, advance/retreat, cycles are identified which correspond to alternating periods of relatively cold-based ice, (associated with a high aspect ratio and net growth), and wet-based ice with a lower aspect ratio, characterised by streaming. The timing and dynamics of these events are determined through a combination of basal thermomechanical switching spatially propagated and amplified through longitudinal coupling, but are modulated and phase-lagged to the oscillations within the NGRIP record of climate forcing. Phases of predominant streaming activity coincide with periods of maximum ice extent and are triggered by abrupt transitions from a cold to relatively warm climate, resulting in major iceberg/melt discharge events into the North Sea and Atlantic Ocean. The broad chronology of the modelled BIIS indicates a maximum extent at ∼20 ka, with fast-flowing ice across its western and northern sectors that extended to the continental shelf edge. Fast-flowing streams also dominate the Irish Sea and North Sea Basin sectors and impinge onto SW England and East Anglia. From ∼19 ka BP deglaciation is achieved in less than 2000 years, discharging the freshwater equivalent of ∼2 m global sea-level rise. A much reduced ice sheet centred on Scotland undergoes subsequent retrenchment and a series of advance/retreat cycles into the North Sea Basin from 17 ka onwards, culminating in a sustained Younger Dryas event from 13 to 11.5 ka BP. Modelled ice cover is persistent across the Western and Central Highlands until the last remnant glaciers disappear around 10.5 ka BP.
Content may be subject to copyright.
Dynamic cycles, ice streams and their impact on the extent, chronology
and deglaciation of the British–Irish ice sheet
Alun Hubbard
, Tom Bradwell
, Nicholas Golledge
, Adrian Hall
, Henry Patton
, David Sugden
Rhys Cooper
, Martyn Stoker
Department of Physical Geography & Quaternary Geology, Stockholm University, SE-106 91, Sweden
Institute of Geography & Earth Sciences, Aberystwyth University, Aberystwyth SY23 3DB, UK
British Geological Survey, Murchison House, West Mains Road, Edinburgh EH9 3LA, UK
Institute of Geography, Drummond Street, University of Edinburgh, Edinburgh EH9 9XP, UK
article info
Article history:
Received 17 June 2007
Accepted 29 December 2008
We present results from a suite of forward transient numerical modelling experiments of the British and
Irish Ice Sheet (BIIS), consisting of Scottish, Welsh and Irish accumulation centres, spanning the last
Glacial period from 38 to 10 ka BP. The 3D thermomechanical model employed uses higher-order physics
to solve longitudinal (membrane) stresses and to reproduce grounding-line dynamics. Surface mass
balance is derived using a distributed degree-day calculation based on a reference climatology from
mean (1961–1990) precipitation and temperature patterns. The model is perturbed from this reference
state by a scaled NGRIP oxygen isotope curve and the SPECMAP sea-level reconstruction. Isostatic
response to ice loading is computed using an elastic lithosphere/relaxed asthenosphere scheme. A suite
of 350 simulations were designed to explore the parameter space of model uncertainties and sensitiv-
ities, to yield a subset of experiments that showed close correspondence to offshore and onshore ice-
directional indicators, broad BIIS chronology, and the relative sea-level record. Three of these simulations
are described in further detail and indicate that the separate ice centres of the modelled BIIS complex are
dynamically interdependent during the build up to maximum conditions, but remain largely independent
throughout much of the simulation. The modelled BIIS is extremely dynamic, drained mainly by
a number of transient but recurrent ice streams which dynamically switch and fluctuate in extent and
intensity on a centennial time-scale. A series of binge/purge, advance/retreat, cycles are identified which
correspond to alternating periods of relatively cold-based ice, (associated with a high aspect ratio and net
growth), and wet-based ice with a lower aspect ratio, characterised by streaming. The timing and
dynamics of these events are determined through a combination of basal thermomechanical switching
spatially propagated and amplified through longitudinal coupling, but are modulated and phase-lagged
to the oscillations within the NGRIP record of climate forcing. Phases of predominant streaming activity
coincide with periods of maximum ice extent and are triggered by abrupt transitions from a cold to
relatively warm climate, resulting in major iceberg/melt discharge events into the North Sea and Atlantic
Ocean. The broad chronology of the modelled BIIS indicates a maximum extent at w20 ka, with fast-
flowing ice across its western and northern sectors that extended to the continental shelf edge. Fast-
flowing streams also dominate the Irish Sea and North Sea Basin sectors and impinge onto SW England
and East Anglia. From w19 ka BP deglaciation is achieved in less than 2000 years, discharging the
freshwater equivalent of w2 m global sea-level rise. A much reduced ice sheet centred on Scotland
undergoes subsequent retrenchment and a series of advance/retreat cycles into the North Sea Basin from
17 ka onwards, culminating in a sustained Younger Dryas event from 13 to 11.5 ka BP. Modelled ice cover
is persistent across the Western and Central Highlands until the last remnant glaciers disappear around
10.5 ka BP.
Ó2009 Elsevier Ltd. All rights reserved.
1. Introduction
The dynamics of the British and Irish Ice Sheet (BIIS) during the
last Glacial cycle are still poorly understood. This has resulted in
*Corresponding author. Institute of Geography & Earth Sciences, Aberystwyth
University, SY23 3DB Aberystwyth, United Kingdom.
E-mail address: (A. Hubbard).
Contents lists available at ScienceDirect
Quaternary Science Reviews
journal homepage:
0277-3791/$ – see front matter Ó2009 Elsevier Ltd. All rights reserved.
Quaternary Science Reviews 28 (2009) 759–777
a number of apparently contradictory hypotheses regarding the
nature, volume and extent of the last BIIS (e.g. Sutherland, 1984;
Stoker et al., 1993; Sejrup et al., 1994, 20 05; Ballantyne et al., 1998;
Bowen et al., 2002; Hall et al., 2003). Consequently, fundamental
questions remain, including:
Where were the ice-sheet accumulation centres and divides?
Were these stable or transitory features throughout the course
of the ice sheet’s evolution?
Was its maximum advance long-lived and climatically driven,
or relatively short reflecting flow instabilities?
How thick was the ice, and what was its configuration and total
volume at maximum extent?
What was its subglacial thermodynamic and hydrological
What was the palaeo-climatic regime of the ice sheet?
Given the scale of these outstanding questions, investigating the
fluctuations, dynamics and palaeo-climate of the last BIIS is an
important undertaking. Furthermore, considered in the context of
recent observations of rapid but episodic glacier acceleration and
thinning from marine-terminating sectors of the Greenland and
West Antarctica Ice Sheets (Shepherd and Wingham, 2007), this
work is especially pertinent. Understanding the demise of an
accessible palaeo-ice sheet with large offshore, marine-based
sectors can provide an analogue, as well as insight and vital context
to the rapid-paced fluctuations and deglaciation observed at the
margins of contemporary polar ice sheets.
The task of reconstructing the BIIS is an exacting one due to the
broad range of palaeo-climatic, environmental and glaciological
conditions under which the ice sheet existed. This is reflected in the
rich and diverse Late Pleistocene geology and geomorphology
found in the British Isles and on the adjacent continental shelf (e.g.
Clark et al., 2004; Bradwell et al., 2008; Greenwood and Clark,
2008). Despite this complexity, the last decade has seen a broad
consensus, or at least a convergence, of views regarding the BIIS;
leading to the notion of an extensive, fast-flowing but relatively
low-aspect maritime ice sheet, the margins of which were groun-
ded below sea-level (e.g. Knutz et al., 2001; Sejrup et al., 2005;
Boulton and Hagdorn, 2006; Carr et al., 2006; Bradwell et al., 2007,
2008; Peck et al., 2007; O
´Cofaigh and Evans, 2007). This new
paradigm has emerged from technological and methodological
developments in geomorphological field-research and data inter-
pretation, particularly in high-resolution airborne remote-sensing,
GIS-based mapping, cosmogenic isotope, luminescence and radio-
carbon dating; along with improved data capture offshore,
including extensive single- and multi-beam bathymetry, real-time
side-scan sonar, and seismic reflection profiling. The combined
evidence indicates an ice sheet characterised by numerous inde-
pendent cold-based accumulation centres along the main upland
belts (e.g. Hall and Glasser, 2003; Ballantyne et al., 2007; Jansson
and Glasser, 2008; Phillips et al., 2006) which were drained and
drawn-down by a series of basally lubricated, fast-flowing, outlet
glaciers and ice streams (e.g. Clapperton, 1971; Stoker and Bradwell,
2005; Golledge and Stoker, 2006; O
´Cofaigh and Evans, 2007;
Bradwell et al., 2008). These ice streams were determined by local
topography onshore, but were mobilised by widespread basal
decoupling over saturated deforming sediments offshore (e.g.
Boulton et al., 1985; Merritt et al.,1995; Bradwell et al., 2007). In the
NW sector of the BIIS, many ice streams and outlet lobes extended
to the continental shelf edge, where they calved into the NE Atlantic
Ocean (Stoker et al., 1993; Knutz et al., 2001; Bradwell et al., 2007;
Peck et al., 2007). In the NE sector ice streams converged at
maximal conditions with Scandinavian ice flowing into the North
Sea Basin (Sejrup et al., 1994, 2005; Graham et al., 2007; Bradwell
et al., 2008). In the southern sector, ice streams terminated on land,
possibly in large proglacial lakes, or in shallow glacio-marine
settings such as the Celtic Sea (Boulton, 1986; Scourse and Furze,
2001; Hiemstra et al., 2006; O
´Cofaigh and Evans, 2007). In the past
decade, the geomorphic and geological imprint of these fast-
flowing corridors in Britain and Ireland have been identified and
mapped using high-resolution digital elevation models. The
identification of mega-scale glacial lineations on the continental
shelf, now at water depths of over 100 m, is taken as primary
evidence that large grounded ice streams drained the NE and NW
sectors of the BIIS (Bradwell et al., 2007, 2008; Graham et al., 2007,
For promoting the dynamic paradigm of the BIIS over the past
three decades, the work of G.S. Boulton (GSB) must be acknowl-
edged. Since the mid-1970s, GSB and his co-workers have adopted
a hybrid approach – combining state-of-the-art modelling
methods, with the prevalent thinking informed by general glacio-
logical principles, and the available geological evidence, to
compose major reconstructions of the ice-sheet form and associ-
ated processes. This research effort has seen a natural evolution –
renewed attempts at BIIS reconstruction being inspired through
improved data sets and, significantly, ideas imported from process
glaciology. In brief, from 1977 onwards GSB’s maximum BIIS
reconstructions have, in essence, experienced sequential reduc-
tions in the critical basal shear-stress (at a rate of around 50% per
decade) which supports and stabilises the ice sheet, from 100 kPa
(Boulton et al., 1977) to 30 kPa (Boulton et al., 1991) in offshore
zones, down to 15–25 kPa (Boulton and Hagdorn, 2006) under
mobile outlet lobes in the most recent time-dependent model. Such
progressive reductions in basal shear-stress have been inspired by
glaciological field insights into the rheology of deforming beds
underneath the Siple Coast ice streams of the WAIS (Alley et al.,
2005). The direct consequence of reducing the basal traction – i.e.
the force (or ‘friction’) that a unit area of the ice–sediment–
substrate interface can support – yields an ice-sheet reconstruction
characterised by progressively lower aspect ratios and a less
viscous ice-mass which inexorably expands to the continental shelf
These modelling developments culminated in the Boulton and
Hagdorn (2006) study (hereafter referred to as B&H), in which
a time-dependent thermomechanical numerical ice-flow model
was applied under varying, user-perturbed, boundary conditions
and forcing functions to provide a set of simulations that best
replicated the geological evidence, with particular emphasis on
simulating the lobes that occupied the southern flanks of the
British mainland within the North Sea and Irish Sea Basins. The
overall form and dynamics of the B&H simulations is firmly and
fundamentally couched in the inclusion of fast-flowing ice stream-
type processes, achieved by preconditioning certain marine sectors
of the model domain with a high decollement’ (i.e. sliding)
parameter. B&H state that their basal preconditioning is coincident
with the existence of extensive soft-sediments that promote basal
deformation, and is the only plausible conditioning that can ach-
ieve the well-defined and geologically constrained outlet lobes
which existed along the low-gradient southern sectors of the North
Sea and Irish Sea Basins. Furthermore, B&H argue that the basal
preconditioning of their model domain is a fundamental prereq-
uisite to the continuous activity of the ice streams and surge lobes
throughout their simulation, which are essential to the effective
draw-down of the central ice-sheet surface, as indicated by the key
palaeo-trimline evidence of Ballantyne et al. (1998). B&H conclude
that the best simulations are those with fixed (preconditioned) ice
streams with velocities of between 500 and 1000 ma
which draw
down the ice-sheet surface and account for 60–84% of the ice flux to
the margin. The impact of such a persuasive reconstruction is
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777760
significant in that it provides a conceptualisation of the BIIS which
is nicely in fashion with that emerging from field observation.
2. Aims and objectives
The implications of such a dynamically enabled BIIS at the Last
Glacial Maximum (LGM), especially in terms of its ice volume,
aspect ratio, sensitivity and response time are far-reaching. The
coupled atmospheric – cryospheric – oceanic system will exhibit
behaviour that partially decouples it from the forcing which initi-
ates and drives it in the first instance. Furthermore, system
behaviour will be complex, with asynchronous ice-maximal extent,
thickness, sea-level change, isostatic loading, together with rapidly
fluctuating runoff and calving-flux. Its behaviour will also become
quasi-stochastic and highly non-linear at critical stages in the ice
sheet’s geometrical and thermodynamical evolution, especially
during phases of basal reconfiguration and concomitant fast-flow.
Drawing all of the available strands of evidence together into
a holistic and coherent framework is a challenge but one in which
numerical modelling has a pivotal and unifying role.
This study aims to build upon B&H’s approach and to this end,
we accept as a fundamental basis, the dynamic, ice-stream-domi-
nated behaviour simulated in their study and extend it, particularly
into the chronological domain, where there are still significant
outstanding questions to be examined. Specifically, we implement
an ensemble suite of numerical experiments in order to explore the
internal and external parameter space, with a motivation to
(1) the potential bounds on reconstructed BIIS extent, thickness,
volume and dynamics over the course of its build up and
(2) the characteristics of the palaeo-climatic envelope that best
yields a BIIS coherent with the observational record, and,
(3) the interaction, phase-lags and extent of any decoupling/non-
linear behaviour between the imposed climate forcing and
dynamic ice-sheet response and its geometrical evolution.
In addition to these objectives, we also address the critically
climatic and rheological assumptions that B&H adopt on a prag-
matic basis, which in the light of recent developments, especially in
marine-ice-sheet theory, can be re-assessed. Specifically, we
implement an improved treatment of the shallow ice approxima-
tion (SIA) employed in the GLIMMER ice-sheet model by utilising
a first-order solution of the ice-flow equations that incorporates
longitudinal stress gradients and enables a more realistic repre-
sentation of ice-stream processes (Paterson,1994). The SIA assumes
that the force balance of each discrete modelled column of ice can
be considered independent of others around it and that its gravi-
tational driving force (or basal shear-stress) is exactly balanced by
the traction (local friction) at its bed. Such an assumption poorly
represents the processes operating under ice streams where
extensive zones are hydraulically lubricated and decoupled
(essentially zero basal traction) and the necessary resistive forces
are supplied either by lateral shearing with cold-based ice at their
margins or longitudinal coupling up and down stream. In contrast,
application of our model, incorporating a higher-order rheology
with non-local basal decoupling and concomitant fast-flow,
precludes the necessity to precondition areas of the domain with
a high decollement parameter to a priori condition fast-flow to
geographically specific areas.
A formal boundary solution for determining the evolving
grounding line and ice-calving is also incorporated so that expan-
sion of marine sectors of the ice sheet to the shelf edge is also not
aana priori model outcome. This avoids the common model
assumption that all ice is lost at a predetermined depth (200 m in
the case of B&H), which not only neglects marine-shelf dynamics
but also preconditions the ice-sheet to automatically advance to the
shelf edge where the required depth is exceeded. A more realistic
glacio-meteorological solution to the surface mass-balance distri-
bution is also adopted based on a modified temperature-index
method exploiting a reference (1961–1990) climatology. This
addresses the mass-balance elevation shortcomings in B&H, which
fails to account for key feedbacks between ice-sheet evolution and
climate but equally importantly, also predefines the spatial pattern
of the modelled ice sheet’s divides and accumulation centres which
are fixed by the predetermined accumulation gradient. Given the
ability to realistically model mass balance using a temperature-
index based approach (using temperature, precipitation and
incoming radiation), a prescriptive mass-balance calculation seems
overly limiting. Finally, and aided by the availability of ever-faster
computing power, we endeavour to apply this model at finer grid
resolutions that enable individual outlet glaciers and ice streams to
be resolved yielding a more comprehensive comparison with
empirical evidence.
3. The model
3.1. Introduction
The 3D thermomechanical model applied is a first-order
implementation of the ice-flow equations based on the
approaches developed and applied by Pollard and DeConto (2007),
Marshall et al. (2005) and Hubbard (1999, 2000) and equates to
the L1L2 classification of higher-order model as defined by the
Hindmarsh (2004) schema. The model solves first-order terms
through a heuristic combination of grounded ice-sheet and ice
shelf equations which are solved iteratively to yield terms for the
vertically averaged longitudinal stress and basal traction from
which basal motion and internal deformation components are
derived. Longitudinal stresses become increasingly important in
high-resolution models encompassing complex relief and where
ice streaming and fast-flow are prevalent. The model is based on
that used to reconstruct the LGM ice sheet across Iceland
(Hubbard et al., 2006) but incorporating an addition boundary-
layer treatment for the grounding-line transition zone developed
by Schoof (2007) and implemented on the basis of Pollard and
DeConto (2007).
Isostatic loading is calculated using the commonly implemented
elastic lithosphere/relaxed asthenosphere (ELRA) scheme identi-
fied by Le Meur and Huybrechts (1996) as a reasonable approach in
the absence of a full spherical earth model. The coupled flow model
has been tested against a suite of higher-order intercomparison
experiments (Pattyn et al., 2008) as well as a number of indepen-
dent applications to alpine glaciers (Hubbard et al., 1998), to the
Scottish Younger Dryas ice cap (Hubbard, 1999; Golledge, 2007;
Golledge et al., 2008) and to the Icelandic LGM ice sheet (Hubbard
et al., 2006).
3.2. Boundary conditions
The input boundary distributions required are mean and
maximum basal topography, mean annual air temperature (MAAT)
and annual temperature range (derived from mean January/
February and mean July/August) along with the distribution of
annual precipitation. The model is applied to a nested hierarchy of
finite-difference grids from 10 to 2.5 km. The mean and maximum
topographies were derived from the NEXTMap digital elevation
model for Britain, 90 m Shuttle Radar Topographic Mission (SRTM)
data for Eire, Northern Ireland and the Isle of Man, and a mosaic of
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 761
high-resolution British Geological Survey bathymetric survey data
of the North Sea and Irish Sea Basins. The Atlantic abyss and west
coast of Eire were derived from the GEBCO 1-min data set (http:// These data sets
were interpolated onto a 250-m extended British Ordnance Survey
projection, melded and gridded to the required resolution (Fig. 2d).
Reference climatic distributions were derived from multiple-
regression analysis of monthly and annual temperature, and
precipitation distributions derived from the 1961–1990 data by the
United Kingdom Climatic Impacts Programme (UKCIP). Within the
limits of the model domain they were regressed against Northing,
Easting and elevation (yielding temperature and precipitation lapse
rates), and additional measures of continentality were derived for
precipitation under an index of cumulative elevation above sea-
level from the west coast and the annual temperature range. This
multiple-regression analysis yields R
values of 0.92 and 0.96 for
January and July temperature and 0.65 for precipitation (Fig. 2a–
2c); all values and coefficients are provided in Table 1. Though not
a perfect representation of the recent climate, these regressions do
capture the main features of the synoptic climate distribution, and
furthermore, enable the palaeo-climate to interact directly with the
evolving ice-sheet geometry.
3.3. Mass balance
Surface mass balance is determined by a positive degree-day
(PDD) scheme applied according to Laumann and Reeh (1993) and
relates total melt to cumulative positive temperature throughout
the year on monthly intervals. Monthly temperature is calculated
from the MAAT perturbed by a sinusoidal function whose
maximum and minimum amplitude is determined by mean
monthly July and January temperatures. Daily cumulative PDDs for
each month are calculated using a probability function based on
a relationship between the standard deviation of daily to mean
monthly temperature. Precipitation is assumed to be evenly
distributed throughout the year and accumulates when MAAT is
below a snowfall threshold of 1
C. Despite the implicit limitations
of such a non-deterministic scheme (see van der Veen, 2002 for
a thorough deconstruction of both degree-day and energy balance
models), their general ability to simulate glacier responses in
contemporary environments (e.g. Braithwaite, 1995; Jo
et al., 1995) provides some justification for its application to an era
when very little is known about the variables required for a more
sophisticated energy balance approach. The coupled model is
driven by perturbations in MAAT based on either a calibrated proxy
record for transient dynamic experiments or a single perturbation
for steady-state sensitivity experiments. Global eustatic sea-level
change is an additional forcing variable based on the SPECMAP
time-series for all experiments. Palaeo-climate forcing was imple-
mented through a 10-year running mean NGRIP
O record (North
Greenland Ice Core Project members, 2004; http://www.ncdc.noaa.
gov/paleo/icecore/greenland/ngrip/ngrip-data.html) scaled to the
prescribed palaeo-temperature forcing with an additional offset is
required. A bulk aridity function is introduced whereby the
precipitation distribution is linearly suppressed with the amplitude
of the imposed temperature perturbation. Hence, according to this
configuration, based on the NGRIP time-series, maximum cooling is
synonymous with the greatest aridity at 29.2 ka BP (Fig. 1).
3.4. Basal and calving dynamics
The model includes grounding-line dynamics and the formation
and flow of ice shelves when buoyancy is met. The processes and
physics at the grounding line have been the source of much debate
over the last decade (e.g. Hindmarsh and Le Meur, 2001; Vieli and
Payne, 2005; Pattyn et al., 2006) though a recent analytical
boundary-treatment by Schoof (2007) has provided a potential
solution for the marine-ice-sheet modelling community. Specifi-
cally, Eq. (16) of Schoof (2007) is used in our model to define the ice
flux at the grounding line as a function of ice thickness linearly
interpolated between the adjacent nodes that bracket floating and
grounded ice. Depending on whether the calculated analytical flux
from Eq. (16) is greater or lesser than that computed from the
heuristic flow equations, then the flux condition is imposed on
either the outer or inner node, enabling the grounding line to
fluctuate freely according to the constraint imposed by the Schoof
(2007, Eq. (16)) condition. Ice flow across the domain occurs
through the processes of internal deformation and basal motion
(Weertman, 1964) when basal temperatures attain pressure
melting point. Ice-calving into the ocean at the margin is calculated
using a standard empirical function relating calving-flux to ice
thickness, water depth and calving front geometry (Brown et al.,
1982). Geothermal heat flux varies little across the domain and is
kept constant at 55 mW m
(Rollin et al., 1993). Key rheological
relationships, variables and parameters are given in Table 2.
3.5. Implementation and limitations
Under the above boundary conditions, the model was integrated
forward in time with a time step of between 0.1 and 10 days from
an assumed ice-free and isostatically relaxed state. Despite the
additional rheological and climatic improvements in the model, it
must be noted that any model offers only a virtual representation of
the complexity inherent within the natural system. The process of
modelling involves pragmatic simplifying assumptions that are
common at some level to all models, despite their apparent
sophistication. Many of the primary thermodynamical processes
operating within ice sheets, especially those at its bed which
control its dynamic response are initiated on length scales
Table 1
Parameters, errors and RMS values from multiple-regression analysis of mean
January and July temperature and annual precipitation distributions across the
model domain. Northing and Easting regressed in British Ordinance Survey Units
(km) and elevation in m asl.
Parameter Value Error t-value
Independent: column (easting) /column (elev); dependent: July-temp
Y-intercept 16.38422 0.01577
Easting 0.00205 2.96886E5
Northing 0.00373 1.61369E5
Elev 0.00599 2.38319E5
(COD) Adj. R
Root-MSE (SD)
0.95776 0.95775 0.34911
Independent: column (easting) /column (elev); dependent: Jan-temp
Y-intercept 6.66361 0.01852
Easting 0.00423 3.48713E5
Northing 0.0025 1.89539E5
Elev 0.00616 2.79922E5
(COD) Adj. R
Root-MSE (SD)
0.92089 0.92085 0.43005
Independent: column (dtemp) /column (elev); dependent: precip
Y-intercept 2798.97474 45.59747
dtemp 143.17114 4.20148
Easting 0.84895 0.04273
Northing 0.03147 0.01625
Elev 1.23598 0.02208
(COD) Adj. R
Root-MSE (SD)
0.65459 0.65442 323.18397
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777762
significantly less than the grid resolution applied here. The general
principle is that such sub-grid processes can be aggregated in both
time and space so that they are adequately represented within the
bulk generalised description and operation of the model at the
resolution applied. Such an approach becomes problematic though
when the bulk sensitivity of system behaviour to sub-grid
processes, such as, grounding-line mechanics or weakening/failure
within the ice or deforming substrate is lacking adequate universal
physical description. The effect of such processes can be empirically
parameterised by, for example, preconditioning predefined zones
of the basal domain, but such an ad hoc approach does not neces-
sarily aid overall understanding of the system. In such a case, the
internal logic and consistency of the governing equations and
boundary conditions which lie at the heart of the approach are
compromised and as such, even though the overall fit of the model
with reality may be improved, it comes at a price – namely that the
overall integrity of the explanation offered is correspondingly
Whereas a large technical and computational investment is
made here to address deficiencies in rheology, basal processes and
Calendar Years
δ18O ‰
δ temp
87 6 5 4 3 2 1
Fig. 1. Time-series with Marine Isotope Stages (MIS) of NGRIP
O isotopic signature from 39.2 to 10 ka Calendar BP and the scaled 100-year running mean record used to perturb
the 1961–1990 reference climatology (data from:
Fig. 2. Model boundary distributions derived from multiple-regression analysis in Table 1. (a) Mean January temperature. (b) Mean July temperature. (c) Mean annual precipitation.
(d) Melded domain topography and bathymetry at 1-km resolution derived from NEXTMap (5 m), SRTM (90 m), BGS offshore (250 m) and GEBCO (10) data sets. (e) The BRITICE and
Olex on and offshore GIS used to provide direct empirical constraints on model validation, overlain with place names referred to in the text.
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 763
the spatial resolution of the ice-flow model, the coupling of the
model to a realistic atmosphere and ocean is limited. In effect, the
climate that the modelled ice sheet experiences at its upper
boundary condition is based on a set of statistical relationships
derived from present reference distributions of precipitation and
temperature and how these relate to accumulation and melt.
Though these do evolve to track the changing ice-sheet geometry,
they do so in a non-deterministic manner based on lapse rate/
elevation relationships which are static and flawed. Furthermore,
the imposed climate forcing itself is applied in a greatly simplified
manner; expressed as a relative perturbation in MAAT and
a percentage precipitation reduction that does little justice to how
synoptic climate change may express itself in terms of a locality’s
specific ‘weather’. However, these issues do attenuate and aggre-
gate themselves out, such that local micro-climate effects might
only be significant to seasonal accumulation/ablation at the local
scale, but less important on the macro-scale once an area has been
inundated by ice whose origin is many hundreds of kilometres
Coupled 3D Earth-system models become extraordinarily
complex as additional processes and parameterisations are incor-
porated. Much of the pragmatic utility of the modelling approach
adopted here is derived exactly from its relative simplicity; even
though there are over a dozen free parameters to tune, these only
express themselves in terms of two degrees of freedom. First, the
‘internal’ which encompasses the ice sheet’s bulk (but highly
heterogeneous) effective viscosity, whether it be determined by
differences in substrate, internal fabric, temperature, water supply,
impurity content and pressure and the resulting patchiness of basal
slip/stick. The second is more nebulous and comprises the mass-
balance condition at the surface, bed or margin. It is this coupling to
an atmosphere and/or ocean which makes this approach a signifi-
cantly more complex and challenging undertaking. A straight-
forward path is deliberately adopted for the surface mass-balance
condition and is justified on the basis that degree-day models and
their parameters are acceptable, even preferable, when or where
little is known of the multitude of climatic variables required to
drive a more complex model. Coupling to an intermediate General
Circulation Model (GCM) is an option but given that our ultimate
goal is to define broad linkages between the palaeo-climate (be it
reduced in complexity) and the geological legacy, then the intro-
duction of up to 200 additional variables into the bulk parameter
space with even a relatively simple (i.e. intermediate complexity)
GCM is unwarranted. Above all, we do not simulate the ice sheet per
se – the modelling approach is too crude, the processes are too
subtle and poorly understood for any such claim to be methodo-
logically justifiable. Our aim is to explore some of the processes and
dynamics of the BIIS and relate them to its bulk geometrical
evolution and the climatic change that drives it. We fully accept
that large-scale mismatches are inevitable and recognise that these
are intrinsic to the approach and part of advancing overall under-
standing of the system’s behaviour.
4. Experimental design and palaeo-environmental record
The suite of investigative numerical experiments must be set
within a palaeo-environmental framework that broadly complies
with numerous and in some cases disparate lines of geological and
geomorphological constraint. However, the initial- and end-point
members of the experiment need to be clearly defined and
4.1. Initial ice-free conditions
Computational limitations preclude numerical experiments
which cover the full 120 ka spanning the last Glacial cycle. Hence,
pragmatism dictates that the model should be initiated from as
near an ice-free condition as can be identified demarking some
convenient onset staging-post from which to launch a ‘Late
Devensian Glaciation’. Sejrup et al. (1994) identified a convenient
initiation point of rapid ice-growth marking the end of the
Ålesund Interstadial at ca 33,000
C BP, which they argue led to
coalescence of the British and Scandinavian ice sheets across the
North Sea. This conception resonates with recent work across
Scotland which indicates that build up of the last BIIS occurred
after a period of minimal glaciation when ice cover was confined
within Loch Lomond Stadial limits. This period, named the Tolsta
Interstadial (Whittington and Hall, 2002), spans from w38 to
31 ka BP at the type site of Tolsta Head, Lewis. Radiocarbon-dated
organic deposits indicative of cold, but ice-free, conditions found
at Balglass Burn (39.8–32.8 ka BP; Brown et al., 2006) and Sourlie
(39.4–35.2 ka BP; Bos et al., 2004) in the central lowlands of
Scotland lend support for an interstadial with much reduced ice
cover at this time.
4.2. Lateglacial conditions
Considerably more is known about the palaeo-climate and
environmental conditions during the Lateglacial period (ca 11.5–
15 ka BP) since it captures the last major glacial event that affected
Britain and Ireland. This period has been the subject of a number of
palaeo-climatic and ice-sheet modelling studies of increasing
resolution and sophistication (Payne and Sugden, 1990; Hubbard,
1999; Golledge et al., 2008). The Lateglacial period was marked by
climatic instability characterised by abrupt and intense climatic
Table 2
Parameters, constants and values used in the ice flow, mass balance and isostatic
components of the model.
Symbol Parameter Value Units
Density of ice 910 kg m
Density of seawater 1028 kg m
GGravity 9.81 m s
nGlen flow-law exponent 3
TTemperature – K
(pressure melt corrected) T8.7 10
aThermal parameter
T* <263.15 1.1410
T* 263.15 5.47 10
QCreep activation energy
T* <263.15 60 kJ mol
T* 263.15 139
RGas constant 8.314 J mol
Deformation enhancement 2.5
Thermal conductivity 2115.3þ7.93 (T273.15) J m
Specific heat capacity 3.1 10
Internal frictional heating J m
GGeothermal heat flux 55 mW m
mSliding-law exponent 1–3
Sliding-law coefficient 1.810
m kPa
Calving parameter 24.4 a
dtTime step 0.25–0.00325 a
Finite-difference interval 2.5 10
West–east array size 550
North–south array size 700
Daily air temperature sinusoid based on T
Snow–rain threshold 1.0
Rfr Refreezing ratio 0.07
PDD coefficient for ice 0.008 m
PDD coefficient for snow 0.003 m
Ice–ice free correction 1.0
DFlexural rigidity 5.0 10
Response time 3000 a
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777764
changes throughout northwestern Europe (Witte et al., 1998). Both
ice core data and palaeoecological proxies indicate a brief warm
period during the interval 15–14 ka BP at the very close of the last
full Stadial, possibly lasting no more than 500 years, before rapid
cooling ensued during the Older Dryas (ca 14.2–13.7 ka BP). This
cycle was temporarily interrupted by a climatic reversal from ca
13.7–13.1 ka BP, immediately prior to the deep and rapid thermal
decline marking the onset of the Younger Dryas Stadial. Cold
conditions persisted until 11.6 ka BP (Alley, 2000), after which an
extremely rapid climatic warming of 5–10
C took place in a matter
of decades (Dansgaard et al., 1989; Alley, 2000; Alley et al., 1993).
During the 1000-year return of full-glacial conditions in Scotland,
the climate experienced considerably greater seasonality than at
present (Denton et al., 2005), related to a reduction in the ther-
mohaline circulation that led to a southward migration of the
oceanic polar front to below 50
N(Ruddiman and McIntyre, 1981)
concomitant with the expansion of winter sea ice into lower lati-
tudes (Alley, 2000). During this period small corrie and mountain
glaciers formed as far south as Snowdonia, the Brecon Beacons and
SW Ireland (Sissons, 1979).
4.3. Modelling strategy and parameter space
Given the above palaeo-environmental constraints, a suite of
numerical experiments was initiated to explore the parameter
space of model uncertainties and sensitivities. The investigation
was designed to run from w38 ka BP, assuming ice-free conditions,
through to the Holocene at w10 ka BP using the North Greenland
oxygen isotope record (NGRIP) as a proxy climatic driver. Over 350
ensemble experiments were initiated in which key climatic, rheo-
logical and mechanical variables and parameterised relationships
were perturbed in a systematic manner to yield output that can be
compared against a set of observational criterion taken from the
available empirical record. The strategy is to define a subset of
experiments that passes empirical scrutiny and which provides
a basis for the geometrical reconstruction of the BIIS as well as our
conclusions regarding the associated palaeo-climatic conditions
and glaciological processes. The ensemble approach also permits
estimation of uncertainty in the reconstructions, that is the range of
BIIS volumes and palaeo-climatic conditions that yields acceptable
results. However, although the parameter space sampling strategy
is deliberately broad and systematic it is far from exhaustive.
Furthermore, and as discussed above, the model does not represent
all aspects of the system; there are significant simplifications,
missing processes and issues of scale which compromise the effi-
cacy of the approach. This said, the model is up to date with respect
to ice rheology and has been successfully tested against detailed
observations from numerous glaciers and larger ice masses along
with hypothetical benchmark and intercomparison experiments
and hence, we believe it is reasonably fit for purpose given present
state of knowledge and computational capability.
4.4. Experiment optimisation and selection
There exists a potentially vast collection of observable criteria
against which model efficacy can be judged. A real and significant
problem with a high-resolution modelling approach is that it offers
apparent detail at the local level which is unjustified given the
broad methodological assumptions implicit, as discussed above. For
this reason it is pragmatic to take a synoptic, regional scale
abstraction of the observable record, but still attempt to honour the
implications of the record contained at particularly significant
localities. The task here has benefited significantly from a number
of regional reconstructions including the BRITICE compilation
(Clark et al., 2004; Evans et al., 2005), and the interpretation of
high-resolution onshore and offshore digital terrain models (e.g.
Bradwell et al., 2008; Greenwood and Clark, 2008). A recent review
of the offshore evidence by Bradwell et al. (2008) indicates large,
dynamic, internally forced lobes which advanced to the shelf edge
along the North Atlantic margin and converged with the Fenno-
scandian ice sheet in the North Sea Basin. Greenwood and Clark
(2008) develop the ideas of Warren (1985) to find strong support
for a considerably more extensive and dynamic Irish Ice Sheet than
previously thought, with streaming flow on the continental shelf
off western Ireland. This reconstruction also supports the pio-
neering work of Scourse and Furze (2001) who proposed the
concept of a mobile and dynamic ice lobe extending across the
Celtic Sea as far south as the Isles of Scilly.
High-level trimline evidence from North West Scotland and
Ireland (Ballantyne et al., 1998, 2006, 2008; Stone and Ballantyne,
2006) also provide critical vertical constraints on ice-sheet geom-
etry, though their specific interpretation in terms of ice surface
altitude and thermodynamic environment is being re-assessed.
Recent cosmogenic isotope surface-exposure dating of mountain
top sites combined with the geomorphic evidence (Ballantyne and
Hall, 2008; Phillips et al., 2008; Stone and Ballantyne, 2006) indi-
cates prolonged, cold-based ice cover and provides a much-needed
insight into ice-sheet thermal regime and thickness. In terms of the
definition of broad experimental constraints, ice limits (vertical or
horizontal) are not used for de facto model optimisation and their
role in specific experiment validation is not crucial. Distal limits as
well as vertical limits are open to interpretation and at present are
still being evaluated and re-assessed. Furthermore, they may not
per se be a direct manifestation of the core processes driving the
wider ice-sheet configuration, thermodynamics and palaeo-
climate but are likely, at least in part, to reflect short-lived, rela-
tively stochastic excursions whose specific temporal and spatial
imprint is currently well beyond the scope of any deterministic
modelling approach. Furthermore, and on a practical level, ice
limits are not easily implemented as a direct boundary constraint
within a forward-marching time-dependent model and so for the
purposes of this study, modelled ice geometry is left as a free and
unconstrained output variable.
In addition to limits dictating the extent of the BIIS, there are
other lines of evidence which deserve at least equal ranking in
evaluating model efficacy. Lineations, ice-sheet bedforms, drumlin
flow-sets and erratic paths represent unambiguous, though
temporally incomplete, constraints on past ice flow and indirectly
the associated thermal regime and ice thickness. Here, we use the
ice-directional record from the BRITICE Geographical Information
System (Clark et al., 2004;
staff/clark_chris/britice.html), averaged onto a 5 km
domain as a direct constraint on modelled basal flow-regime under
the condition that basal flow vector orientation has to coincide for
at least 10 years and within 15
for a particular experiment to be
accepted. This approach is crude at best and does not do justice to
the detail in the BRITICE compilation nor, in effect to the sophisti-
cation within the model, but it does provide a primary and easily
implemented/automated criterion for experiment selection. The
relative sea level (RSL) record also provides a primaryconstraint on
palaeo-ice distribution. There has been significant empirical and
geophysical modelling research carried out to resolve the sea-level
record from isolation basins, raised beaches and other indicators of
RSL (Peltier,1998; Lambeck and Purcell, 2001; Shennan et al., 2005)
determined by the interplay between the eustatic and isostatic
history – a cumulative reflection of the temporal configuration of
ice loading. Along with the ice-directional record, honouring the
broad relative sea-level history at four type sites is a primary
constraint. Sufficient and persistent modelled ice is required to
provide the necessary isostatic depression to account for the main
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 765
characteristics of the raised beach/isolation basin record around the
5. Results
Of the 350 ensemble experiments conducted, a subset of w16
were identified as acceptable candidates for an ‘optimal’ solution as
defined by the flow direction and RSL criteria outlined in the
previous section. The scope of the present paper precludes a full
sensitivity analysis of the parameter space explored in the
ensemble experiments, however, a sample of the resulting volume
time-series (Fig. 3a) from the complete ensemble demonstrates the
potentially wide range of outcomes, from a massive, w2.5-km thick
BIIS that covers the entire landmass and adjacent shelf for virtually
the entire duration of the last glaciation, to a limited small-scale
glaciation episodically confined to mountain areas during the very
coldest phases. Such a spectrum encompasses all outcomes in
terms of potential ice-sheet geometrical configuration, thickness
and concomitant volume. Individual model trajectory was highly
sensitive to choice of boundary conditions and internal and
external parameterisation – particularly with respect to climatic,
geothermal and ice softness variables, and surprisingly less so with
respect to sliding and calving parameterisation. The scientific
problem is one that is under-determined; there is simply not
enough critical empirical evidence to definitively determine
a unique solution to the problem. Having said that, there is little
leeway in the resulting envelopes of sensible parameter/climate
forcing sets which yield a reconstruction compliant with the
established RSL and the flow-direction record. Hence, the majority
of the ensemble simulations are simply implausible on funda-
mental observational grounds and can be discounted on the basis of
the geomorphic and isostatic footprint. The experiments identified
as acceptable all share broadly similar chronological characteristics,
although the ice-sheet geometry, ice-stream dynamics and the
mass-balance regime can differ significantly on centennial to
millennial timescales. Two particularly noteworthy experiments
from the optimal subset which cannot be discounted on the basis of
their RSL and basal imprint are of interest since they yield
maximum ice conditions somewhat earlier than the majority of the
experiments: at w24 and 29 ka (Fig. 3a). However, for the purposes
of this paper we focus our attention on three specific experiments,
which should not be considered definitive reconstructions but do
Fig. 3. Time-series of: (a) scaled NGRIP climate forcing; (b) modelled ice volume from a representative sample of the entire suite of ensemble experiments; (b) modelled ice-sheet
volume and area for the subset of three optimal experiments; (c) net accumulation, ablation and iceberg discharge for the median optimal experiment E109b8, and (d) modelled ice-
sheet volume and area of the median optimal experiment with purge-phases which coincide with 50% or more of the ice bed at PMP overlain in grey.
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777766
appear to encompass the generalised dynamic response of the BIIS
to reasonable climatic forcing. They arguably demonstrate behav-
iour which provides an insight into the ice sheet’s sensitivity,
response and flow-regime, which is in tune with the emerging
paradigm of BIIS ice dynamics, extent, thickness and glacial history.
The experiments singled-out demonstrate the interplay
between temperature perturbation and precipitation suppression
that constitutes the imposed climatic forcing. The first experiment
(E102b2) was forced with
C and
P¼50%, the second
experiment (E109b8) with
C and
P¼75% and the
third (E109b2) with
C and
P¼70%. All of these
experiments include a permanent temperature offset of
C. The volume, area, net accumulation, net ablation and
calving-flux trajectory over the 28-ka simulation are shown in
Fig. 3c–e. Plan-form maps of (a) cumulative percentage ice cover,
(b) ice-sheet geometry and flow corresponding to the time slice of
maximal areal extent and (c) cumulative time that basal ice was at
pressure melting point (PMP) are shown in Fig. 4a-i to c-iii. High-
resolution .avi animations of these simulations at 50-year intervals
can be viewed at
model and the corresponding sequence of velocity maps at 30
key time slices over the duration of the simulation are provided for
the median experiment (E109b8) in Fig. 5, which forms the basis of
the following chronological synopsis which captures the general
form and chronology of all three experiments. In the following
section all dates are quoted in model years before present.
5.1. 38–32 ka
The main feature of the first w6 ka of the E109b8 experiment
are the three climatically induced oscillations that produce a small
land-based ice sheet, which undergoes relatively slow build up
then rapid retreat and thinning roughly in-phase with the
Greenland Stadial/Interstadial cycles that drive it. Glaciation during
these periods is limited to terrestrial areas of Mainland Scotland,
Orkney and Shetland, but a number of ephemeral, cold-based ice
caps develop across the Hebrides, the Southern Uplands and the
mountainous regions of Ireland, Wales and Northern England (the
‘Celtic Fringe’) during the very coldest stages. The three distinct
phases of activity peak at w35.3, 33.6 and 32.3 ka BP when Scottish
ice streams into the Moray Firth, the Minch and the Central
Lowlands. The intervening warm phases have restricted ice,
allowing deposition of interstadial sediments in the Scottish
Central Lowlands. Before complete deglaciation occurs, a return to
cold conditions marking the onset of the next stadial initiates
a renewed period of positive mass balance and subsequent ice-
sheet growth. The ice sheet is never in equilibrium with the climate
and the zenith and nadir of each cycle lags behind that of the
climate forcing by between 200 and 500 years. Although accumu-
lation rates scale well to the magnitude of forcing during each cycle,
transient ice-growth during these phases is controlled by the short,
w1.5 ka duration of each stadial.
5.2. 32–27.5 ka
After 32 ka BP, stadial conditions persist for a prolonged w3ka
phase to yield a significantly larger, more extensive ice-sheet over
Scotland, which by 30.35 ka has advanced down major valley
systems into offshore shallows to coalesce with the independent
ice centres over Orkney, Harris, and Mull, and to converge with ice
from the Southern Uplands across the central lowland belt. The
Moray Firth and other large outlets experience streaming behav-
iour at this time. The Shetland ice cap advances into shallow
offshore waters as do ice caps over Connemara, Erris and SW
Ireland which are drained by fast-flowing, actively calving outlet
glaciers. Ice caps across the uplands of Northern Ireland, the
northern Pennines and North and Mid-Wales also converge. The
climatic nadir at 30 ka imposed by the NGRIP record yields a cold-
based and frigid phase of ice-growth across the entire domain, the
eventual convergence of Lake District Ice with Pennine Ice across
the Howgills and the linking of ice cover over the Brecon Beacons
with that from the central Cambrian Mountains. At this time, small,
cold ice caps build up in peripheral uplands at Dartmoor, Exmoor
and the North Yorkshire Moors. This frigid period is a precursor to
a 200-year phase of rapid, widespread areal expansion through ice
streaming and outlet glacier advance into adjacent shelf areas of
the Atlantic Margin and North Sea Basin from Ireland and Scotland
from 29.3 ka. During this period, outlets from the Welsh Ice Cap
stream down the Dyfi, Mawddach and Dwyryd estuaries into
Cardigan Bay and ice advances into Conway Bay. Further rapid
expansion occurs by 29.1 ka with the development of ice streams
on the deformable sediments across much of the adjacent shelf. An
ice stream from the Firth of Forth coalesces with ice from the Tweed
and Tay, spilling into the North Sea Basin and advancing south as far
as the Tees estuary. The Moray Firth Ice Stream continues to surge
episodically and is fed by streaming ice from the Spey. Ice from
Western Scotland, County Antrim and the Solway Firth pours into
the Irish Sea Basin, feeding, at this stage, a fast-flowing outlet
glacier which almost reaches the Isle of Man. The Outer Hebrides
appear to have been largely overwhelmed at this stage and ice
centres across Ireland converge to completely inundate the central
Irish lowlands. With the arrival of Greenland Interstadial (GI) 4, ice
thins and retreats rapidly to a core area centred on Western Scot-
land when another cold cycle leads to a less extensive advance and
subsequent retreat culminating in a small ice sheet centred over
Scotland by the end of GI-3 at 27.5 ka.
5.3. 27.5–19.4 ka
A sustained cold climate from 27.2 ka onwards provides the
consistent conditions required for the development of a large ice-
mass which ultimately advances across much of the domain. For
the initial 3 ka, ice development is similar in nature to that of the
previous stadial build up between GI-5 to 4, with cold-based,
independent ice centres gradually expanding and advancing
offshore and southwards, interspersed with episodic ice stream
switching across the west coast of Scotland and into the Irish Sea
Basin. At 23.7 ka the Scottish Ice Sheet eventually converges with
Shetland ice, and from this time onwards a relatively temperate but
still cool climatic regime triggers a phase of intense dynamical ice-
stream activity across all sectors, which unlike previous cycles does
not lead to the massive and widespread draw-down and concom-
itant retreat. Wet-based-driven fast-flow is widespread across
virtually all lowland and marine sectors into the North Sea and
Atlantic with extremely rapid switching between ice streams that
compete and penetrate deep into core zones of the ice sheet for
mass. Throughout this period an extended Irish Sea Ice Stream
(ISIS) with its onset zone in the central Scottish lowlands advances
rapidly down St George’s Channel to converge with episodically
surging outlets from the Welsh Ice Cap across the Lleyn Peninsula
and from the entire eastern sector of Ireland. At 22.65 ka sustained
warm conditions lead to a local retreat of Welsh and southern Irish
ice but the ISIS, with its large catchment area, continues its advance
southwards. Another phase of warming at 20.8 ka prompts
a widespread mobilisation of wet-bedded ice across extensive
lowland and shallow marine sectors yielding maximal ice cover
across the domain at 20.1–19.9 ka. Large, mobile, surge lobes
dominate the southern sectors of the Irish Sea and North Sea, and
the Atlantic coast and extend to the north coast of SW England,
Norfolk and to the continental shelf edge, respectively. However,
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 767
Fig. 4. Plan-form distributions of: (a) cumulative ice cover as a percentage of the total simulation time. The maximum ice sheet (non-synchronous) footprint corresponds by white
shading (% ice cover >0%). (b) modelled ice-sheet surface geometry and associated flowlines for the time-slice correspondingto the maximum areal extent. (c) cumulative time that
the ice-sheet bed was at PMP expressed as a percentage of the total simulation time. Persistent frozen basal conditions given by black shading (%PMP<2.5%). The figures
correspond to experiments: (i) E102b2, (ii) E109b8 and, (iii) E109b3.
Fig. 4. (continued).
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 769
Fig. 5. Key time-slices from between 35.95 and 11.65ka BP for the optimal experiment E109b8.
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777770
the less vigorous margins of the ice sheet are beginning to retreat at
this stage, and the Scottish-based catchment of the ISIS is diverted
into the Atlantic by the southerly migration of the former Scottish-
Irish ice divide towards the Isle of Man. For the next 700 years, ice
margins fed by intensely switching, dynamic ice streams are rela-
tively stable but tend to retreat slightly in the southwest and
Atlantic sector and advance in the eastern sector into the North Sea
Basin, concurrent with a general shifting of divide structure and the
ice sheet’s ‘centre of the gravity’ eastwards. A broad east-west
divide develops from the Cairngorms to Aberdeen and across the
North Sea Basin and is intersected by another broad divide running
roughly south from Shetland.
5.4. 19.4–17.4 ka
For 2 ka from 19.4 ka onwards, all sectors of the ice sheet
undergo active, dynamic and extremely rapid thinning and retreat
in response to sustained warmer climatic conditions, which the
surface mass and flux regime of the ice sheet cannot sustain. This
rapid retreat phase is accompanied by constant and highly dynamic
ice-stream activity, which includes constant switching and cross-
competition across core central areas and results in short-lived
standstills or occasional minor surge-initiated forays that lead to
limited local and short-lived advances. However, the overall picture
is one of large-scale rapid reorganisation, if not complete collapse
across all shallow marine sectors of the ice sheet to a compact,
mainland-based central core zone which covers much of Scotland
and impinges on NE Ireland. The Shetland and Orkney ice caps
actively persist after separation, along with a number of isolated ice
fields left stranded across upland areas of Ireland, Wales and
England. Many of these stagnate and perish but a handful, espe-
cially those which separate from the retreating ice sheet later on,
survive through to the advent of cooler conditions at 17.4 ka.
5.5. 17.2–13.8
A period of relative cooling offsets retreats and stabilises a much
reduced and thinner ice sheet which extends over much of Scotland
across to the northern Pennines. This climatic trigger gives the
necessary stimulus for the ice sheet to stream across western and
eastern marine sectors. Significant streaming activity from the
Tweed, Forth, Tayand Moray Firth results in substantial surges onto
the shallow shelf of the North Sea Basin, whilst advances down the
Solway, St George’s Channel and across the west coast are less
pronounced. By 15.7 ka continued warming further reduces cover
to an ice cap limited to the Western Highlands and Cairngorms with
small ice centres persisting in the Southern Uplands and northern
Pennines, along with ice in Shetland and Orkney. Isolated ice fields
and small ice caps persist in Snowdonia, Wicklow and Connemara
but after a final limited phase of surging at 15.6 ka, ice cover almost
completely disappears from the domain. However, some small
areas of activity persist in the Western Highlands and Spey catch-
ments throughout Greenland Interstadial 1.
5.6. 13.8–10.4 ka
The final event, the Younger Dryas, is launched from a limited
but persistent, wet-based core centred on Loch Linnhe and Ben
Nevis which freezes and expands in response to the onset of rapid
and severe glacial conditions across much of the Western High-
lands at 12.8 ka. For 1.2 ka ice builds up over Western Highland
areas and to a limited extent over the Cairngorm Plateau with
smaller pockets of ‘glaciation’ experienced across the mountainous
of regions of NW Scotland and England, Wales and Ireland,
particularly across Snowdon, mid-Wales, Cumbria, Skye, Mull, the
Outer Hebrides, Wicklow, Connemara and the Pennines. These
satellite ‘ice fields’ vary in size ranging from small-scale corrie
glaciers associated with the w200 years of coldest temperatures, to
more active ice caps with limited outlet glaciers in NW Ireland and
western Scotland. The mainly cold-based Scottish ice cap persists
throughout the stadial, but experiences a late phase of wet-based
surge activity at the onset of warming at 11.6 ka before undergoing
rapid but active retreat. Modelled ice cover persists into the early
Holocene in a handful of key areas, mainly around Ben Nevis, but
the complete demise of the last remnant of British glacier ice is
achieved by w10.5 ka, by which time the Irish Sea Basin is
completely inundated by rising sea-levels.
6. Discussion
The subset of model experiments which conform to the RSL and
ice-directional records, used as a primary constraint, all share
several key characteristics of the reconstructed BIIS complex. These
can be used as a framework for further discussion:
(1) The BIIS undergoes rapid fluctuations in size at centennial to
millennial timescales,
(2) The BIIS experiences repeated but episodic and highly transient
zones of fast-flowing warm-based ice,
(3) The BIIS maintains persistent slow-flow conditions in key
areas, which remain under cold-based ice,
(4) The BIIS develops multiple ice centres/domes and divides
which link them and migrate freely, and
(5) The BIIS experiences significant and major changes in ice flow
and basal orientation.
6.1. Ice-sheet fluctuations and dynamics
The dynamism of the simulated ice sheet is one of the striking
features of the experiments. Fluctuations both in terms of spatial
limits and flow units over this magnitude and time-scale are not
apparent in previous models of the last BIIS and significantly extend
the interpretation implicit to B&H into the chronological domain.
The dynamism, rapid and abrupt shifts, and reorganisations of the
ice-sheet margin and divide structure become transiently decou-
pled from the climatic forcing that drives it. These changes are well
illustrated during the numerous ‘binge-purge’ cycles that repeat-
edly dominate the simulation. The intense, cold-based ‘binge phase’
enables a core, high viscosity ice-mass to build up in thickness and
volume and advance primarily by viscous creep, and subsequently,
by small fast-flow units when the pressure melting point is attained
over selected zones of the bed. However, at the apex of the cold
phases the ice sheet is limited in extent and has a high aspect ratio
corresponding to thick, cold ice and large volume. This is an
essential precursor to the subsequent purge-advance phase, in that
it provides the required mass and thermal inertia to enable wide-
spread and extensive streaming and thinning to take place. It is not
the classic ‘binge-purge’ free-oscillator mechanism described by
MacAyeal (1993) since there is a constant, low-level ice-stream
activity that deters any tendency for widespread internally driven
thermomechanical meltdown. Hence, the ice sheet self-regulates
through ice streaming but by doing so, it spreads and thins, thereby
becoming more vulnerable to external climatic amelioration. When
such an external stimulus does occur, the warmer ice associated
with it advects through the ice-mass, induces basal melting and
mobilisation, which is rapidly propagated through longitudinal
coupling to yield widespread fast-flow and a binge-advance phase.
This is inevitably followed by a phase of equally rapid purge-retreat
and reorganisation as the ice sheet’s net mass-balance regime
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 771
becomes unsustainable. There are a number of critical climatic,
glaciological and geomorphological implications to this modified
binge-purge/advance–retreat cycle.
A major consequence is that periods of maximum areal extent
are not coincident with maximum cold, stadial conditions, though
these conditions are a necessary precursor. Instead, phases of
maximal streaming and areal extent, if not bulk volume, will be
phase-locked to the transition stages into warmer conditions. Given
a fairly short thermal response time of a maritime BIIS (<1 ka) in
which warm surface ice can be advected through to the bed rela-
tively rapidly, then phases of maximal ice-sheet extent will often
coincide with interstadial warm climate conditions. Furthermore,
the relatively short-lived warm-based, ice streaming purge-retreat
phase will be greatly over-represented in the geomorphic record
thereby dominating the long-term landscape legacy. This is espe-
cially true during purge-retreat phases which characterise degla-
ciation when melt-water is abundant within the basal and
proglacial landscape.
The periodic purge-advance phases are also associated with
massive iceberg flushing events (Fig. 3d) as streaming ice advances
into deepening water on the continental shelf and delivers large
fluxes of ice-rafted debris (IRD) to the seafloor. These calving-flux
events are apparently non-linear, as their magnitude does not
necessarily directly reflect the bulk volume, extent or indeed
dynamical vigour of the ice sheet associated with each event. For
example the calving event at 23.7 ka of >1000 km
is dispro-
portionately large compared to later events at 21 and 20 ka when
the ice sheet was significantly larger and more dynamic (Fig. 3d).
Although a somewhat perplexing result at first, it follows that the
scale of iceberg flushing events is highly sensitive to the overall
system state and stability, which governs the precise ice sheet, sea
level and climatic configuration at that time. Such a finding has
important implications for the wider interpretation of specific IRD
layers within the context regional offshore record which warrants
further investigation.
On the millennial scale, ice-sheet dynamics and associated
maximum limits can become completely decoupled from the
climate that drives them in the first instance. Hence, the maximal
extent of the BIIS – during a purge-advance phase of pervasive
fast-flow and streaming at 19–21 ka – coincides with a period of
relatively benign climate in the NGRIP record. It is, however, the
long-lived cold phases (i.e. 27–24 ka) that provided the critical
mass and thermal inertia for the necessary ice reservoir to accu-
mulate in the first instance. After 24 ka the BIIS is sufficiently
large with enough mass and thermal inertia to significantly
modify its own climate through temperature-elevation feedbacks.
Hence it persists and continues to grow; albeit in a controlled and
limited streaming-phase under a moderate climate. At the same
time, localised ice caps in marginal locations responding to this
forcing experience a negative mass-balance and decline, whilst
the ISIS and other major ice streaming outlets continue to
advance. The corollary is also true, with minor, short-lived and
internally or externally induced advances occurring at any time
and at any locality during the subsequent deglaciation sequence.
The threshold associated with the absolute ‘tipping-point’ of the
final, large-scale deglaciation from 19 ka onwards is not a wholly
internal nor externally forced response; neither is it stochastic. By
this stage, the over-extended, low-aspect ratio ice sheet has made
itself unsustainable in terms of overall mass and flux budget
under sustained warm conditions and has become so large that
reorganisation is inevitable. As a result, the BIIS goes into rapid
freefall – losing almost 600,000 Gtonnes of mass (up to w2 m sea-
level equivalent) in under 2000 years (Fig. 3e). This generates
vast amounts of melt-water runoff, which peaks at over 500
Gtonnes a
at w19 ka BP.
The mass balance–elevation relationships across the large
retreating Irish Sea and North Sea lobes are very similar to those
observed across the western margin of the Greenland ice sheet
(GIS) at the Arctic Circle over the last decade (van de Wal et al.,
2005) ranging from 6ma
at sea-level to þ0.5 m a
at 1500 m
asl. Given that the long surface profiles and gradients across the
larger BIIS outlet lobes are also similar to those across the southern
GIS margin, then it is likely that large supra-glacial lakes will have
repeatedly formed and drained in a similarly catastrophic manner
to those observed in Greenland by Das et al. (2008). These events,
which can occur on an annual time-scale across the entire southern
GIS margin (Box and Ski, 2007), rapidly decant large (e.g.
w19 Mtonnes in 2 h) volumes of surface-derived melt-water into
the basal hydrological system through a process of fracture prop-
agation and frictional conduit widening through up to 1000 m of
ice. The impact of such basal hydrological tapping events is
profound, at least in the immediate locality of the injection point,
and may well provide a valid glacio-hydrological explanation for
the formation of the large, incised tunnel valley systems, up to
350 m deep and 39-km long, that have been cut into sediments
perpendicular to reconstructed palaeo-ice-sheet margins within
the North Sea and Irish Sea Basins (Wingfield, 1990; Lonergan et al.,
2006; Bradwell et al., 2008; Kristensen et al., 2008). The macro-
scale dynamical implications of such events is yet to be determined
but recent InSAR results from the western margin of the GIS indi-
cate seasonal surface accelerations of up to 200% (Joughin et al.,
2008). Though these direct surface-to-basal dynamical processes
are not incorporated in the current model of the BIIS, it is a poten-
tially useful test-bed. Such enhanced dynamical response could
only have hastened the retreat and rapid demise of the BIIS at this
6.2. Periodic development of fast-flow
The former existence of fast-flowing ice streams in the BIIS has
been firmly established, for example, in the Minch (Stoker and
Bradwell, 2005; Bradwell et al., 2007), in Strathmore (Golledge and
Stoker, 2006), and the Tweed catchments (Everest et al., 2005),
along with the identification and limits of the ISIS (Scourse and
Furze, 2001). The ‘activity index’ of the ice streams – their switching
in both space and time – in the three experiments is unprecedented
and potentially open to question. However, recent work by Hulbe
and Fahnestock (2007) reconstructing Siple Coast/Ross Sea ice-
stream dynamics indicates not only century-scale stagnation and
reactivation cycles but also lateral communicationwith adjacent ice
streams transmitted through geometrical and hydrological
changes, observations that are consistent with the simulations of
the BIIS presented here. We therefore propose that the switching
and dynamism seen in the BIIS model reconstructions is real.
Fig. 7a–o reveal a series of 50-year time-slices corresponding to
22.6–21.9 ka BP which demonstrate a spatial and temporal hier-
archy of ice stream behaviour that is fractal in nature but very much
in tune with the observation and modelling carried out by Hulbe
and Fahnestock (2007) across the Ross Ice Shelf. The major ice
streams, such as those which occupy the Irish Sea Basin and Sea of
the Hebrides shelf are persistent features, though their individual
tributaries, onset zones and their sphere of influence over the
entire catchment fluctuate significantly – constantly competing
with each other over the 700-year period. Other less persistent but
geographically determined ice streams, such as in the Minch, also
demonstrate remarkably dynamic behaviour. The Minch ice stream
initiates at 22.45 ka with two onset zones located off the present
day coastline around the Summer Isles and another further south
between Raasay and Skye. The ice stream develops rapidly with its
main lobe surging to the continental shelf edge over the next
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777772
100 years whilst also expanding its catchment area inland as the ice
sump is drawn-down. At 22.3 ka piracy occurs and its southern
tributary is captured by an ice stream initiating due west across
Lewis, however, over the next 200 years the main trunk of the
Minch ice stream thickens and expands headward, sourcing ice
from as far east as the Moray Firth for a limited period. The main ice
divide and dome centred over the Cairngorms at this time is in
a state of perpetual warping and recovery as it is systematically
drawn-down by vigorously competing ice streams from several
sectors. Even more temporally ephemeral ice streams develop in
the Moray Firth and extend NE out into the North Sea Basin and
also, for short periods, deflect NW across Buchan from the Moray
and Tweed catchments. The surge lobe which develops along the
southeast Scottish coast to Northumberland, and extends into
Yorkshire almost as far as Flamborough Head, is fed by a combina-
tion of transient ice streams and tributaries originating from the
Tay, Forth, Tweed and even the Dee-Don catchments, all of which
are switching on centennial timescales. At 21.9 ka, marginal ice
stream activity abruptly shuts down due to a combination of cold
ice being rapidly advected into the basal-system and overall surface
6.3. Maintenance of slow-flow conditions
Despite the purge-advance and retreat cycles described above,
which will dominate the geomorphological record, the majority of
the glacial simulation is characterised by cold-based, thick ice
associated with the binge-cycle that has a minimal geomorphic/
geological impact other than that it is a fundamental prerequisite
in order to derive the long-term isostatic and RSL footprint. A
permanent thin, low-aspect ice sheet which is subject to constant
ice streaming and fast-flow does not have sufficient volume and
mass to isostatically depress the lithosphere to the extent implied
by the RSL record. The former existence of thick, cold ice during
the last glaciation has long been suspected over the inherited,
persistent and largely non-glacial landscapes found across upland
Britain and Ireland (e.g. Sugden, 1968). But only recently has this
been confirmed by cosmogenic exposure-age analysis in northern
Scotland (Phillips et al., 2006; Phillips et al., 2008) and southern
and western Ireland (Ballantyne et al., 2006; 2008). Our model
simulations concur with this interpretation. Ballantyne and Hall
(2008) have recently concluded that high-level trimlines, formerly
classified as demarking the maximum ice-sheet height across the
Northern Highlands, represent an upper limit of glacial erosion
probably corresponding with an englacial thermal boundary. Such
a re-interpretation would lend further support to the model
experiments presented here. Modelled cumulative ice cover of less
than 2.5%, at pressure melting point, should provide a reasonable
indication of the relict, upland landscapes across Britain and
Ireland which were protected by cold-based ice (Fig. 4c-i–c-iii),
within the grid resolution and thermomechanical limitations of
the model. Cursory comparison with the literature (cited above)
suggests that these persistently cold-based zones are fairly
well predicted by the model, though a more rigorous analysis is
called for.
6.4. Ice-flow directions and ice limits
All three simulations honour the ice-directional and erratic
train record contained within the BRITICE reconstruction. Main
flow line travel paths west out of the Western Highland fjo
systems, from the Southern Highlands into East Lothian and Fife,
NE across the Aberdeenshire coastline and from Ailsa Craig west
into Co. Antrim and south into Galloway are all successfully
reproduced at some stage in the simulations, if not at the LGM
(i.e. the time-slices of maximum areal extent (Fig. 4b-i–b-iii)). As
discussed, there is significant and rapid flow-switching occurring
throughout the course of the simulation which is accompanied by
ice divide and ice centre migration and complex ice-directional
switching. Fig. 6 gives the rose diagrams associated with cumula-
tive basal ice-flow orientation over the entire E109b8 experiment
for selected localities across the domain. These records should
correspond with the palaeo-flow record observed from ice contact
features and lineations documented in BRITICE and other workers
such as the Solway Firth (Golledge and Merritt, 2005), the Minch
(Bradwell et al., 2007), and the Tweed (Everest et al., 2005).
However, some key localities within the model domain indicate
a complexity that may be real but are not fully appreciated or
preserved in the empirical record due to subsequent erosion. The
Aberdeen area has a dominant flow axis trending WNW to ESE but
also records flow directions to the NE and NW, the latter possible
corresponding with a known phase of onshore carry of erratics
from the western North Sea (Merritt et al., 2003). The record from
the Isle of Man is extremely complex and seems to signify its
central location as a major interactive junction for ice from
a variety of directions. The mouth of the Dyfi indicates the main
direction of ice flow perpendicular to the coast associated with the
ISIS but also a local outlet glacier flowing SW into Cardigan Bay.
Similarly, the records from the Lleyn, Orkney, Shetland, Spey,
Tweed and Forth all indicate a strong interplay between local,
hinterland-derived ice flow interacting with or against the larger
regional-streaming flow. We propose that the Sarns which extend
up to 25 km into Cardigan Bay are likely associated with the final
warm-based, purge-advance phases of the Welsh Ice Cap. The Irish
Sea Basin record indicates strong bifurcation of ice which appears
to originate either from the Irish Sea Basin to the NNE, or from
more locally derived ice source in Southern Eire to the NNW. It is
this latter component of flow, indicative of an Irish source area that
may hold the key to resolving issues of the southern limit of the
Irish Sea Ice Stream.
The limits of the last ice sheet in England and Wales are long
established and the modelled limits appear to match these fairly
well though there are some significant and important problems
and discrepancies. The limits of the last BIIS are now recognised to
lie along the shelf edge from west of Shetland to the Barra fan NW
of Donegal (Bradwell et al., 2008). Even without the bolstering or
diverting effect of Scandinavian ice, the model reproduces all of
these well. Non-synchronous maximum ice extent (Fig. 4a-i–a-iii)
indicates that ice extended offshore to the Atlantic shelf edge across
all western sectors including Ireland. It is also likely that Ireland
experienced complete ice inundation over its cumulative glacial
history even if this did not occur synchronously during a particular
event. There exists a time-transgressive element to maximum
extent which is apparent in the dynamics of the ISIS which reached
its maximum extent at w21 ka, and the North Sea lobe which
attained it maximum southerly position at w19.3 ka. Maximum
limits do not agree particularly well with the conventional limits,
but many of these have been recently revised or are in the process
of being significantly modified. A case in point is the work of
Greenwood and Clark (2008) who have reconstructed the chro-
nology and limits of the Irish Ice Sheet and argue strongly for 100%
inundation of the terrestrial landmass with western limits placed
well offshore towards the continental shelf edge. Interestingly,
a case is also made for a southerly migration of the Scottish–Irish
ice divide in the later stages of her reconstruction – and this is a key
feature of the three experiments presented, and which in turn leads
to a curtailment of the ISIS activity southwards.
In two of the three experiments presented, the modelled ISIS
does not advance across the Isles of Scilly, despite the fact that all
three indicate ice advancing across Pembrokeshire onto the north
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 773
Devon coast. Here, the model is not consistent with the recent
evidence. We would not at all wish to claim ascendancy or to
question the important work on the southern limits of the ISIS and,
in particular, the glaciation of the Isles of Scilly (Scourse and Furze,
2001). However, despite all the shortcomings and limitations of the
modelling approach utilised in this study, it does have a consistent
internal logic based on contemporary glaciological principles that
seems to suggest that the Irish Sea Basin story is not yet resolved.
Without preconditioning certain large but critically limited zones
of the Irish Basin to a priori streaming, it is difficult to achieve
a single ‘surge-advance’ south to the Scilly Isles without a broad
piedmont type-lobe impinging onshore across much of SW
England. The reconstruction inferred in the E109b2 experiment
with a high precipitation scenario across western Britain may be
exaggerating the case somewhat but this simulation does tanta-
lisingly still meet all of the available ice-directional and RSL
constraints. Further enhancing precipitation rates across SW Eire
up to and exceeding present day values would yield a considerably
more western dominated Irish Sea ice-mass that would bring
modelled ice limits, especially those associated with the ISIS in SW
England, into line. However, there is little palaeo-climatic evidence
nor GCM modelling to support a wetter LGM than present across
southern Eire. Within the limitations of this study, we do not
pursue this though further investigation is clearly required.
The experiments presented also indicate significant excursions
of wet-based ice into areas of southern England, where little
evidence of recent glaciation has been found, This may not present
such a major problem given that ourmodel indicates ice was at this
extended limit for less than 1 ka. The experiments also provide
support for a possible glacial mechanism for the movement of
Preseli erratics as a transport trajectory which overrides parts of
northern Pembrokeshire and was subsequently deflected south-
eastwards across the Bristol Channel into SW England, cannot be
completely discounted.
6.5. Development of multiple ice centres/domes
Evidence for discrete ice centres in Scotland on Shetland, the
Cairngorms, Mull, Skye, the Southern Uplands and the Outer Heb-
rides were identified and reviewed by Sissons (1967)and Sutherland
(1984). Furthermore, significant ice accumulation centres in the
English Lake District, North, Mid and South Wales, and SW Ireland
can be seen from our modelling experiments. We also identify
a Cairngorm ice centre which plays a significant part in ice-sheet
build up, as originally stated by Sutherland (1984); and a further ice
centre based on Orkney, not previously identified. The Outer Heb-
rides appear to lack an independent ice dome over the southern isles
but a cold ice cap that builds on Harris deflects the Minch Ice Stream
northwards and around Tolsta Head. A fast-flowing calving outlet
glacier derived from an ice cap over the hills of NW Ireland appears
to be a persistent feature of the Lateglacial common to all three
optimal experiments; though there appears tobe little observational
Shetland WEire Dyfi Eden Spey
NIreland Solway Orkney Tiree Aberdeen
Forth StKilda IofMan SEire IrishSea
Moray LleynTweedCaithness
Fig. 6. Rose diagrams of total cumulative basal velocity orientation signature over the entire simulation at selected localities for the experiment E109b8.
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777774
evidence to support this. Small, cold-based ice caps and ice fields
develop at marginal locations during the coldest stages of the
simulation, the most controversial of which are found over Dart-
moor, Exmoor and the North Yorkshire Moors. Observational
evidence for cold-based ice cover across these marginalupland areas
is currently lacking but could provide ideal validation targets.
7. Conclusions
In this paper we have applied a climatically coupled higher-order
ice-sheet model to the British Isles down to 2.5-km resolution. By
incorporating longitudinal stresses and newgrounding-line physics,
our model is able to capture the dynamics of the former British–Irish
Ice Sheet palaeo-ice sheet and provide additional insight and chro-
nological context to the work of Boulton and Hagdorn (2006).
Specifically, the new modelling simulations presented here lend
strong support to the dynamical, fast-flow paradigm that has
emerged in the last decade. However,this needs to be qualified since
we also find that these fast-flow streaming phases are part of a larger
binge–purge cycle. In essence, this cycle is moderatedby long phases
of frigid, cold-based ice build up (‘binges’) that provide the ice sheet
with its mass, thermal inertia and the required isostatic signature
from which to launch streaming ‘purge’ phases. The trigger for such
events appears to be a interplay between internal thermodynamic
coupling and external climatic forcing that occur during rapid
transition stages from cold to warm conditions, and is accompanied
by extensive and dynamic ice streaming and interior draw-down.
Longitudinal coupling plays a significant role in the mobilisation of
extensive marine sectors of the ice sheet and these purge-advance
phases are accompanied by large-scale calving events which release
significant iceberg and freshwater runoff flux into the N Atlantic. The
NGRIP isotope record, used to force the model, contains numerous
stadial/interstadial cycles and the BIIS modelled in this study
becomes phase-locked to these cycles, resonating and amplifying
them. The wider implications for the offshore sediment record and
an understanding of the broad forcing/response couplings between
the global ocean – atmosphere – cryosphere systems are potentially
profound and undoubtedly complex. Equally so, the onshore
geomorphological imprint of such a dynamically fluctuating BIIS is
correspondingly complex, but it is hoped that this work and further
refined BIIS experiments may provide an initial synoptic framework
from which individual sites can be assessed and interpreted. In this
way, we hope that our model can be critiqued and improved in the
Specifically, with particular reference to our optimal subset of
numerical experiments we find support for:
(1) Intense phases of cold-based and warm-based activity, in
which ice streams play a defining role and draw-down large ice
fluxes from the ice sheet interior.
Fig. 7. Time-slices at 50-year intervals of dynamic ice-stream activity for 700 years from 22.6 ka BP.
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 775
(2) Intense persistent and ephemeral ice stream and outlet glacier
activity, which displays switching and dynamical fluctuations
on a century or less time-scale.
(3) The existence of a dynamic and extensive Irish Sea Ice Stream
that can advance onto the Scilly Isles but only in conjunction
with ice impinging on the SW coast of England.
(4) Complete glaciation of Ireland extending offshore across the
west coast. This may be time-transgressive and is not neces-
sarily coincident with the Last Glacial Maximum.
(5) Extensive glaciation across the North Sea Basin extending to
the Atlantic shelf edge and encompassing an independent
Shetland ice cap.
(6) Deglaciation in under 2 ka which decanted ca þ2 m of global
sea-level equivalent, under conditions analogous to those
across the SW Greenland Ice Sheet margin today. This leads to
a hypothesis that the North Sea and Irish Sea tunnel valley
systems may have a supra-glacial melt-lake drainage origin.
(7) Significant widespread ‘inherited’ upland landscapes protected
under persistent cold-based ice cover. We also find strong
support for the re-interpretation that high-level trimlines
mapped across NW Scotland actually represent an upper limit
of glacial erosion, not glaciation.
(8) Independent, persistently cold-based ice caps over Scottish,
Irish, Welsh and English upland areas with warm, surging
outlet glaciers and lobes discharging on all fronts. More
controversially, other cold-based ice caps existed over Dart-
moor, Exmoor and the North Yorkshire Moors.
ALH acknowledges the generous support from a BGS–BUFI grant
[ESB49900096], a research fellowship provided by the University of
Stockholm, SUCLIM initiative, a Royal Society Short Visit grant from
which parts of this study were supported and also generous
support for fieldwork from the Carnegie Trust for the Universities of
Scotland. ALH is also grateful to the organisers and those present at
the 2007 St. Andrews QRA meeting on the British and Irish Ice
Sheet, from which many of the ideas and stimulus came about. He is
also indebted to Colm O
´Cofaigh, Andreas Vieli and Danny McCar-
roll for careful reviews and useful discussions on modelling with
Richard Hindmarsh, Frank Pattyn and Christian Schoof, and
particularly to David Pollard for passing on his grounding-line
boundary-layer implementation.
Alley, R.B., 2000. The Younger Dryas cold interval as viewed from central Greenland.
Quaternary Science Reviews 19, 213–226.
Alley, R.B., Clark, P.U., Huybrechts, P., Joughin, I., 2005. Ice-sheet and sea-level
changes. Science 310, 456–460.
Alley, R.B., Meese, D.A., Shuman, C.A., Gow, A.J., Taylor, K.C., Grootes, P.M.,
White, J.W.C., Ram, M., Waddington, E.D., Mayewski, P.A., Zielinski, G.A., 1993.
Abrupt increase in Greenland snow accumulation at the end of the Younger
Dryas event. Nature 362, 527–529.
Ballantyne, C.K., Hall, A.M., 2008. The altitude of the last ice sheet in Caithness and
east Sutherland, Northern Scotland. Scottish Journal of Geology 44, 169–181.
Ballantyne, C.K., McCarroll, D., Nesje, A., Dahl, S.O., Stone, J.O., Fifield, L.K., 1998.
High-resolution reconstruction of the last ice sheet in NW Scotland. Terra Nova
10, 63–67.
Ballantyne, C.K., McCarroll, D., Stone, J.O., 2006. Vertical dimensions and age of the
Wicklow Mountains ice dome, Eastern Ireland, and implications for the extent
of the last Irish ice sheet. Quaternary Science Reviews 25, 2048–2058.
Ballantyne, C.K., McCarroll, D., Stone, J.O., 2007. The Donegal ice dome, NW Ireland:
dimensions and chronology. Journal of Quaternary Science 22, 773–783.
Ballantyne, C.K., Stone, J.O., McCarroll, D., 2008. Dimensions and chronology
of the last ice sheet in Western Ireland. Quaternary Science Reviews 27,
Bos, J.A.A., Dickson, J.H., Coope, G.R., Jardine, W.G., 2004. Flora, fauna and climate of
Scotland during the Weichselian Middle Pleniglacial – palynological, macro-
fossil and coleopteran investigations. Palaeogeography, Palaeoclimatology,
Palaeoecology 2004, 65–100.
Boulton, G.S., 1986. Push-moraines and glacier-contact fans in marine and terres-
trial environments. Sedimentology 33, 677–698.
Boulton, G.S., Hagdorn, M., 2006. Glaciology of the British Isles Ice Sheet during the
last glacial cycle: form, flow, streams and lobes. Quaternary Science Reviews 25,
Boulton, G.S., Jones, A.S., Clayton, K.M., Kenning, M.J., 1977. A British Ice Sheet Model
and Patterns of Glacial Erosion and Deposition in Britain. British Quaternary
Studies: Recent Advances. Oxford University Press., pp 231–246.
Boulton, G.S., Peacock, J.D., Sutherland, D.G., 1991. Quaternary. In: Craig, G.Y. (Ed.),
Geology of Scotland. The Geological Society, London, pp. 503–543.
Boulton, G.S., Smith, G.D., Jones, A.S., Newsome, J., 1985. Glacial geology and
glaciology of the last mid-latitude ice sheets. Journal of the Geological Society,
London 142, 447–474.
Bowen, D.Q., Phillips, F.M., McCabe, A.M., Knutz, P.C., Sykes, G.A., 2002. New data for
the Last Glacial Maximum in Great Britain and Ireland. Quaternary Science
Reviews 21, 89–101.
Box, J.E., Ski, K., 2007. Remote sounding of Greenland supraglacial melt lakes:
implications to sub-glacial hydraulics. Journal of Glaciology 181, 257–265.
Bradwell, T., Stoker, M.S., Golledge, N.R., Wilson, C., Merrit, J., Long, D., Everest, J.D.,
Hestvik, O.B., Stevenson, A., Hubbard, A., Finlayson, A., Mathers, H., 2008. The
northern sector of the Last British Ice Sheet: maximum extent and demise.
Earth Science Reviews 88, 207–226.
Bradwell, T., Stoker, M.S., Larter, R., 2007. Geomorphological signature and flow
dynamics of The Minch palaeo-ice stream, northwest Scotland. Journal of
Quaternary Science 22, 609–617.
Braithwaite, R.J., 1995. Positive degree-day factors for ablation on the Greenland
Ice Sheet studied by energy-balance modelling. Journal of Glaciology 41,
Brown, C., Meier, M., Post, A., 1982. Calving speed of Alaskan tidewater glaciers,
with application to Columbia Glacier. USGS Professional Paper 1258-C, 13 pp.
Brown, E.J., Rose, J., Coope, R.G., Lowe, J.J., 2006. An MIS 3 age organic deposit from
Balglass Burn, central Scotland: palaeoenvironmental significance and impli-
cations for the timing of the onset of the LGM ice sheet in the vicinity of the
British Isles. Journal of Quaternary Science 22, 295–308.
Carr, S.J., Holmes, R., van der Meer, J.J.M., Rose, J., 2006. The Last Glacial Maximum in
the North Sea Basin: micromorphological evidence of extensive glaciation.
Journal of Quaternary Science 21, 131–153.
Clapperton, C.M., 1971. The location and origin of glacial meltwater phenomena in
the eastern Cheviot Hills. Proceedings of the Yorkshire Geological Society 38,
Clark, C.D., Evans, D.J.A., Khatwa, A., Bradwell, T., Jordan, C.J., Marsh, S.H.,
Mitchell, W.A., Bateman, M.D., 2004. Map and GIS database of glacial landforms
and features related to the last British Ice Sheet. Boreas 33, 359–375.
Dansgaard, W., White, J.W.C., Johnsen, S.J., 1989. The abrupt termination of the
Younger Dryas climate event. Nature 339, 532–534.
Das, S.B., Joughin, I., Behn, M.D., Howat, I.M., King, M.A., Lizarralde, D., Bhatia, M.P.,
2008. Fracture propagation to the base of the Greenland Ice Sheet during
supraglacial lake drainage. Science 320, 778–781.
Denton, G.H., Alley, R.B., Comer, G.C., Broecker, W.S., 2005. The role of seasonality in
abrupt climate change. Quaternary Science Reviews 24, 1159–1182.
Evans, D.J.A., Clark, C.D., Mitchell, W.A., 2005. The last British Ice Sheet: a review of
the evidence utilised in the compilation of the Glacial Map of Britain. Earth-
Science Reviews 70 (3/4), 253–312.
Everest, J.D., Bradwell, T., Golledge, N.R., 2005. Subglacial landforms of the Tweed
Palaeo-Ice Stream. Scottish Geographical Journal 121, 163–173.
Golledge, N., Stoker, M.S., 2006. A palaeo-ice stream of the British Ice Sheet in
eastern Scotland. Boreas 35, 231–243.
Golledge, N.R., 2007. An ice cap landsystem for palaeoglaciological reconstructions:
characterizing the Younger Dryas in western Scotland. Quaternary Science
Reviews 26, 213–229.
Golledge, N.R., Hubbard, A., Sugden, D.E., 2008. High-resolution numerical simu-
lation of Younger Dryas glaciation in Scotland. Quaternary Science Reviews 27,
Golledge, N.R., Merritt, J.W., 2005. Palaeo-iceflow and glacial chronology in the
Solway Lowlands interpreted from a new 1.5 m-resolution DSM. Presented at:
International Conference on Glacial Sedimentary Processes and Products,
August 22–27, 2005. University of Wales, Aberystwyth.
Graham, A.G.C., Lonergan, L., Stoker, M.S., 2007. Evidence for Late Pleistocene ice
stream activity in the Witch Ground Basin, central North Sea, from 3D seismic
reflection data. Quaternary Science Reviews 26, 627–643.
Graham, A.G.C., Lonergan, L., Stoker, M.S., 2009. Seafloor glacial features reveal the
extent and decay of the last British Ice Sheet, east of Scotland. Journal of
Quaternary Science 24, 117–138.
Greenwood, S., Clark, C., 2008. Subglacial bedforms of the Irish Ice Sheet. Journal of
Hall, A.M., Glasser, N.F., 2003. Reconstructing the basal thermal regime of an ice
stream in a landscape of selective linear erosion: Glen Avon, Cairngorm
Mountains, Scotland. Boreas 32, 191–207.
Hall, A.M., Peacock, J.D., Connell, E.R., 2003. New data for the Last Glacial Maximum
in Great Britain and Ireland: comments on the paper by Bowen et al. (2002).
Quaternary Science Reviews 22, 1551–1554.
Hiemstra, J.F., Evans, D.J.A., Scourse, J.D., Furze, M.F.A., McCarroll, D., Rhodes, E.,
2006. New evidence for a grounded Irish Sea glaciation of the Isles of Scilly, U.K.
Quaternary Science Reviews 25, 299–309.
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777776
Hindmarsh, R., 2004. A numerical comparison of approximations to the Stokes
equations used in ice sheet and glacier modeling. Journal of Geophysical
Research 109, doi:10.1029/2003JF000065.
Hindmarsh, R.C., Le Meur, E., 2001. Dynamical processes involved in the retreat of
marine ice sheets. Journal of Glaciology 47, 271–282.
Hubbard, A., 1999. High-resolution modeling of the advance of the Younger Dryas
Ice Sheet and its climate in Scotland. Quaternary Research 52 (1), 27–43.
Hubbard, A., Blatter, B., Nienow, P., Mair, D., Hubbard, B., 1998. Comparison of
a three-dimensional model for glacier flow with field data from Haut Glacier
d’Arolla, Switzerland. Journal of Glaciology 44 (147), 368–378.
Hubbard, A., 2000. The verification and significance of three approaches to longi-
tudinal stresses in high-resolutions models of glacier flow. Geografiska Annaler
82, 471–487.
Hubbard, A., Sugden, D.E., Dugmore, A., Norddahl, H., Pe
´tursson, H.G., 2006. A
modelling insight into the Icelandic Last Glacial Maximum ice sheet. Quater-
nary Science Reviews 25, 2283–2296.
Hulbe, C., Fahnestock, M., 2007. Century-scale discharge stagnation and reactivation
of the Ross ice streams, West Antarctica. Journal Of Geophysical Research 112,
F03S27, doi:10.1029/2006JF000603.
Jansson, K.N., Glasser, N.F., 2008. Modification of peripheral mountain ranges by
former ice sheets: the Brecon Beacons, Southern UK. Geomorphology 97,
´hannesson, T., Sigurdsson, O., Laumann, T., Kennett, M., 1995. Degree-day glacier
mass-balance modelling with applications to glaciers in Iceland, Norway and
Greenland. Journal of Glaciology 41 (138), 345–358.
Joughin, I., Das, S.B., King, M.A., Smith, B.E., Howat, I.M., Moon, T., 2008. Seasonal
speedup along the western flank of the Greenland Ice Sheet. Science 320,
Knutz, P.C., Austin, W.E.N., Jones, E.J.W., 2001. Millennial-scale depositional cycles
related to British ice sheet variability and North Atlantic paleocirculation since
45 kyr B.P., Barra Fan, U.K. margin. Palaeoceanography 16, 53–64.
Kristensen, T.B., Piotrowski, J.A., Huuse, M., Clausen, O.R., Hamberg, L., 2008. Time-
transgressive tunnel valley formation indicated by infill sediment structure,
North Sea – the role of glaciohydraulic supercooling. Earth Surface Processes
and Landforms 33, 546–559.
Lambeck, K., Purcell, A.P., 2001. Sea-level change in the Irish Sea since the Last
Glacial Maximum: constraints from isostatic modelling. Journal of Quaternary
Science 16, 497–506.
Laumann, T., Reeh, N., 1993. Sensitivity to climate change of the mass balance of
glaciers in southern Norway. Journal of Glaciology 39 (133), 656–665.
Le Meur, E., Huybrechts, P., 1996. A comparison of different ways of dealing with
isostasy: examples from modelling the Antarctic ice sheet during the last glacial
cycle. Annals of Glaciology 23, 309–317.
Lonergan, L., Maidment, S.C.R., Collier, J.S., 2006. Pleistocene subglacial tunnel
valleys in the central North Sea basin: 3-D morphology and evolution. Journal of
Quaternary Science 21, 891–903.
MacAyeal, D.R., 1993. Binge/purge oscillations of the Laurentide Ice Sheet as
a cause of the North Atlantic’s Heinrich events. Paleoceanography 8 (6),
Marshall, S.J., Bjo
¨rnsson, H., Flowers, G.E., Clarke, G.K.C., 2005. Simulation of Vat-
¨kull ice cap dynamics. Journal of Geophysical Research 110, F03009,
Merritt, J.W., Auton, C.A., Firth, C.A., 1995. Ice-proximal glaciomarine sedimentation
and sea-level change in the Inverness Area, Scotland: a review of the deglaci-
ation of a major ice stream of the British Late Devensian ice sheet. Quaternary
Science Reviews 14, 289–331.
Merritt, J.W., Auton, C.A., Connell, E.R., Hall, A.M., Peacock, J.D., 2003. Cainozoic
Geology and Landscape Evolution of North-east Scotland. British Geological
Survey, Edinburgh, 178 pp.
´Cofaigh, C., Evans, D.J.A., 2007. Radiocarbon constraints on the age of the
maximum advance of the British–Irish Ice Sheet in the Celtic Sea. Quaternary
Science Reviews 26, 1197–1203.
Paterson, W.S.B., 1994. The Physics of Glaciers, third ed. Pergamon, Oxford.
Pattyn, F., et al., 2008. Benchmark experiments for higher-order and full Stokes ice
sheet models (ISMIP-HOM). The Cryosphere Discussions 2 (1), 111–151.
Pattyn, F., Huyghe, A., De Brabander, S., De Smedt, B., 2006. The role of transition
zones in marine ice sheet dynamics. Journal of Geophysical Research (Earth
Surface) 111 (F2), F02004, doi:10.1029/2005JF000394.
Payne, A.J., Sugden, D.E., 1990. Topography and ice sheet growth. Earth Surface
Processes and Landforms 15, 625–639.
Peck, V.L., Hall, I.R., Zahn, R., Grousset, F., Hemming, S.R., Scourse, J.D., 2007. The
relationship of Heinrich events and their European precursors over the past
60 ka BP: a multi-proxy ice-rafted debris provenance study in the North East
Atlantic. Quaternary Science Reviews 26, 862–875.
Peltier, W.R., 1998. Postglacial variations in the level of the sea: implications for
climate dynamics and solid-earth geophysics. Reviews of Geophysics 36,
Phillips, W.M., Hall, A.M., Ballantyne, C.K., Binnie, S., Kubik, P.W., Freeman, S., 2008.
Extent of the last ice sheet in northern Britain tested with cosmogenic
exposure ages. Journal of Quaternary Science 23, 101–107.
Phillips, W.M., Hall, A.M., Mottram, R., Fifield, L.K., Sugden, D.E., 2006. Cosmogenic
Be and
Al exposure ages of tors and erratics, Cairngorm Mountains, Scot-
land: Timescales for the development of a classic landscape of selective linear
glacial erosion. Geomorphology 73, 222–245.
Pollard, D., DeConto, R.M., 2007. A coupled ice-sheet/ice-shelf/sediment model
applied to a marine-margin flowline: forced and unforced variations. In:
Hambrey, M.J., Christoffersen, P., Glasser, N.F., Hubbard, B. (Eds.), Glacial Sedi-
mentary Processes and Products. International Association of Sedimentologists
Special Publication, 39. Wiley–Blackwell.
Rollin, K.E., Kirby, G.A., Rowley, W.J., Buckley, D.K., 1993. Atlas of Geothermal
Resources in Europe: UK Revision. British Geological Survey Technical Report
WK/95/07. British Geological Survey, Nottingham.
Ruddiman, W.F., McIntyre, A., 1981. The mode and mechanism of the last deglaci-
ation: oceanic evidence. Quaternary Research 16, 125–134.
Schoof, C., 2007. Ice sheet grounding line dynamics: steady states, stability and
hysteresis. Journal of Geophysical Research (Earth Surface) 112, F03S28,
Scourse, J.D., Furze, M.F.A., 2001. A critical review of the glaciomarine model for
Irish sea deglaciation: evidence from southern Britain, the Celtic shelf and
adjacent continental slope. Journal of Quaternary Science 16 (5), 419–434.
Sejrup, H.P., Haflidason, H., Aarseth, I., King, E., Forsberg, C.F., Long, D.,
Rokoengen, K., 1994. Late Weichselian glaciation history of the northern North
Sea. Boreas 23, 1–13.
Sejrup, H.P., Hjelstuen, B.O., Torbjørn Dahlgren, K.I., Haflidason, H., Kuijpers, A.,
`rd, A., Praeg, D., Stoker, M.S., Vorren, T.O., 2005. Pleistocene glacial history
of the NW European continental margin. Marine and Petroleum Geology 22
(9/10), 1111–1129.
Shennan, I., Bradley, S., Milne, G., Brooks, A., Bassett, S., Hamilton, S., 2005. Relative
sea-level changes, glacial isostatic modelling and ice-sheet reconstructions
from the British Isles since the Last Glacial Maximum. Journal of Quaternary
Science 21, 585–599.
Shepherd, A., Wingham, D., 2007. Recent sea-level contributions of the Antarctic
and Greenland Ice Sheets. Science 315, 1529–1532.
Sissons, J.B., 1967. The Evolution of Scotland’s Scenery. Oliver and Boyd, Edinburgh,
259 pp.
Sissons, J.B., 1979. Palaeoclimatic inferences from former glaciers in Scotland and
the Lake District. Nature 278, 518–521.
Stoker, M.S., Bradwell, T., 2005. The Minch palaeo-ice stream, NW sector of the
British–Irish Ice Sheet. Journal of the Geological Society 162, 425–428.
Stoker, M.S., Hitchen, K., Graham, C.C., 1993. United Kingdom Offshore Regional
Report: the geology of the Hebrides and the West Shetland Shelves, and
Adjacent Deep Water Areas. H.M.S.O., London.
Stone, J., Ballantyne, C.K., 2006. Dimensions and deglacial chronology of the Outer
Hebrides Ice Cap, northwest Scotland: implications of cosmic ray exposure
dating. Journal of Quaternary Science 21, 75–84.
Sugden, D.E., 1968. The selectivity of glacial erosion in the Cairngorm Mountains,
Scotland. Transactions of the Institute of British Geographers 45, 79–92.
Sutherland, D.G., 1984. The Quaternary deposits and landforms of Scotland and the
neighbouring shelves: a review. Quaternary Science Reviews 3, 157–254.
van der Veen, C.J., 2002. Polar ice sheets and global sealevel: how well can we
predict the future? Global and Planetary Change 32, 165–194.
Vieli, A., Payne, A., 2005. Assessing the ability of numerical ice sheet models to
simulate grounding line migration. Journal of Geophysical Research 110,
F01003, doi:10.1029/2004JF000202.
van de Wal, R.S.W., Greuell, W., van den Broeke, M.R., Reijmer, C.H., Oerlemans, J.,
2005. Surface mass-balance observations and automatic weather station data
along a transect near Kangerlussuaq, West Greenland. Annals of Glaciology 42,
Warren, W.P., 1985. Stratigraphy. In: Edwards, K.J., Warren, W.P. (Eds.), The
Quaternary History of Ireland. Academic Press, London, pp. 39–65.
Weertman, J., 1964. The theory of glacier sliding. Journal of Glaciology 5, 287–303.
Whittington, G., Hall, A.M., 2002. The Tolsta Interstadial, Scotland: correlation with
D-O cycles GI-8 to GI-5? Quaternary Science Reviews 21, 901–915.
Wingfield, R., 1990. The origin of major incisions within Pleistocene deposits of the
North Sea. Marine Geology 91, 31–52.
Witte, H.J.L., Coope, G.R., Lemdahl, G., Lowe, J.J., 1998. Regression coefficients of
thermal gradients in northwestern Europe during the Last Glacial–Holocene
transition using beetle MCR data. Journal of Quaternary Science 13, 447–454.
A. Hubbard et al. / Quaternary Science Reviews 28 (2009) 759–777 777
... A newly published interpretation of those data (O'Cofaigh et al., 2021) suggests the retreating ice margin reached the Aran Islands by ~19.5 ka and that the retreat rate dropped significantly thereafter, with deglaciation of interior Connemara taking a further two millennia ( Fig. 12 of O'Cofaigh et al. (2021); see Section 5.2 for discussion). Modelling simulations by Roberts et al. (2020) depict the ice sheet then retreating eastward into the Irish midlands in a manner similar to simulations by Hubbard et al. (2009). ...
... This configuration is noteworthy because, coupled with the absence of evidence for regional flow reversals, it suggests that the post-LGM ice margin retreated back to its Connemara accumulation zone, as proposed by earlier depictions (Kinahan and Close, 1872;Charlesworth, 1929;Orme, 1965;Warren, 1991Warren, , 1992Greenwood and Clark, 2009;C. Clark et al., 2012), rather than eastward into the Irish midlands (e.g., Hubbard et al., 2009;Roberts et al., 2020). The pattern of quasi-radial ice flow and margin retreat described here is broadly, if not perfectly, consistent with the regional ice divide proposed by Warren (1992) and Warren and Ashley (1994), but harder to reconcile with models depicting a single, centrally located ice divide on the island of Ireland (e.g., McCabe, 1985;Knight, 2000). ...
... Clark et al., 2012), we note that this age aligns broadly with the range presented by Roberts et al. (2020). However, unlike that previous study, we suggest that neither the new 10 Be dataset nor the provenance of indicator erratics supports the conceptual model of time-transgressive eastward retreat of the ice margin (Hubbard et al., 2009;Roberts et al., 2020). Instead, they describe a virtually instantaneous (within the resolution of our data) removal of a regional ice mass, the eastern portion of which was flowing approximately east out of Connemara immediately prior to its demise (Figs. ...
Full-text available
Late Pleistocene stadials are traditionally viewed as periods of severe, year-round cooling centred on the North Atlantic, where the abrupt onset and termination of stadial conditions is widely, though not unequivocally, thought to reflect the impact on Atlantic Meridional Circulation of meltwater fluxes from fringing ice masses. Widespread subaerial melting of land-based ice, however, implies a state of glacier-climate disequilibrium that, according to some conceptual models, had to persist for the duration of each stadial event rather than immediately prior to it. Yet, to date, the spatial extent of any such melting remains unclear, as do the magnitude and duration of any cognate atmospheric warming. To help address this ambiguity, we reconstructed the timing and nature of deglaciation of the Connemara ice centre, Ireland, which, being located immediately downwind of the North Atlantic, affords an ideal location for assessing terrestrial expressions of stadial and non-stadial conditions. We report 15 new cosmogenic beryllium-10 ages of glacial erratic boulders that reveal the timing and magnitude of fluctuations in this sector of the former Irish ice sheet during Heinrich Stadial 1 (HS1). Coupled with geomorphic mapping, these results indicate rapid, widespread deglaciation of the Connemara ice centre at approximately 17 ka, under full stadial conditions, and that the ice margin retreated back to regional source areas rather than eastward towards the centre of the island. Recognising the potential divergence from the traditional view of cold North Atlantic stadials, we propose that the rate and nature of deglaciation in Connemara during HS1 likely reflects enhanced subaerial melting during the summer ablation season.
... 30-14.5 ka BP) play an important role in the development and improvement of numerical ice-sheet models used to predict future ice-sheet responses to climate change (e.g. Hubbard et al., 2009;Clark et al., 2012;Gandy et al. 2019). A well-preserved glacial 'footprint' indicates that the British-Irish Ice Sheet was highly dynamic, reflecting its relatively small size and southerly position on the eastern Atlantic margin. ...
... Recent attempts at computer-based modelling of the evolution of the British-Irish Ice Sheet have succeeded in simulating the marine margins of the ice sheet well, but have only limited success in modelling terrestrial margins, particularly during the later recession of the IIS (e.g. Hubbard et al., 2009;Gandy et al., 2019). The work presented here indicates that this reflects a poor understanding of the controls on ice recession rates across Ireland during the Last Glacial Termination and that much work remains to be done in reconstructing those margins. ...
Full-text available
The retreat of the last Irish Ice Sheet across the Irish midlands resulted in the formation of an ice‐contact lake, Palaeolake Riada, along its eastern margin during recession. However, the full extent and characteristics of the lake and its impact on ice dynamics are unknown, as glacigenic landforms and deposits relating to the lake have not been mapped in sufficient detail to allow reconstruction of ice‐marginal behaviour. This paper presents new mapping of subglacial, ice‐marginal and glaciolacustrine landforms and deposits in the Irish midlands. Glacial landforms and landform associations across this wide area have been identified using aerial photographs and high‐resolution digital elevation models derived from LiDAR and radar data. Reconstructions of successive ice‐margin retreat positions are combined with glacio‐isostatically adjusted digital elevation models to identify the probable position of lake outflows and reconstruct the development of the proglacial Palaeolake Riada through time. These reconstructions indicate that at its largest, Palaeolake Riada had an ice‐contact margin of 113 km, and covered an area of over 2300 km2. Initially, ice recession was concentrated around meltwater discharge outlets, but these became less important as the lake expanded and thermo‐mechanical erosion occurred along the entire subaqueous margin.
... Several published models argue that during the last glaciation the BIIS (Devensian) and FIS (Weichselian) converged within the central part of the North Sea basin (e.g., Boulton and Hagdorn, 2006;Carr et al., 2006;Graham et al., 2007Graham et al., , 2011Bradwell et al., 2008;Sejrup et al., 2009Sejrup et al., , 2016Hughes et al., 2016). However, the maximum extents of these ice sheets are as yet poorly constrained (see Fig. 1) (e.g., Jansen et al., 1979;Catt, 1991;Clark et al., 2004;Hubbard et al., 2009;Brooks et al., 2008;Sejrup et al., 2009Sejrup et al., , 2016. This uncertainty partly resulted from the fact that, until recently, very little was known about the Quaternary sequence underlying Dogger Bank; an isolated, approximately Proceedings of the Geologists' Association xxx (xxxx) xxx ⁎ Corresponding author. ...
... Cotterill et al. (2017a) suggested, based upon unpublished sediment provenance data, that Dogger Bank was overridden by the FIS from the north. The western ice mass, however, is most likely to have been the North Sea lobe of the BIIS which is widely regarded to have extended into this part of the southern central North Sea during the Late Devensian (Figs. 1 and 13) (e.g., Jansen et al., 1979;Clark et al., 2004;Carr et al., 2006;Brooks et al., 2008;Hubbard et al., 2009;Evans et al., 2021); its maximum offshore extent being equated with the distribution of the Bolders Bank Formation (see Fig. 13) Cameron et al., 1992;Dove et al., 2017). The North Sea lobe was fed by ice from the Firth of Forth that flowed southwards through the North Sea Basin broadly parallel to the contemporary coastline of eastern England Evans et al., 2021, and references therein). ...
High-resolution 2D seismic data from the western side of Dogger Bank (North Sea) has revealed that the glacigenic sediments of the Dogger Bank Formation record a complex history of sedimentation and penecontemporaneous, large-scale, ice-marginal to proglacial glacitectonism. The resulting complex assemblage of glacial landforms and sediments record the interplay between two separate ice masses revealing that Late Devensian ice sheet dynamics across Dogger Bank were far more complex than previously thought, involving the North Sea lobe of the British and Irish Ice Sheet, advancing from the west, interacting with the Dogger Bank lobe which expanded from the north. The active northward retreat of the Dogger Bank lobe resulted in the development of a complex assemblage of arcuate thrust-block moraines (≤ 15 km wide, > 30 km long) composed of highly folded and thrust sediments, separated by sedimentary basins and meltwater channels filled by outwash. The impact of the North Sea lobe was restricted to the western margin of Dogger Bank and led to deep-seated (100–150 m thick) glacitectonism in response to ice-push from the west. During the earlier expansion of the North Sea lobe, this thrust and fold complex initially occupied a frontal marginal position changing to a more lateral ice-marginal position as the ice sheet continued to expand to the south. The complex structural relationships between the two glacitectonic complexes indicates that these ice masses interacted along the western side of Dogger Bank, with the inundation of this area by ice probably occurring during the last glaciation when the ice sheets attained their maximum extents.
... Reconstructing the nature and behaviour of palaeo-glaciers enables a better understanding of the long-term (centennial to multi-millennial) character of glaciers within the Earth-atmosphere system (e.g. Clark 1997;Boulton et al. 2001;Clark et al., 2018;Evans et al. 2009;Hubbard et al. 2009;Ely et al. 2021). In particular, the relative sensitivity of mid-latitude glaciers to modern climate change makes them especially important to consider compared to other regions around the world. ...
Full-text available
The extent of the Southern Alps icefield in New Zealand is well-constrained chronologically for the last glacial cycle. The sediment-landform imprint of this glacial system, however, offers insight into ice-marginal processes that chronological control cannot. We present the first detailed investigation of sediments along the southwestern shores of Lake Tekapo, South Island. We identify seven lithofacies, from which a five-stage palaeoglaciological reconstruction of depositional and glaciotectonic events is proposed: (i) ice-marginal advance and deposition of outwash gravels in lithofacies (LF) 1; (ii) ice-marginal recession and the development of an ice-contact lake, manifest in rhythmite deposition and iceberg rafting of dropstones (LF 2), followed by a depositional hiatus; (iii) ice-marginal recession recorded in ice-proximal aggradation of glaciofluvial hyperconcentrated flows (LFs 3, 4); (iv) ice-marginal advance documented by glaciotectonic disturbance and localized hydrofracturing, coeval with the deposition of delta foresets and a subglacial diamicton/till (LFs 5, 6); (v) final stages of ice-marginal recession and deposition of outwash gravels in LF 7. Two infrared stimulated luminescence ages were obtained from the glaciolacustrine sediments and, whilst the dating has some limitations, the sediments pre-date both the global and local Last Glacial Maximum. Overall, this sequence, consistent with sediment fills recorded elsewhere across South Island, suggests recurrence of processes from different glacial advances and the role of topographic constraints on maintaining lake positions.
... Throughout this period, the marineterminating BIIS was characterised by several ice streams, radiating from the interior of the ice sheet. The Minch Ice Stream (MIS) flowed from onshore Scotland towards the Hebrides Shelf in a north-western direction ( Fig. 2 Rignot et al., 2004;Hubbard et al., 2009;Pritchard et al., 2009). ...
Palaeo ice-sheet reconstructions are considered a key approach to increase our understanding of past climate change and how this impacts on the cryosphere. These reconstructions have shown that ice sheets can have a relatively fast response to climate and ocean forcing mechanisms. This has raised concerns about the future stability of ice sheets in a warming world, especially those that are marine-based or marine-terminating, such as the Greenland and Antarctic Ice Sheets. However, predictions of ice sheets stability are complex and their long-term accuracy remains a major weakness in climate change science. Palaeoglaciological reconstructions offer one critical approach to improve our understanding of how ice sheets respond to climatic drivers over full glacial cycles or through dynamic periods of ice sheet history. As such, palaeo-ice sheets act as useful analogues for helping to determine the important drivers that can influence ice sheets in the future. This study examines the southern margins of two former ice sheets: the British-Irish ice Sheet (BIIS) and the Newfoundland Ice Sheet (NIS). These are located on opposite sides of the North Atlantic Ocean but at similar latitudes. Both ice sheets were grounded below sea level, were drained by ice streams and had extensive marine margins potentially exposed to changes in large-scale ocean and atmospheric circulation. Therefore, they represent good analogues for modern marine-terminating ice sheets. New marine geophysical and sediment data were analysed across the Celtic Sea shelf, between Ireland and the UK, which was occupied by the Irish Sea Ice Stream (ISIS), the largest outlet of the BIIS. Geomorphological mapping shows a large meltwater drainage system, including tunnel valleys, beneath the central axis of the former ISIS. This evidence implies significant and erosive meltwater release, potentially influencing the rapid retreat across the shelf. At this stage (~25 cal BP), the ISIS was also retreating close to the southern Irish coastline. Some 30 km off the coast, a relatively large grounding-zone wedge was formed, together with a sequence of morainic ridges, which are capped by glacimarine laminated and massive muds. These features show a stepped retreat of the ISIS margin towards to the coastline. The difference in behaviour of the retreating ice sheet near the present-day coast compared to that in the central axis was probably governed by topographical and geological controls, including variations in water depth and the presence of bedrock outcrops acting as pinning points. On the other side of the Atlantic Ocean, new data on the southern shelf of Newfoundland were analysed. Here, fjords served as outlets for sediment-laden meltwater draining the former NIS. Intense and widespread calving occurred across the NIS marine margin following its extension to the shelf edge. When the ice sheet reached the present-day coastline, it stabilised at the mouth of the fjords and formed a series of moraines that record a still-stand of the ice margin. The combination of new and extant data suggest that the still-stand occurred between 16.3 ka cal BP and 15 ka cal BP. Stratified glacimarine sediments accumulated at the mouth of the fjords during a period of prevailing cold-water conditions when relative sea level was ~30 m higher than today. Comparison between the two study areas shows different topographic settings and asynchronous ice-sheet behaviour during the last deglaciation. The onset of retreat between the two ice sheets is around 10 ka apart. A comparison of the results against existing proxy data from the North Atlantic Ocean highlighted that deglaciation of both shelves was initiated in the absence of ocean warming, when eustatic sea level was at a minimum. Internal glaciological factors were therefore most likely responsible for the demise of both marine sectors. This demonstrates that marine-terminating ice margins can internally trigger their own demise in very different glaciological settings and within overall cold conditions. Such information provides additional data for ice sheet numerical models that investigate links between rate and pattern of retreat and the drivers of ice sheet variability.
... The key difference in these three models is in the development of the input BIIS reconstructions. The Kuchar model incorporates the output from a dynamic ice sheet model (Hubbard et al., 2009); both the B2011 and BC2020 have ice sheet loading histories produced using geomorphological data, with the latter adopting the output from a simple plastic ice sheet model (Gowan et al., 2016) constrained to the larger landform dataset from the BRITICE-CHRONO project (i.e Bradwell et al., 2019;Clark et al., 2017). Each GIA model is combined with a specific optimum Earth model determined by comparisons to the British Isles sea-level database (Shennan et al., 2018). ...
Full-text available
The record of ice-sheet demise since the last glacial maximum (LGM) provides an opportunity to test the relative importance of instability mechanisms, including relative sea-level (RSL) change, controlling ice-sheet retreat. Here we examine the record of RSL changes accompanying the retreat of the Minch Ice Stream (MnIS) of northwest Scotland during the deglaciation following the LGM as well as use the record to provide additional age constraints on a local late-glacial readvance known as the Wester Ross Readvance. We use new and existing records of RSL change obtained from isolation basins in Wester Ross along the flanks of the former MnIS to test available glacial-isostatic adjustment (GIA) predictions of the deglacial RSL history for the region. Using these GIA model predictions we examine the nature of RSL change across the retreating front of the MnIS through the early deglaciation. Our new radiocarbon ages from these basins confirm the timing of deglaciation within the inner trough of the former MnIS as well as refines the age of the Wester Ross Readvance, both established by earlier cosmogenic-based studies. We find that the Wester Ross Readvance culminated around 15.8 ± 0.1 ka, slightly earlier than recent suggestions. Near Gairloch, Wester Ross, RSL fell from a marine limit ∼20 m above present at ∼16.1–16.5 ka. Three isolation basins record RSL fall over the following ∼0.8 ka allowing a comparison between GIA predictions and RSL observations. Our new analyses suggest that the rate of RSL rise increased at the ice front, in concert with the MnIS encountering a landward sloping bed potentially aiding the rapid retreat of the MnIS from 17.6 to 16.4 ka BP. This observation suggests that GIA during deglaciation does not necessarily induce a stabilizing RSL change to marine-based ice streams as some models have suggested. Along indented ice margins, the RSL field at the front of individual ice streams may be governed by the regional GIA signal driven by the ice sheet as a whole, rather than the local ice front. In addition, the stabilizing impact of post-glacial rebound is dependent on an Earth rheology weak enough to respond quickly to the ice-sheet retreat. In the case of the MnIS, the RSL experienced at the front of the ice stream was likely governed by the earlier ice mass extent, the larger ice masses lying to the east and south of the highly indented ice front, and the relatively strong Earth rheology beneath the British Isles. Thus, the geometry of the ice sheet margins, such as those in Greenland and Antarctica today, and the Earth rheology beneath them need to be taken into account when considering the stabilizing impact of post-glacial rebound on marine ice sheet retreat.
... The Late Devensian ice sheet expanded from mountain source areas after *35 to 32 ka, and for much of its existence was drained by several large ice streams (Clark et al. 2012;Hughes et al. 2014;Ballantyne and Small 2019). One of these, the Minch Ice Stream, dominated the ice-sheet flow configuration in NW Scotland (Bradwell et al. 2007;Hubbard et al. 2009;Bradwell and Stoker 2015a). The trunk of this ice stream occupied The Minch and continued northwestwards across the continental shelf within a well-defined cross-shelf trough, probably following the route of previous ice streams that developed during earlier periods of ice-sheet glaciation. ...
The far northwest of mainland Scotland is renowned for its scenery, structural complexity and geodiversity, and is designated as a UNESCO Global Geopark. The region is bisected by the Moine Thrust Zone (MTZ) , west of which a foreland of undeformed Archaean gneiss supports inselbergs of Neoproterozoic and Palaeozoic rocks, and east of which are thrust-stacked, deformed metasedimentary rocks of the Neoproterozoic Moine Supergroup. The MTZ forms a north–south belt within which rocks were extensively thrust and folded during the Caledonian Orogeny. Successive Pleistocene glaciations have resulted in an array of erosional landforms: troughs, rock basins, cirques, glacially steepened inselbergs, extensive areas of knock-and-lochan terrain and clusters of glacial megagrooves. During the last and earlier ice-sheet glaciations, the region sourced northwestward-flowing ice feeding the Minch Ice Stream, which extended far across the adjacent shelf, but by ~15 ka the last ice sheet had retreated to its mountain heartland. The Loch Lomond Stade (~12.9 to 11.7 ka) witnessed reoccupation of the main mountain axis by a substantial (~350 km²) icefield, and cirque glaciers formed on peripheral mountains; the extent of the former is mainly delimited by multiple recessional moraines, the latter by end-moraine belts. Lateglacial and Holocene landforms include outwash or delta terraces at fjord heads, sea stacks, beaches backed by sand dunes, rock-slope failures, relict talus accumulations, and active periglacial and aeolian features on high ground. Karst terrain developed on dolostones comprises sinkholes, resurgences and extensive cave networks formed by water-table lowering due to Middle and Late Pleistocene valley deepening.
Full-text available
Palaeo-glacial landforms can give insights into bed roughness that currently cannot be captured underneath contemporary-ice streams. A few studies have measured bed roughness of palaeo-ice streams but the bed roughness of specific landform assemblages has not been assessed. If glacial landform assemblages have a characteristic bed-roughness signature, this could potentially be used to constrain where certain landform assemblages exist underneath contemporary-ice sheets. To test this, bed roughness was calculated along 5 m × 5 m resolution transects (NEXTMap DTM, 5 m resolution), which were placed over glacial landform assemblages (e.g. drumlins) in the UK. We find that a combination of total roughness and anisotropy of roughness can be used to define characteristic roughness signatures of glacial landform assemblages. The results show that different window sizes are required to determine the characteristic roughness for a wide range of landform types and to produce bed-roughness signatures of these. Mega scale glacial lineations on average have the lowest bed-roughness values and are the most anisotropic landform assemblage.
New relative sea‐level (RSL) data constrain the timing and magnitude of RSL changes in the southern Isle of Skye following the Last Glacial Maximum (LGM). We identify a marine limit at ~23 m OD, indicating RSL ~20 m above present c. 15.1 ka. Isolation basin data, supported by terrestrial and marine limiting dates, record an RSL fall to 11.59 m above present by c. 14.2 ka. This RSL fall occurs across the time of global Meltwater Pulse 1A, supporting recent research on the sources of ice melting. Our new data also help to resolve some of the chronological issues within the existing Isle of Skye RSL record and provide details of the sub‐Arctic marine environment associated with the transition into Devensian Lateglacial climate at c. 14.5 k cal a bp, and the timing of changes in response to the Loch Lomond Stadial climate. Glacio‐isostatic adjustment (GIA) model predictions of RSL deviate from the RSL constraints and reflect uncertainties in local and global ice models used within the GIA models. An empirical RSL curve provides a target for future research.
The Quaternary Period in Scotland was characterized by major climatic shifts and the alternation of glacial and temperate conditions over a wide range of timescales. The extent of multiple glaciations prior to the Mid-Pleistocene Transition (1.25–0.70 Ma) is uncertain, but thereafter up to ten major episodes of ice-sheet expansion occurred. Glacial erosion by successive glaciers and ice sheets created a range of terrain types: glaciated mountains, zones of areal scouring, landscapes of selective linear erosion, drift-mantled terrain of differential erosion and areas of limited glacial modification. The last ice sheet (~35–14 ka) extended to the shelf edge in the west and was confluent with the Fennoscandian Ice Sheet in the North Sea Basin; during its existence, it experienced marked changes in configuration, in part driven by the development of major ice streams. Subsequent glaciation during the Loch Lomond Stade (~12.9–11.7 ka) was restricted to a major icefield in the western Highlands and smaller glaciers in peripheral mountain areas. Contrasting glacial landsystems occupy terrain inside and outside the limits of the Loch Lomond Stadial glaciers. Postglacial landscape changes have been characterized by Lateglacial periglaciation and paraglacial landscape modification, mainly in the form of rock-slope failure and the accumulation (then later erosion) of paraglacial sediment stores and incision and terracing of glacigenic valley fills. Shore platforms of various ages formed around rock coastlines during the Quaternary, glacio-isostatic uplift has resulted in the formation of Lateglacial and Holocene raised beaches, and reworking of glacigenic deposits has provided sediments for present-day beach and dune systems.
Full-text available
The bedrock isostatic response exerts a strong control on ice-sheet dynamics and is therefore always taken into account in ice-sheet models. This paper reviews the various methods normally used in the ice-sheet modelling community to deal with the bedrock response and compares these with a more sophisticated full-Earth model. Each of these bedrock treatments, five in total, is coupled with a three-dimensional thermomechanical ice-sheet model under the same forcing conditions to simulate the Antarctic ice sheet during the last glacial cycle. The outputs of the simulations are compared on the basis of the time-dependent behaviour for the total ice volume and the mean bedrock elevation during the cycle and of the present rate of uplift over Antarctica. This comparison confirms the necessity of accounting for the elastic bending of the lithosphere in order to yield realistic bedrock patterns. It furthermore demonstrates the deficiencies inherent to the diffusion equation in modelling the complex deformation within the mantle. Nevertheless, when characteristic parameters are varied within their range of uncertainty, differences within one single method are often of the same order as those between the various methods. This overview finally attempts to point out the main advantages and drawbacks of each of these methods and to determine which one is most appropriate depending on the specific modelling requirements.
Full-text available
A degree-day glacier mass-balance model is applied to three glaciers in Iceland, Norway and Greenland for which detailed mass-balance measurements are available over a period of several years. Model results are in good agreement with measured variations in the mass balance with elevation over the time periods considered for each glacier. In addition, the model explains 60-80% of the year-to-year variance in the elevation-averaged summer season mass-balance measurements on the glaciers, using a single parameter set for each glacier. The increase in ablation on the glaciers due to a warming of 2° C is predicted to range from about 1 m w.e. year−1 at the highest elevations to about 2.5 m w.e. year−1 at the lowest elevations. Predicted changes in the winter balance (measured between fixed date) are relatively small, except at the lowest elevations on the Icelandic and Norwegian glaciers where the winter balance is significantly reduced. Equilibrium-line altitudes are raised by 200-300 m on the Icelandic and Norwegian glaciers. Except at the highest elevations, the winter balance of the Icelandic and Norwegian glaciers is predicted to decrease even if the warming is accompanied by a 10% increase in the precipitation. No firm evidence of a climate-related variation in the degree-day factors or in the temperature lapse rate on the same glacier could be found. The model, furthermore, reproduces large variations in the mass balance with elevation and from year to year on each glacier using the same parameter set. We assume, therefore, that these parameters will not change significantly for the climate scenarios considered here.
A degree-day model developed for parameterizing melt rates on the Greenland ice sheet is adapted to the climatic conditions on glaciers in southern Norway. The model is calibrated by means of observed average mass-balance-elevation relationships (1963–90) for three glaciers in a west-east transect in southern Norway and 30 year normals (1961–90) of temperature and precipitation observed at nearby climate stations. The calibration gives a surprisingly small variation of the model parameters (degree-day factors for snow-and ice-melt, and precipitation-elevation gradient) from one glacier to another. The derived values of the parameters are used to estimate the change of the mass-balance-elevation relationship for different climatic scenarios. The study indicates that a low-lying glacier in the maritime, high-precipitation environment near the Atlantic coast is more sensitive to both temperature and precipitation changes than the high elevated glaciers in the dry, more continental climate farther away from the coast. However, all of the glaciers studied will lose mass in a warmer climate, unless the warming is accompanied by a dramatic increase in the precipitation of 25–40% deg−1 warming.
Ice ablation is related to air temperature by the positive degree-day factor. Variations of the positive degree-day factor in West Greenland are studied using an energy-balance model to simulate ablation under different conditions. Degree-day factors for simulated and measured ice ablation at Nordbogletscher and Qamanârssûp sermia agree well with values around 8 mm d−1 °C−1. Degree-day factors for snow are less than half those for ice. Energy-balance modelling shows that degree-day factors vary with summer mean temperature, surface albedo and turbulence but there is only evidence of large positive degree-day factors at lower temperatures and with low albedo (0.3). The greatest effect of albedo variations (0.3–0.7) is at lower temperatures while variations in turbulence have greater effect at higher temperatures. Current models may underestimate runoff from the Greenland ice sheet by several tenths because they use a degree-day factor for melting ice that is too small for the colder parts of the ice sheet, i.e. the upper ablation area and the northerly margin.
Lithology, lithic petrology, planktonic foraminiferal abundances, and clastic grain sizes have been determined in a 30 m-long core recovered from the Barra Fan off northwest Scotland. The record extends back to around 45 kyr B.P., with sedimentation rates ranging between 50 and 200 cm/kyr. The abundance of ice-rafted debris indicates 16 glacimarine events, including temporal equivalents to Heinrich events 1–4. Enhanced concentrations of basaltic material derived from the British Tertiary Province suggest that the glacimarine sediments record variations in a glacial source on the Hebrides shelf margin. Glacimarine zones are separated by silty intervals with high planktonic foraminifera concentrations that reflect an interstadial circulation regime in the Rockall Trough. The results suggest that the last British Ice Sheet fluctuated with a periodicity of 2000–3000 years, in common with the Dansgaard-Oeschger climate cycle.
Firn line altitudes of Loch Lomond Advance glaciers in Scotland and the Lake District imply that snowfall was associated mainly with south to southeasterly air streams preceding fronts. Precipitation was high in the south-west Grampians and very low in the north-west Cairngorms and Speyside. The junction of polar and relatively warm ocean waters at the latitude of south-west Ireland was associated with many depressions following more southerly tracks than at present. It is suggested that during ice-sheet growth the zone of maximum snowfall moved southwards over the British Isles.
The theory of the sliding of glaciers presented by the author in earlier papers has been generalized (1) by taking into account the resistance to sliding offered by obstacles both smaller and larger than the controlling obstacles and (2) by relaxing the assumption that ice is always in intimate contact with the bed at the down-stream side of an obstacle. The sliding velocities and controlling obstacle sizes which are found from the generalized theory are approximately the same as those found from the earlier theory. A new result obtained from the present theory is that a water layer an order of magnitude smaller in thickness than the height of the controlling obstacles can cause an appreciable increase in the sliding velocity. The generalized theory contains Lliboutry’s sliding theory as an extreme limiting case. For certain thicknesses of a glacier the sliding velocity is a double-valued function of the shear stress exerted at the bed.
Terrestrial and marine subglacial landforms in eastern Scotland are used to evaluate previously unsubstantiated notions of ice streaming within the British Ice Sheet (BIS) in this area during the last glacial cycle. Employing both regional and local-scale data sets, we describe onshore landform-sediment assemblages, offshore geomorphology and stratigraphy, and reconstructed palaeo-ice flow patterns. The results and their glaciological significance are discussed in the context of stratigraphical and geomorphological frameworks established by earlier workers, and are compared with modelled reconstructions for the BIS in this area. We conclude that the Main Late Devensian ice sheet in eastern Scotland hosted a zone of fast-flowing ice at least 100 km long and 45 km wide, akin to a contemporary ice stream. This sector - the Strathmore Ice Stream - flowed through a combination of basal sliding on meltwater-lubricated rigid beds and by deforming unconsolidated basal substrates.