Search

Over the past decade, Antarctic sea ice extent exhibited a sequence of record maxima, followed by a rapid decline in 2015/16, and record minima since. In this presentation, we show that this sudden and remarkable ice loss marks an abrupt transition from a high to a low ice state that cannot be explained by year-to-year variability. Instead, it is most likely associated with a longer term variability arising from ice–ocean feedbacks. The abrupt transition was preceded by a multi-decadal increase in persistence and variance of the sea ice anomalies, an increasing upper Southern Ocean density stratification, and an accumulation of heat at the subsurface; suggesting a decoupling of the surface from the subsurface ocean. During this period, the sea ice anomalies shifted from being structured predominantly regionally and seasonally to a largely circumpolar and interannual regime. In 2015/16, the upper ocean density stratification in the ice-covered region suddenly weakened, leading to a release of heat from the subsurface, contributing to the sea ice decline during winter. Our analysis suggests that the sudden sea ice loss in 2015/16, and the persisting low ice conditions since, arose from a systematic change in the physical state of the coupled circumpolar ice–ocean system. This change will have wide implications for global climate, ecosystems, and the Antarctic Ice Sheet.

A major source of uncertainty in future sea-level projections is the ocean-driven basal melt of Antarctic ice shelves. Whereas ice sheet models require a kilometer-scale resolution to realistically resolve ice shelf stability and grounding line migration, global or regional 3D ocean models are computationally too expensive to produce basal melt forcing fields at this resolution. To bridge this resolution gap, we introduce the 2D numerical model LADDIE (one-Layer Antarctic model for Dynamical Downscaling of Ice–ocean Exchanges) which allows for the computationally efficient modeling of basal melt rates. The model is flexible, and can be forced with output from coarse 3D ocean models or with vertical profiles of offshore temperature and salinity. In this study, we describe the model equations and numerics. To illustrate and validate the model performance, we apply the model to two test cases: the small Crosson-Dotson Ice Shelf in the warm Amundsen Sea region, and the large Filchner-Ronne Ice Shelf in the cold Weddell Sea. At ice-shelf wide scales, LADDIE reproduces observed patterns of basal melt and freezing that are also well reproduced by 3D ocean models. At scales of 0.5–5 km, which are unresolved by 3D ocean models and poorly constrained by observations, LADDIE produces plausible basal melt patterns. Most significantly, the simulated basal melt patterns are physically consistent with the applied ice shelf topography. These patterns are governed by the topographic steering and Coriolis deflection of meltwater flows, two processes that are poorly represented in basal melt parameterisations. The kilometer-scale melt patterns simulated by LADDIE include enhanced melt rates in basal channels, in some shear margins, and nearby grounding lines. As these regions are critical for ice shelf stability, we conclude that LADDIE can provide detailed basal melt patterns at the essential resolution that ice sheet models require. The physical consistency between the applied geometry and the simulated basal melt fields indicates that LADDIE can play a valuable role in the development of coupled ice–ocean modeling.

Results from CMIP5 have previously suggested that ensemble regression techniques or model selection may provide solutions to the challenge of making projections of future Antarctic sea ice area (SIA) in the presence of large historical biases. Here, we revisit and extend such analysis incorporating the CMIP6 ensemble, which shows modest improvements in some aspects of sea ice simulation and in particular a reduction of inter-model spread in historical SIA. We focus on the strongest forcing scenarios analysed, CMIP5 RCP85 and CMIP6 SSP5.85.In summer (February) the historical climatology of SIA is a strong linear constraint on projections of SIA in both generations. This is because the strong forcing leads to the loss of the majority of summer SIA in each model, so that the models that start with greater SIA exhibit greater reductions. Differences between CMIP5 and CMIP6 are largely explained by the fact that, compared to CMIP6, CMIP5 contains many more models that have very large positive biases in historical SIA and do not lose the majority of ice.In winter (September), a much smaller proportion of SIA is lost, but inter-model spread in SIA climatology still explains just under half the variance in projections of SIA change, in both CMIP5 and CMIP6. The mean historical winter climatology is similar between generations, as is the regression slope of SIA change against SIA climatology. However, there is a greater reduction of SIA in CMIP6 than CMIP5. We find this to be statistically related to greater global mean warming in CMIP6 than CMIP5, and therefore potentially to the larger climate sensitivity in the CMIP6 ensemble.These findings imply that, depending on season, a different balance of local (SIA climatology) and global (GMST change) drivers can be used to explain the inter-model and inter-generation spread in projections of SIA loss. They also firmly tie our ability to project Antarctic SIA loss to our understanding of the fidelity of higher CMIP6 climate sensitivity. Questions remain about whether models are correct in their simulation of Antarctic SIA sensitivity to global surface temperature.

A UKESM climate model which is coupled annually to the BISICLES ice sheet model to enable a two way interactions in Antarctica has been developed and run through a small ensemble of four SSP1-1.9 & SSP5-8.5 scenario members. Under the extreme anthropogenic forcing, all the initial condition ensemble members develop strong melting under the cold & large Ross and Filchner-Ronne ice-shelves, where it starts after the first half of simulation period for the former and in the last decade of the run for the latter. Despite that, during the 85 years timescale of these scenario runs, the stronger radiative forcing has positive effects on the ice-sheet mass gain through increasing precipitation on grounded ice regions which offsets the impact of basal melting in ice discharge across the grounding lines.

In this presentation we describe how models of the Greenland and Antarctic ice sheets have been incorporated in the global U.K. Earth System model (UKESM1) with a two-way coupling that passes fluxes of energy, water and the locations of ice surfaces between the component models. Offline, file-based coupling is used throughout to pass information between the components, which is both physically appropriate and convenient within the UKESM1 structure. Ice sheet surface mass balance is computed in the land surface model using sub-gridscale multi-layer snowpacks. Icebergs calved from the ice sheets are fed into a Langrangian iceberg drift scheme in the ocean. Ice shelf basal melt is explicitly calculated in cavities resolved by the ocean model, and ice sheet and shelf geometries are kept consistent in all components. We demonstrate that our coupled model remains stable when simulating changes in ice sheet height, extent and grounding-line position of hundreds of kilometres.

Antarctic Bottom Water (AABW) supplies the lower limb of the global overturning circulation, ventilates the abyssal ocean and sequesters heat and carbon on multidecadal to millennial timescales. AABW originates on the Antarctic continental shelf, where strong winter cooling and brine released during sea ice formation produce Dense Shelf Water, which sinks to the deep ocean. The salinity, density and volume of AABW have decreased over the last 50 years, with the most marked changes observed in the Ross Sea. These changes have been attributed to increased melting of the Antarctic Ice Sheet. Here we use in situ observations to document a recovery in the salinity, density and thickness (that is, depth range) of AABW formed in the Ross Sea, with properties in 2018–2019 similar to those observed in the 1990s. The recovery was caused by increased sea ice formation on the continental shelf. Increased sea ice formation was triggered by anomalous wind forcing associated with the unusual combination of positive Southern Annular Mode and extreme El Nino conditions between 2015 and 2018. Our study highlights the sensitivity of AABW formation to remote forcing and shows that climate anomalies can drive episodic increases in local sea ice formation that counter the tendency for increased ice-sheet melt to reduce AABW formation. Interacting atmospheric circulation patterns are responsible for a recent reversal of a decades-long decline in deepwater formation on the Antarctic shelf, according to an analysis of in situ and remote sensing data from the Ross Sea.

The contact between ice shelves and relatively warm ocean waters causes basal melt, ice shelf thinning, and ultimately ice sheet mass loss. This basal melt, and its dependence on ocean properties, is poorly understood due to an overall lack of direct observations and a difficulty in explicit simulation of the circulation in sub-shelf cavities. In this study, we compare a number of parameterisations and models of increasing complexity, up to a 2D ‘Layer’ model. Each model is aimed at quantifying basal melt rates as a function of offshore temperature and salinity. We test these models in an idealised setting (ISOMIP+) and in a realistic setting for the Amundsen Sea Embayment. All models show a comparable non-linear sensitivity of ice-shelf average basal melt to ocean warming, indicating a positive feedback between melt and circulation. However, the Layer model is the only one which explicitly resolves the flow direction of the buoyant melt plumes, which is primarily governed by rotation and by the basal topography of the ice shelves. At 500m resolution, this model simulates locally enhanced basal melt near the grounding line, in topographical channels, and near the western boundary. The simulated melt patterns for the Amundsen Sea ice shelves are compared to satellite observations of ice shelf thinning and to 3D numerical simulations of the sub-shelf cavity circulation. As detailed melt rates near the grounding line are essential for the stability of ice sheets, spatially realistic melt rates are crucial for future projections of ice sheet dynamics. We conclude that the Layer model can function as a relatively cheap yet realistic model to downscale 3D ocean simulations of ocean properties to sub-kilometer scale basal melt fields to provide detailed forcing fields to ice sheet models.

In recent years, the ice shelf of Pine Island Glacier has experienced several significant calving events. It is understood that the presence of the ice shelf in conjunction with a subglacial ridge provide a strong topographic barrier to warm Circumpolar Deep Water spilling onto the continental shelf, but it is not known how this barrier will respond to this recent, and possible future, calving events. In this presentation, I shall present results of numerical simulations of ocean circulation under Pine Island Glacier, which indicate a strong sensitivity to such calving events, and discuss the implications of these results for the overall stability of the glacier.

In recent years, there have been great advances in coupled ice sheet-ocean modelling, to the point where ice-ocean interactions can be represented in global climate models — with potential to greatly improve forecasting of marine ice-sheet loss and sea level rise in the coming century and beyond. However, initialisation of coupled ice sheet-ocean models has not yet been properly examined; and initialisation approaches applied to ocean and coupled atmosphere-ocean models may not be appropriate due to the long time scales inherent in dynamic ice sheets. Moreover, as ocean melt rates and ice-shelf geometry strongly influence each other, nonphysical transients in incorrectly initialised coupled ice-ocean models may persist for longer than in ice-sheet models alone.In this work, two approaches to coupled initialisation are considered using a synchronously coupled ice-ocean model. The two approaches are based on two commonly used approaches to ice sheet model initialisation: “snapshot” calibration, where ice-sheet basal and internal parameters are configured to optimise fit with observed surface velocity; and “transient” calibration, where these parameters are configured to jointly optimise fit with velocity and geometry change; however, the transient calibration makes use of the ocean component to ensure the ice model is not subject to “initialisation shock” from ocean melting. The approaches are applied to Smith Glacier, a small but fast-thinning glacier in West Antarctica, and the model is forced under ocean warming scenarios in multidecadal runs. Initially there is much faster retreat seen in the Snapshot-calibrated simulation, but this difference decays over several decades, and ultimately the Transiently-calibrated model sees more retreat.The experiments further suggest that Smith Glacier is not likely to exhibit Marine Ice Sheet instability in the next century. But the methods discrepancy has strong implications for glaciers which are susceptible to this instability.

In this study, satellite-derived observations of sea ice concentration, drift, and thickness are combined to provide a climatology and inter-annual monthly variability of the sea ice volume budget over the growth season for the CryoSat-2 period Octobre 2010 to April 2019. This allows the first wintertime observational decomposition of the dynamic (advection and divergence) and thermodynamic drivers of ice volume change.Dynamic and thermodynamic processes will be separated by applying similar methods to Holland and Kwok (2012), which decomposed the governing equation of ice concentration, C:[1] ∂C/∂t + ∇.(uC) = f_C - rwhere u is ice drift motion, f_C is thermodynamic freezing or melting, and r is the concentration change from mass-conserving mechanical ice redistribution processes which convert ice area to ice thickness, such as ridging and rafting. ∂C/∂t represents ice intensification and ∇.(uC) represents ice flux divergence, the dynamic contribution to ice concentration.Equation 1 can be rearranged and the dynamic contribution, ∇.(uC), expanded to show the contributions from advection, (-u.∇C) and divergence, (-C∇.u), determining four terms:[2] ∂C/∂t = -u.∇C - C∇.u + f - rThe governing equation of the volume budget is of the same form but combines the thickness and concentration data:[3] ∂Ch/∂t = -u.∇Ch - Ch∇.u + f_Chwhere h is the thickness of the ice and the resulting product of the two datasets, Ch, is the effective thickness. If Ch were to be multiplied by the grid cell area this would give the volume, V. It is not necessary to take this step because the area remains constant and does not influence the relative values of the terms. However, when deriving an Arctic-wide climatology spatial integration across the grid cell areas is required. Ridging is taken into account by effective thickness change, therefore, it is not included in the calculations.The method is replicated using model simulations from the Centre for Climate Observation and Modelling (CPOM)-modified Los Alamos sea-ice model (CICE), providing a test of the model’s ability to calculate the volume budgets but also identifying unrealistic growth regimes in the CryoSat-2 observational datasets. Sensitivity to several observational datasets is performed to provide an estimated uncertainty of the budget calculations.The observational results show ice gain in the central Arctic is dominated by ice freezing with contributions from convergence. Divergence at the coastlines of the Arctic form an ice sink where freezing generates new ice. Advection is shown to drive ice equatorward and induce melting at the ice edge where ice becomes thermodynamically unstable. The dynamic components are found to grow in influence throughout the growth season.The 2016/17 winter growth season budget shows reduced thermodynamic intensification and stronger dynamic tendencies which may be in response to thin initial ice and an exceptionally warm winter. Compared to the observed volume budget, the CICE model displays similar patterns of thermodynamic freezing, however, dynamic components in the central Arctic are significantly reduced whilst they are over-amplified at the ice edge.

In this talk I will present preliminary results of direct numerical simulations of ice melting in a turbulent stratified shear flow. The model solves the evolution of the turbulent fluid phase and of the diffusive solid ice phase, due to melting and freezing, in a fully coupled way. This is done by combining a Direct Numerical Simulation (DNS) code with a novel formulation of the equations for the solid and liquid phases of water based on the phase-field method. DNS enables turbulent motions to be simulated without approximation, i.e. solving Navier Stokes equations, while the phase-field method allows the ice-ocean interface to be rough and evolve in response to melting. I will present results on the turbulent boundary layer and on the self-generated basal topography at the ice-water interface. The ultimate goal of this work is to propose a new DNS-based parameterization of the melting process at rough ice-ocean boundaries that takes into account the effects of temperature and salt stratification, and flow velocities.

Ice shelf dynamics play a key role in the climate. Melt-rates along the ice shelf-ocean interface are an important aspect in determining the character of global sea level rise. A representation of ice shelf melt is currently implemented in various z-level General Circulation Models (GCMs) by employing parameterisations of the small scale boundary layer dynamics. However, these parameterisations are strongly dependent on the near boundary flow and at the spatial scales for which GCMs are intended the boundary layer is not well resolved. We investigate the ability of a GCM in representing these small scale boundary effects. This is done using MITgcm in an idealised setting with a sloping ice-ocean interface.

The interaction between Ice-Shelves and the Ocean is an important component of the response of ice sheets to future warming oceans. Observational data in the ocean boundary layer beneath ice shelves is limited and the turbulent flow in the boundary layer is not well characterised. Our work uses small scale (9m depth) direct numerical simulations (DNS) of the Ice-Shelf-Ocean Boundary Layer, inspired by field observations made beneath the George VI Ice Shelf. Here, warm water has been observed directly beneath the ice shelf, and yet the observed melt rates are modest. To study this scenario, we simulate a forced turbulent flow underlying an ice shelf where the ice base is represented by a dynamic melting boundary condition. As the ice melts, a pool of relatively cold, fresh water develops below the ice base. Thermal diffusion causes the underlying water to cool and can drive turbulent convection. At the same time, the salinity gradient in the halocline is stabilising, but develops over a longer time scale. As a result, two flow regimes exist: one with active turbulent convection driven by double-diffusion of heat and salt, and the other with stratified turbulence leading to mixing of the halocline. By varying control parameters, we identify the transition between the flow regimes in terms of the temperature contrast (thermal driving) and the level of turbulence in the far field. We consider the behaviour of the diapycnal buoyancy flux near the ice base, and it provides insight into the drivers of both the double-diffusive convection and its modification by ambient turbulence. Finally, we discuss how the double-diffusive process we have described applies to real-world ice-shelf ocean boundary layers, and how it may be quantified within observations.

In recent decades, rapid ice-shelf disintegration along the Antarctic Peninsula has had a global impact through enhancing outlet glacier flow, and hence sea level rise, and the freshening of Antarctic Bottom Water. Ice shelf thinning due to basal melting results from the circulation of relatively warm water in the underlying ocean cavity. However, the effect of sub-shelf circulation on future ice-shelf stability cannot be predicted accurately with computer simulations if the geometry of the ice-shelf cavity is unknown. To address this deficit for Larsen C Ice Shelf, west Antarctica, we integrate new water-column thickness measurements with existing observations. We present these new data here along with an updated bathymetry grid of the ocean cavity. Key findings include relatively deep seabed to the south-east of the Kenyon Peninsula, along the grounding line and around the key ice shelf pinning point of Bawden Ice Rise. In addition, we can confirm that the cavity’s southern trough stretches from Mobiloil Inlet to the open ocean. These areas of deep seabed will influence ocean circulation and tidal mixing, and will therefore affect the basal-melt distribution. These results will help constrain models of ice-shelf cavity circulation with the aim of improving our understanding of sub-shelf processes and their potential influence on ice shelf stability. The data set comprises all point measurements of seabed depth and a gridded data product, derived using additional measurements of both offshore seabed depth and the thickness of grounded ice. We present all new depth measurements here as well as a compilation of previously published measurements used in the gridding process. The gridded data product is included in the supplementary material. The underlying seismic data sets which were used to determine bed depth and ice thickness are available at https://doi.org/10.5285/315740B1-A7B9-4CF0-9521-86F046E33E9A (Brisbourne et al., 2019), https://doi.org/10.5285/5D63777D-B375-4791-918F-9A5527093298 (Booth, 2019), https://doi.org/10.5285/FFF8AFEE-4978-495E-9210-120872983A8D (Kulessa and Bevan, 2019) and https://doi.org/10.5285/147BAF64-B9AF-4A97-8091-26AEC0D3C0BB (Booth et al., 2019).

Persistent outflow of supercooled Ice Shelf Water (ISW) from beneath McMurdo Ice Shelf creates a sub-ice platelet layer (SIPL) having a unique crystallographic structure under the sea ice in McMurdo Sound (MMS), Antarctica. A new frazil-ice-laden ISW plume model that encapsulates the combined nonlinear effects of the vertical distributions of supercooling and frazil ice concentration (FIC) on frazil ice growth is applied to MMS, and is shown to reproduce the observed ISW supercooling and SIPL distributions. Using this model, the dependence of SIPL thickening rate on ISW supercooling in MMS is investigated. Results are found to be sensitive to the choice of frazil ice suspension index, which determines the vertical distribution of FIC. For each suspension index, SIPL thickening rate can be expressed as an exponential function of ISW supercooling. The complex dependence on FIC highlights the need to improve frazil ice observations within the ice-ocean boundary layer.

2016 was an interesting year in the Arctic, with record low sea ice at the start of the year, but a summer (September) Arctic sea ice extent that was higher than expected by most seasonal forecasts. Here we explore the 2016 Arctic sea ice state in terms of its monthly sea ice cover, placing this in context of the sea ice conditions observed since 2000. We demonstrate the sensitivity of monthly Arctic sea ice extent and area estimates, in terms of their magnitude and annual rankings, to the ice concentration input data (using two widely used datasets) and to the methodology used to convert concentration to extent (daily or monthly extent calculations). We use estimates of sea ice area to analyse the relative 'compactness' of the Arctic sea ice cover, highlighting anomalously low compactness in the summer of 2016 which contributed to the higher than expected September ice extent. Two cyclones that entered the Arctic Ocean during August appear to have driven this low concentration/compactness ice cover, but were not sufficient to cause more widespread melt out and a new record low September ice extent. We use concentration budgets to explore the regions and processes (thermodynamics/dynamics) contributing to the monthly 2016 extent/area estimates highlighting, amongst other things, rapid ice intensification across the central eastern Arctic through September. Two different products show significant early melt onset across the Arctic Ocean in 2016, including record early melt onset in the North Atlantic sector of the Arctic. Our results also show record late 2016 freeze up in the Central Arctic, North Atlantic. and the Alaskan Arctic sector in particular, associated with strong sea surface temperature anomalies that appeared shortly after the 2016 minimum (October onwards). We explore the implications of this low summer ice compactness for seasonal forecasting, suggesting that sea ice area could be a more reliable metric to forecast in this more seasonal, 'New Arctic', sea ice regime.

Coupled ice sheet-ocean models capable of simulating moving grounding lines are just becoming available. Such models have a broad range of potential applications in studying the dynamics of marine ice sheets and tidewater glaciers, from process studies to future projections of ice mass loss and sea level rise. The Marine Ice Sheet-Ocean Model Intercomparison Project (MISOMIP) is a community effort aimed at designing and coordinating a series of model intercomparison projects (MIPs) for model evaluation in idealized setups, model verification based on observations, and future projections for key regions in the West Antarctic Ice Sheet (WAIS). Here we describe computational experiments constituting three interrelated MIPs for marine ice sheet models and regional ocean circulation models incorporating ice shelf cavities. These consist of ice sheet experiments under the Marine Ice Sheet MIP third phase (MISMIP+), ocean experiments under the ice shelf-ocean MIP second phase (ISOMIP+) and coupled ice sheet-ocean experiments under the MISOMIP first phase (MISOMIP1). All three MIPs use a shared domain with idealized bedrock topography and forcing, allowing the coupled simulations (MISOMIP1) to be compared directly to the individual component simulations (MISMIP+ and ISOMIP+). The experiments, which have qualitative similarities to Pine Island Glacier Ice Shelf and the adjacent region of the Amundsen Sea, are designed to explore the effects of changes in ocean conditions, specifically the temperature at depth, on basal melting and ice dynamics. In future work, differences between model results will form the basis for evaluation of the participating models.

The catastrophic collapses of Larsen A and B ice shelves on the eastern Antarctic Peninsula have caused their tributary glaciers to accelerate, contributing to sea-level rise and freshening the Antarctic Bottom Water formed nearby. The surface of Larsen C Ice Shelf (LCIS), the largest ice shelf on the peninsula, is lowering. This could be caused by unbalanced ocean melting (ice loss) or enhanced firn melting and compaction (englacial air loss). Using a novel method to analyse eight radar surveys, this study derives separate estimates of ice and air thickness changes during a 15 year period. The uncertainties are considerable, but the primary estimate is that the surveyed lowering (0.066 ± 0.017 m yr−1) is caused by both ice loss (0.28 ± 0.18 m yr−1) and firn air loss (0.037 ± 0.026 m yr−1). Though the ice loss is much larger, ice and air loss contribute approximately equally to the lowering. The ice loss could be explained by high basal melting and/or ice divergence, and the air loss by low surface accumulation or high surface melting and/or compaction. The primary estimate therefore requires that at least two forcings caused the surveyed lowering. Mechanisms are discussed by which LCIS stability could be compromised in future, suggesting destabilisation timescales of a few centuries. The most rapid pathways to collapse are offered by a flow perturbation arising from the ungrounding of LCIS from Bawden Ice Rise, or ice-front retreat past a "compressive arch" in strain rates.

An ocean mixed layer model has been incorporated into the Los Alamos sea ice model CICE to investigate regional variations in the surface-driven formation of Antarctic shelf waters. This model captures well the expected sea ice thickness distribution, and produces deep (> 500 m) mixed layers in the Weddell and Ross shelf seas each winter. This results in the complete destratification of the water column in deep southern coastal regions (leading to HSSW formation) and also in some shallower regions (no HSSW formation) of these seas. Shallower mixed layers are produced in the Amundsen and Bellingshausen seas. By deconstructing the surface power input to the mixed layer, we show that the freshwater flux from sea ice growth/melt dominates the evolution of the mixed layer in all seas, with a smaller contribution from the surface heat flux. The Weddell and Ross shelf seas receive an annual surplus of energy at the surface, the Amundsen shelf sea energy input in autumn/winter is balanced by energy extraction in spring/summer, and the Bellingshausen shelf sea experiences an annual surface energy deficit, through both a low energy input in autumn/winter and the highest energy loss in spring/summer. An analysis of the sea ice mass balance demonstrates the contrasting mean ice growth, melt and export in each region. The Weddell and Ross shelf seas have the highest annual ice growth, with a large fraction exported northwards each year, whereas the Bellingshausen shelf sea experiences the highest annual ice melt, driven by the advection of ice from the northeast. A linear regression analysis is performed to determine the temporal and spatial correlations between the autumn/winter mixed layer power input and several atmospheric variables. The temporal mean Weddell and Ross autumn/winter power input shows stronger spatial correlation to several atmospheric variables compared to the Amundsen and Bellingshausen. In contrast the spatial mean autumn/winter power input shows stronger temporal correlation to several atmospheric variables, in the Amundsen and Bellingshausen. All regions show strong temporal correlation between the autumn/winter surface power input and the meridional wind speed except the Ross, which instead shows moderate correlation to the zonal wind speed. Further regressions demonstrate that this is probably due to the Ross shelf-sea geometry and impact of the ocean turning angle on ice motion, with a more zonal (eastward) wind preventing ice build up along the Cape Adare coast in the eastern Ross shelf sea, increasing ice export.