IAPSO Best Practice Research Group: REconciling Cross-platform Observations of Ice-sheLf melt (RECOIL)

IAPSO Best Practice Research Group: REconciling Cross-platform Observations of Ice-sheLf melt (RECOIL) Dan Goldberg1, Andrew Meijers2, Elizabeth Case3, Sue Cook4, Peter Davis2, Pierre Dutrieux2, Noel Gourmelen1, Chad A. Greene5, Tore Hatterman6,7,8, Patrick Heimbach9,10, Karen J. Heywood11, Won Sang Lee12, Luc Lenain13, Katie Lowery2,14, Ken Mankoff15,16, Felicity McCormack17, Yoshihiro Nakayama18,19, Keith Nicholls2, Madelaine Rosevear20,4, Alessandro Silvano21, and Anna Wahlin22 1School of GeoSciences, University of Edinburgh, UK 2British Antarctic Survey, Cambridge, UK 3Institute for Marine and Atmospheric research Utrecht (IMAU), Utrecht University, Netherlands 4Australian Centre for Excellence in Antarctic Science, University of Tasmania, AU 5Jet Propulsion Laboratory, California Institute of Technology, Pasadena, USA 6Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research, Bremerhaven, Germany 7Akvaplan-niva AS, Tromsø, Norway 8Norwegian Polar Institute, Tromsø, Norway 9Oden Institute for Computational Engineering and Sciences, U. Texas-Austin, USA 10Jackson School of Geosciences, U. Texas-Austin, USA 11School of Environmental Sciences, University of East Anglia, Norwich, UK 12Division of Glacier & Earth Sciences, Korea Polar Research Institute, Incheon, Republic of Korea 13Scripps Institution of Oceanography, U. California. USA 14School of Earth and Environment, University of Leeds, Leeds, UK 15NASA Goddard Institute for Space Studies, New York,USA 16Autonomic Integra LLC, New York, USA 17School of Earth, Atmosphere and Environment, Monash University, AU 18Thayer School of Engineering, Dartmouth College, Hanover, NH, USA 19Institute of Low Temperature Science, Hokkaido University, Sapporo, Japan 20Department of Mechanical Engineering, University of Melbourne, AU 21School of Ocean and Earth Science, University of Southampton. UK 22Department of Marine Sciences, University of Gothenburg, Sweden Abstract One of the principal factors influencing mass loss and sea level contribution from the Antarctic Ice Sheet currently and in future centuries is the melting of ice shelves by the ocean. These floating extensions of the ice sheet buttress against ice-sheet movement, and melt-driven ice-shelf thinning can potentially lead to rapid ice-sheet loss. In addition, the meltwater from the ice shelves influences sea ice formation as well as ocean density stratification and circulation, and can trigger critical feedbacks in the global carbon and energy cycles. However, ice shelf-ocean interactions are not well understood or well quantified. Melt (and refreeze) rates depend on processes that cover a broad range of scales, from the centimeter-scale ice-shelf boundary layer processes to the large-scale circulation within ice-shelf cavities and beyond. Both remain incredibly challenging to observe, limiting our ability to derive and validate model parameterizations of, and constrain simulations of, submarine melt. A range of methodologies are employed to measure ice-shelf melt rates, broadly grouped as: (1) Satellite-based – using satellite altimetry and velocity together with estimates of surface melt and compaction to estimate broad melt patterns; (2) Oceanographic – quantifying spatially-integrated (bulk) melt via tracer and heat balance; and (3) Geophysical – high fidelity melt estimation through direct observations of the ice-shelf base at discrete locations. Each method differs in spatial and temporal extent and resolution, and each has limitations. Where melt estimates overlap, there is potential to independently validate estimates. However, this requires both a careful consideration of the methods used in data collection as well as a standard basis for comparison – considerations which, to date, have been lacking. Thus opportunities exist in terms of planned and future campaigns to ensure that measurements are as comparable as possible. Here we propose an outline framework for ensuring such intercomparability. We make two “levels” of recommendation. In the first (low-cost), we make a set of recommendations by which to evaluate and plan proposed field campaigns to maximise the degree to which melt measurements can be compared. In the second (high cost), we propose a bold “ice-shelf laboratory” concept in which a single ice shelf can be instrumented as much as possible and biases in large-scale and low-cost measurement platforms can be better understood. We further propose several candidate ice shelves for such “supersites”. We intend that this framework will inform the community in the planning stages for future internationally-led field programmes. 2 Contents Abstract 2 1 Introduction 5 2 Workshop: Aims and Format 9 3 Current Methodologies: Overview and Challenges 11 3.1 ApRES and in-situ radar . . . . . . . . . . . . . . . . . . . . . . . 11 3.2 Satellite based methods . . . . . . . . . . . . . . . . . . . . . . . . 12 3.2.1 Measuring elevation and elevation change . . . . . . . . . . 13 3.2.2 Assumptions.......................... 13 3.2.3 Surface processes . . . . . . . . . . . . . . . . . . . . . . . 14 3.3 Oceanographic-based methods . . . . . . . . . . . . . . . . . . . . 14 3.4 Potential sources of discrepancy . . . . . . . . . . . . . . . . . . . 15 4 The role of modelling 19 4.1 Ice-sheetmodels............................ 19 4.2 Global and regional ocean models . . . . . . . . . . . . . . . . . . 20 4.3 Stateestimates ............................ 21 5 Critical Regions 22 5.1 CriticalRegimes ........................... 22 5.1.1 Grounding Zones . . . . . . . . . . . . . . . . . . . . . . . 22 5.1.2 PinningPoints ........................ 23 5.1.3 ShearZones.......................... 23 5.1.4 Regions with Potential Regime Shifts . . . . . . . . . . . . 24 5.1.5 Areas of particularly high melt rates or potential glacial instability ........................... 24 5.2 Importance of diverse Ice-Ocean boundary conditions . . . . . . . 25 6 Leveraging Existing Data and Models 27 6.1 Remote Sensing ice-shelf melt products . . . . . . . . . . . . . . . 27 6.2 Oceanographic-based meltwater flux estimates . . . . . . . . . . . 27 6.3 ApRES/radar based estimates . . . . . . . . . . . . . . . . . . . . 28 6.4 Modelling and assimilation products . . . . . . . . . . . . . . . . 28 3 7 A Framework for reconciling estimates of ice-shelf melt 32 7.1 Framework Aims, Themes, and Values . . . . . . . . . . . . . . . 32 7.2 What Can We Do Now? Leveraging Existing Data and Networks 32 7.3 What Could We Do? A Collaborative and Integrative Approach . 34 7.3.1 Observations ......................... 34 7.3.2 Modeling ........................... 35 7.3.3 Finalthoughts ........................ 36 4 Chapter 1 Introduction The mass of ice held in the Antarctic Ice Sheet depends on its sources and sinks: the processes adding and removing mass, respectively. The former is predominantly due to snowfall on its surface, while the latter is mainly due to iceberg calving and ocean-driven melting at the base of ice shelves – the floating extensions of the ice sheet. While losses due to calving and submarine melt, respectively, are close in magnitude at a continental scale – O(103) Gt/a each are lost to calving and melting, respectively [Depoorter et al., 2013] – the balance between the two, and the degree, patterns, and regime of melting within each ice shelf, have high levels of variability among Antarctic ice shelves. While some ice shelves are exposed to relatively warm waters (3-4◦C above the in-situ melting point) and lose thickness due to melt at rates on the order of 100 m/yr or greater [Milillo et al., 2019, Gourmelen et al., 2025], other ice-shelf cavities currently contain waters that are cooled by atmospheric processes and which drive little melt and even cause freeze-on [Joughin and Vaughan, 2004]. There are a number of reasons why it is vital to constrain and understand sub-ice shelf melt rates at both an ice-shelf and a continental scale. One is understanding and forecasting sea level rise. Ice shelves play a strong role in the force balance of ice sheets, acting as anchors against acceleration of glacier flow into the ocean (often termed “buttressing” ). Thinning and removal of mass can reduce buttressing and lead to heightened losses from the ice sheet, as seen on the Antarctic Peninsula following the collapse of Larsen B ice shelf [Scambos et al., 2004]. Within certain sectors of Antarctica, acceleration could lead to retreat of grounding lines (i.e. the locations where the ice sheet goes afloat), potentially triggering an instability whereby losses accelerate in the presence of inland-deepening bedrock [Weertman, 1974]. Another impact of melt is the addition of freshwater to the Southern Ocean. A shift in the large-scale balance and transport of freshwater could cause changes in sea-ice distribution [Richardson et al., 2005, Bronselaer et al., 2018], affecting the heat uptake of the Southern Ocean, as well as changes in production of Antarctic Bottom Water, the upwelling of Circumpolar Deep Water or sinking of Mode and Intermediate waters - all of which contribute to the global overturning circulation and associated heat and carbon storage by the ocean. Climate model studies suggest strengthening freshwater inputs could have measurable effects on global ocean circulation and climate before the turn of the century [Li et al., 2023]. Ad5 ditionally, ice-shelf melt processes contribute to recirculation of nutrients that are important to biological processes on continental shelves. For example, iron cycling, a key component in spring and summer blooms of carbon-fixing phytoplankton, has been shown to depend critically on ice-shelf melt in certain regions [Gerringa et al., 2012, Herraiz-Borreguero et al., 2016]. In order to better understand and anticipate future sea-level rise and climate change, as well as how our actions will affect them, it is vital that we improve the representation of ice-shelf melt in climate models. However, accurate modelling of ice shelves remains extremely challenging. Ice-shelf melting is a process that occurs over a range of scales. Melt is driven by heat and salt transfer across the ice-ocean boundary layer, which is tens of centimeters to meters thick [Rosevear et al., 2021]. The ocean properties that drive it are governed, however, by circulation within the ice-shelf cavity which evolves and transitions at the kilometer scale [Jenkins and Doake, 1991]. The circulation within the cavity is determined, in turn, by regional currents and hydrography on (and off) the continental shelf [Nakayama et al., 2019] and its associated atmospheric forcings. In order to model melt rates as the ice sheet and ice shelves evolve in future decades, it is critical that the relevant processes are sufficiently represented in our models. The Antarctic under-ice shelf environment is extremely observation-poor, however [Nicholls, 2018], making it challenging to provide the key information to enable accurate process representation and parameterisation, and by which models can be validated and improved. Thus, any methods by which spatiotemporal detail of ice-shelf melt can be obtained should be evaluated in order to optimise future in situ observational campaigns. At the same time, the inherent limitations, errors and inconsistencies in such measurements should be understood and, if possible, minimized. Currently there are three main types, or categories, of methods: •In-Situ Geophysical observations. This refers in general to a range of on-ice or under-ice platforms (see Chapter 3.1 below) but for the purposes of this discussion we mainly refer to the use of localised phase-sensitive radar (e.g. ApRES) to determine changes in ice-shelf bottom elevation due to melt. •Satellite. This class of methods refers to a mass conservation approach where satellite-provided observations of ice-shelf elevation change and velocity are combined with estimates of surface mass balance and firn processes in order to generate a spatially-resolved map of sub-ice shelf melt. •Oceanographic. Here, we refer specifically to the “flux gate” method which involves characterizing balances of heat, salt and mass (and potentially chemical tracers) across the ocean-facing entrance of an ice-shelf cavity in order to quantify the generation of freshwater (and hence melt within the cavity). A secondary, yet important, purpose of oceanographic measurements is to provide estimates of the circulation (and thus heat flux) that directly affects melt rate variability. Each has advantages and drawbacks. ApRES provides an accurate measurement of melt, as well as internal ice dynamics such as divergence, shear etc, on 6 time scales ranging from annual down to tidal [Vaˇnkov´a et al., 2020]. However, it is spatially limited, its source of uncertainty includes time-dependent strain rates and areas of complex basal topography, and is limited to regions without marine ice (seawater frozen onto the ice shelf base). Satellite methods convert time-dependent observations of the ice surface and ice velocity to thickness and melt [Adusumilli et al., 2020, Paolo et al., 2023, Gourmelen et al., 2025]. They have comparatively high spatial coverage, but are subject to errors in measurement, as well as uncertainty in surface processes represented by climate models. They are also subject to assumption of hydrostatic balance, which does not hold in grounding zones and narrow channels; as well as assumptions of continuity, which may not hold on shear margins and crevasse fields. Oceanographic methods derive spatially-integrated melt of an ice shelf [Jacobs et al., 1996, Heywood et al., 2016, Jenkins et al., 2018]– encompassing areas that are problematic for both satellite methods and ApRES. This method requires a closed volume budget and an ice shelf where all oceanographic fluxes into and out of the cavity can be sampled (see 3.3). The high density of sustained instrumentation therefore required means that it is not logistically feasible for long term observations (or even year-round). We acknowledge here that the above categorization may not be comprehensive. For instance, airborne radar has been used to provide a melt rate estimate [Khazendar et al., 2016]. Additionally, deployment of upward looking sonar is a promising approach to direct quantification of melt [Stewart et al., 2019, Rosevear et al., 2022]. However our grouping is representative of most of the studies used to date. A key challenge to the use of melt rate estimates in characterizing and projecting future ice-shelf buttressing and freshwater flux is the lack of concordance among measurements. Due in part to the differing spatiotemporal scales of measurement, there are few comparisons across these different observation platforms. While some attempts at comparison have recently been made between ApRES and satellite [Vaˇnkov´a and Nicholls, 2022], there is mixed agreement. Moreover the comparisons are somewhat limited, both spatially and in terms of the iceshelf melt regime considered; they also raise questions regarding the timescale over which comparison is appropriate. More investigation is needed regarding the sources of disagreement. Even among different satellite-based methods, scope for intercomparison is limited due to processing approaches. The lack of formal comparison and validation is then problematic for modelling of ice-ocean interactions: different groups may choose to validate models with data from different sources in different ways, with implications for any projections made with those models. We consider the above discrepancies and information gaps to be a major barrier to improving understanding and to making reliable projections of Antarctic freshwater discharge and sea level rise. In this paper, arising from a workshop held in 2024 (see below), we give a brief overview on our present understanding of Antarctic ice-shelf melt, and current state-of-the-art methods for measuring it. For each of the principal methods we discuss the methodologies, assumptions, and potential sources of error. We also review existing melt-rate estimates, highlight inconsistencies, and consider spatial and temporal overlaps that could be exploited for intercomparison. We then present a formal guidance to be implemented when carrying out 7 either observation or modelling of ice-ocean interaction. This guidance includes standards of operational parameters, time scales and variables of reporting, and a data sharing approach that will enable improved comparison of techniques and outcomes. Where possible, we provide or point to data products that will be useful in following such guidance; otherwise, we call for the creation of such products. Furthermore, we make recommendations for a unique observing system that, if implemented, will allow for cross-validation and reconciliation of measurements of melt rates across observational platforms, with a particular empahasis on satellite observations. The deliverable, if successful, will be greatly strengthened trust in satellite-derived observations of melt. Due to the large spatial scale of satellite measurement, and significant existing body of historical data that may be subject to recalibration or revision, this will be an incredibly powerful result, and a step forward in observing, modelling and predicting Antarctic ice loss and freshening of the Southern Ocean. 8 Chapter 2 Workshop: Aims and Format In 2024, a Best Practice Study Group was formed under the auspices of the International Association for the Physical Sciences of the Oceans (IAPSO), which included funds to host a workshop. It was decided to hold the workshop jointly with the Horizon Europe funded OCEAN ICE project (https://ocean-ice.eu) 2024 workshop in Copenhagen at the Danish Meteorological Institute. This entire document represents the outcomes of this workshop. After extensive debate by the organisers, a list of invitees was generated in order to best represent expertise in the methods and scientific areas discussed. 23 attendees were confirmed but several had to drop out due to health issues or other obligations. The workshop itself was 2.5 days, with one day devoted to scientific “scene setting” talks meant to introduce the key scientific questions, key methodologies, and important issues and concerns. The latter 1.5 days consisted of structured breakout groups. For these groups a strict methodology was adopted from a 2024 meeting on “Observing the Dynamics of the Southern Ocean” held at Caltech. At the start of each session, a question was posed to all groups, and each group then implemented the following: 1. Individual brainstorming: Everyone silently writes down ideas. Each idea goes on a separate note. (10 minutes). 2. Sharing ideas: People take turns sharing the ideas they’ve written and posting them to the group space. (15-20 minutes) 3. Grouping or Clustering: Working together, the break-out team makes groups of similar ideas or concepts. Each group then gets named with a single word or short phrase that best captures the core concept underlying the ideas in that group. (5-10 minutes) 4. Minuted discussion within each breakout group (5 minutes) The breakout questions were as follows: •What are the key science challenges regarding Antarctic ice-ocean interaction, and which variables will address them? 9 For instance, in oceanographic melt-rate calculations, the need to close the volume budget for an ice shelf in the face of uncertainty is a problematic one, with potential disagreement among methods in terms of how this is done. Indeed it may be impossible if there is a net volume flux out of a cavity or across a hydrographic section in front of an ice shelf. The choice of tracers used to close the budget, and in how to find the optimal set of reference velocities, leaves potential for errors and disagreement. These are relatively straightforward standards to establish, however – as are the processes by which chemical and physical tracers are measured. Such methods are also limited to ice shelves where the entirety of the ice shelf front can be easily sampled; this is not always possible where ice shelves have multiple entrances or exits, or where access to ice shelves is prevented by fast ice, icebergs or shallow (or unknown) bathymetry. A major additional source of uncertainty is the determination of meltwater content, the numerical value of which depends critically on a choice of end members, and the assumptions made – for example, how to deal with the region of the water column above the base of the winter mixed layer (the Winter Water core in summer) as the water properties in this layer will be influenced by atmospheric processes (solar warming, sea ice formation or melt, exchange of gases with the atmosphere) as well as the addition of meltwater [Biddle et al., 2019, Zheng et al., 2021]. Ideally, unambiguous tracers such as noble gases can be used [Huhn et al., 2018, Biddle et al., 2019]. In satellite processing, there are various choices to be made regarding data sets used, and algorithms employed, and details can vary among approaches: for instance, different studies have used different time windows for their velocity products [Adusumilli et al., 2020, Gourmelen et al., 2017], which impacts the contribution of velocity divergence (which can be large close to the grounding zone) to thickness change. Similarly, there are no strict standards on how to implement advective change (i.e. translating Eulerian to Lagrangian elevation change). The spatial resolution of a basal melt product is largely determined by that of its input data and can also lead to discrepancies between different products : altimetric resolution is a few 100 meters for altimetry, while surface processes are usually derived at kilometric resolution. The Lagrangian scheme effectively smooths melt over the baseline period of transport, and such smoothing depends on input choices [Pritchard et al., 2012]. Surface mass balance and firn densification introduce further error and disagreement – as discussed above, there are extensive uncertainties in the models describing these processes, in part due to lack of detailed observations. The associated errors in melt are potentially large due to the amplification via hydrostatic inversion. As each study calculating satellite derived melt rates uses slightly different datasets, implement different methodologies and computes melt rates at different spatial and temporal resolutions, untangling the drivers of differences between these datasets is near impossible. In order to compare these estimates with those from other observing methods, we must understand how each of our methodological and data choices affects the result. While finding agreement within a single basal melt rate estimation method is challenging, it is even more difficult to assess results from different platforms against one another. ApRES, while relatively highly accurate, is expensive to deploy over large expanses, and satellite methods suffer from non-hydrostasy 16 close to the grounding line. (The severity of the latter depends on how critical it is to establish grounding-zone melt, see Chapter 5.1.1.) As such, total freshwater flux can be challenging to constrain through either method, limiting comparison with oceanographic methods. Timescales are inconsistent as well: high-resolution remote-sensing melt products typically have low temporal resolution, making comparisons with ApRES and oceanographic methods difficult. Oceanographic measurements integrate the melt of the ice shelf in both time and space, and particularly for large ice shelves melt water residence time may be measured in months or years. This necessarily limits the spatial and temporal scale over which comparisons must be made to similar low temporal resolutions, whilst long term oceanographic observations of necessary resolution are prohibitively expensive and logistically challenging. A key priority for cross-platform comparisons are identifying the correct locations, spatial scales, and time windows for comparison (Fig. 3.1). The lack of coherence across products and across methods is problematic for modelling approaches which make use of melt observations (see Chapter 4). Models of ice-ocean interactions must calibrate, or tune, certain parameters based on fit to observations, in order to provide future projections under climate change. There is no clear standard on how to do this: an ice-ocean model which agrees well with observed shelf-wide meltwater flux from a certain ice shelf may be at odds with remotely-sensed patterns, and vice versa; and currently there is no consensus on which aspect of ice-shelf melt rates is more important to reproduce. Furthermore, there is no agreed set of metrics against which to assess the models. As a result, large uncertainties and disagreement could be injected into projections of ice-sheet loss and melt production under climate change scenarios. 17 Figure 3.1: A schematic showing the wide range of spatial and temporal scales of ice-shelf melt that is detected by different platforms. AUV = Autonomous Underwater Vehicle; ApRES = Autonomous phase-sensitive Radio Echo Sounder; AWS = Automatic Weather Station; SMB = Surface Mass Balance; CTD = Conductivity and Temperature Depth Profiler. Surface/atmospheric variables are in green; oceanographic are in blue; satellite-based are in gray. The short temporal range of satellite measurements is to encompass tidal fluctuations of ice shelves e.g. [Rignot et al., 2024]. 18 Chapter 4 The role of modelling 4.1 Ice-sheet models Due to the role of ice-shelf melt in the loss of ice-shelf buttressing, melting processes can be a strong driver of ice-sheet loss and sea level contribution. Ice-sheet models, therefore, are a useful tool in evaluating the impacts of changes in (or revised assessments of) ice-shelf melt. Arguably, such models are also the only way to conduct future projections based on our best current physical understanding (and simulation capabilities). In the last decade or so, research has shown that melting forced by warming oceans could strongly influence Antarctic sea-level contribution in coming centuries [Cornford et al., 2015, Seroussi et al., 2024]. However, it has also shown that the location of melt (or melt increase) can be as important as the magnitude [Walker et al., 2008, Little et al., 2012, Reese et al., 2018, Goldberg et al., 2019, Morlighem et al., 2021]. Greater impacts on buttressing are seen from melt that occurs: •at greater depths; •closer to the grounding line; •in regions of strongly shearing ice flow. The tools used to investigate oceanic impacts on ice sheets range from onedimensional (i.e., flowline), idealised ice-sheet models to realistic models of the Antarctic margin. A powerful method has arisen in the form of adjoint models of ice-sheet dynamics, allowing grid-scale assessment of the sensitivity of a given metric to changes in the model boundary or initial conditions, or to changes in model parameters [Goldberg and Heimbach, 2013, Goldberg et al., 2019, Morlighem et al., 2021]. Despite being linear approximations, such methods can effectively identify areas (and ice shelves) where change in melt is most impactful. One of the models mentioned above, the ice-sheet component of the Massachusetts Institute of Technology general circulation model (MITgcm), has been coupled to the ocean model [Jordan et al., 2018]. Because both the ocean and ice sheet components are differentiable, this raises the possibility of coupled ice-ocean parameter calibration, state estimation, and initialization for prediction (see Chapter 19 4.3). In all cases, the coupled differentiable system assimilates both oceanographic and glaciological constraints. Such efforts are yet to be investigated. Ice-sheet models have the potential for assessing impacts not only of spatial but temporal variability. The modelling study of [Snow et al., 2017] showed that the degree to which fluctuations in ocean temperature impact grounded ice depends on the time scale of the fluctuations. This variability in response across timescales is important as many parts of the Antarctic margin melt rates exhibit temporal variability, with melt varying on seasonal, interannual, and decadal scales [Dutrieux et al., 2014, Jenkins et al., 2018]. 4.2 Global and regional ocean models In global and large-scale regional simulations with coarse grid resolutions, most studies focus on reproducing integrated ice shelf melt rates across all ice shelves (e.g., [Timmermann and Hellmer, 2013, Kusahara and Hasumi, 2013, Schodlok et al., 2016]). These models typically treat ice shelf melting as a boundary condition, represented as a freshwater flux. Heat and salt transfer coefficients are adjusted to achieve agreement with observations in terms of total Antarctic or individual integrated ice shelf melt rates (e.g., [Nakayama et al., 2017]). In essence, these adjustments target the parameterization of processes that act as boundary conditions in these coarse-resolution models. Significant efforts are being put into assessing how the spatial, including depth, distribution of these freshwater fluxes across the ice-ocean boundary influences subsequent ocean properties and circulation (e.g. SOFIA, https://sofiamip.github.io/). Given the spatial inhomogeneity of ice shelf melt, and many Southern Ocean circulation features (eg. bottom water formation, sea ice distribution), the way in which these features are represented is likely to have a significant impact on the fidelity of global ocean modelling. Regional simulations with finer grid resolutions capable of resolving ice shelf cavity processes demonstrate greater skill in simulating ocean hydrography and ice shelf melting dynamics within the cavities (e.g., [Gwyther et al., 2023, Holland et al., 2023]). For example, these models can reproduce peaks in ice shelf melting near the grounding line. Such models are evaluated using various metrics, including the spatial distribution of ice shelf melt rates, cavity hydrography, temporal variability in melt rates, and the physical processes controlling ice shelf melting [Nakayama et al., 2019]. For instance, thermocline depth is known to play a critical role in determining ice shelf melting. Some studies assess whether their simulations accurately represent the influence of thermocline depth variability on ice shelf melting [Park et al., 2024]. One of the challenges with developing and improving models of ice shelf - ocean interaction is poorly known bathymetry of the sea bed. The bathymetry influences both the amount of turbulent mixing experienced by warm inflows on their way to the grounding zone, and also the inflow pathways within the cavity through topographic steering [Richter et al., 2025]. Lack of knowledge of bathymetry also leads to biases and uncertainties in the tidal currents within ice shelf cavities, which have been shown to influence basal melting [Harrison et al., 2022]. 20 [Goldberg et al., 2020] used an ocean model adjoint to investigate the sensitivity of melt to underand near-ice shelf bathymetry, showing that sensitivity is limited to distinct features such as sills and ridges. Further investigations of this type could help to target locations for future seabed mapping efforts. 4.3 State estimates The interior of the global ocean is largely undersampled, and its scale and relative inaccessibility means that this issue will not be addressed in the near future [Stammer et al., 2000]. In order to constrain important properties at depth, data-constrained modelling, or assimilation, has been used extensively to study ocean phenomena. State estimates are a specific class of assimilation methods which find the forcings, boundary conditions, and physical parameters that yield an optimal misfit with observations in an ocean model [Lee et al., 2009, Stammer et al., 2016]. Importantly, alternative leading methods of assimilation, such as Kalman-type approaches or 3DVar [Lellouche et al., 2013], introduce implicit, nonphysical sources and sinks of heat and salt (see [Wunsch et al., 2023] for a tutorial discussion of the issues). By updating the full model trajectory to agree optimally with observations, state estimates provide an internally physically consistent product. This quality makes state estimates ideal for studying global phenomena such as overturning and heat and energy budgets, and for initial states to test the impact of perturbations in climate forcing or carry out projections. However, four-dimensional ocean state estimates are more computationally expensive and complex than other assimilation frameworks, and require the use of an adjoint model to find an optimal state [Lee et al., 2009]. As a result, only a handful of state estimation systems have been developed [Wunsch et al., 2009, Heimbach et al., 2019]. With improved process representation and observational time series that begin to target seasonal to decadal variability, the concept of dynamically and kinematically consistent parameter and state estimation is also gaining traction in the ice sheet modelling community in the form of transient calibration [Goldberg et al., 2015, Badgeley et al., 2025]. In the context of ice-shelf melt, state estimates do not require direct measurements of state variables. This is beneficial because hydrographic measurements under-ice are sparse or nonexistent, while boundary conditions (e.g. at the ice-shelf front) can provide estimation constraints for inference of sub-ice shelf melt and circulation. To date, ocean state estimates have largely been global and regional in focus [Verdy and Mazloff, 2017, Nguyen et al., 2021], and have not resolved ice-shelf cavities or near-ice shelf circulation on the continental shelf. However, such frameworks are being actively developed: [Nakayama et al., 2021b] provided the first MITgcm state estimate that resolves continental shelf and cavity circulation in a regional model. [Smith, 2021] generated an adjoint-based state estimate of the circulation under Pine Island Ice Shelf, and found, through largescale Bayesian uncertainty quantification, the reduction in melt uncertainty due to various observations. 21 Chapter 5 Critical Regions In the following we identify a number of critical regions – either by geographic location of the ice shelf; or locations or physical regimes within ice shelves – where it was deemed important to constrain melt rates and better understand ice-ocean interactions. 5.1 Critical Regimes 5.1.1 Grounding Zones Grounding zones are critical locations within the ice-shelf cavity in terms of icesheet stability as well as unconstrained processes. Because of the transition from a grounded to a floating regime, there is a strong relationship between ocean depth and dynamic thinning and grounded retreat [Weertman, 1974, Schoof, 2007]. Buttressing – the ability of an ice shelf to transfer stress and limit flow – is affected by melt everywhere in the ice shelf, but in particular by melt close to the grounding line [Walker et al., 2008, Goldberg et al., 2019, Arthern and Williams, 2017]. Thus, knowledge of ocean depth and of drivers of submarine melt in the grounding zone are crucial to understanding and predicting ice-sheet loss. Grounding zone melt is influenced by circulation within the ice-shelf cavity, but also by runoff of subglacial water from the ice sheet [Nakayama et al., 2021a, Davis et al., 2023, Gourmelen et al., 2025]. However, runoff location and magnitude is also poorly constrained. Observations and theory suggest that tidal forcing can drive water under grounded ice [Warburton et al., 2020, Rignot et al., 2024], but the impact of tidal pumping on melt rates remains poorly understood (and poorly observed). Knowledge is also hampered by the inability of existing ocean models to represent circulation in grounding zones: ocean columns are typically tens of meters or less, which is the vertical resolution of most regional models. While some grounding zone in situ measurements exist [Davis et al., 2023, Schmidt et al., 2023], more observations of grounding line location and evolution, as well as processes, are needed in order to better understand the rate and evolution of melt in this critical region. Some developments could be made by studies in areas which do not exhibit warm waters or high melt rates, but may be easier 22 to access due to slow ice movement. 5.1.2 Pinning Points Pinning points are topographic highs upon which ice shelves locally ground [Matsuoka et al., 2015]. Through inducing resistive stresses, pinning points can lead to a modified force balance within the ice shelf and its tributary glacier [Reese et al., 2018, Still et al., 2019]. Hence, pinning points play an important role in ice shelf buttressing, stability, and potential grounding line retreat [Goldberg et al., 2009, MacAyeal et al., 1987, Wild et al., 2021]. Pinning points can also influence the magnitude and variability of ice shelf basal melt rates. Local topographic highs alter ocean circulation within an ice shelf cavity [W˚ahlin et al., 2021], and the delivery of warm ocean water to the grounding line. Cycles of grounding and ungrounding of an ice shelf on a pinning point or topographic high modify ice shelf geometry, with a direct influence on basal melt rate magnitude. This can be seen in patterns of thickness and melt variability on the Totten Glacier ice shelf [Roberts et al., 2018], in the Dotson Melt Channel [Zinck et al., 2023], as well as at Pine Island [Lowery et al., 2025]. This mechanism has also been implicated in the formation of cavities in regions near grounding zones, including in the eastern region of Thwaites Glacier [Milillo et al., 2019]. Despite their importance on ocean-driven ice shelf melt and ice shelf stability, the location and geometry of pinning points are poorly constrained in Antarctica due to a paucity of bathymetry observations, particularly in ice shelf cavities. Addressing this is essential in improving the accuracy of simulated ice sheet evolution. 5.1.3 Shear Zones Shear zones are regions of concentrated shear stresses including, for example, the margins of an embayed ice shelf and ice rises. Resistive stresses generated in these zones buttress the upstream ice. It follows that shear zone weakening has implications for ice shelf stability and potential collapse. Indeed, shear margin weakening has been implicated in the disintegration of the Larsen A and B Ice Shelves [Vieli et al., 2006, Wang et al., 2023]. Ocean-driven melt may be a positive feedback on shear margin thinning and weakening [Alley et al., 2016]; that is, ice shelves are typically thinner in shear margins, leading to a slope in the ice shelf base that can increase plume-driven ocean melt of, and thinning within, the margin. Furthermore, rifts commonly form in shear zones, where effective stresses are higher [Bassis et al., 2024, Benn et al., 2022, Walker et al., 2013, Walker et al., 2015]. Rifts that propagate can eventually lead to ice shelf fracture and calving [Sergienko, 2013, Vaughan et al., 2012]. Ice shelf basal channels formed by buoyant plumes and concentrated in shear zones, may further accelerate the formation of rifts and consequent ice loss through structural weakening [Alley et al., 2016]. More accurate characterisation of basal melt in shear zones is critical for predicting ice shelf evolution, and the long-term stability of the Antarctic Ice Sheet. 23 5.1.4 Regions with Potential Regime Shifts Three main regimes are found on the Antarctic continental shelf: fresh, warm and dense [Thompson et al., 2018]. A fresh regime consists of cold and fresh waters, with little presence of warm CDW (e.g. eastern Weddell Sea). The warm regime instead is typical of areas where ice basal melt rates are high (Amundsen Sea and near the Totten Glacier). Here warm CDW fills the bottom later of the water column and is able to access the ice-shelf cavities. Finally, in dense regimes cold and salty waters form in coastal polynyas, contributing to low rates of basal melt and the production of Antarctic Bottom Water (e.g. Ross Sea, western Weddell Sea, Amery Ice Shelf). A regime shift under global warming represents the transition from a fresh/dense regime to a warm state would result in increased ice shelf basal melt at ice shelves currently experiencing low melting and the reduction in Antarctic Bottom Water formation (e.g. [Hellmer et al., 2012]). Two areas are at present considered at risk: the Ross Sea (dense regime) and the eastern Weddell Sea (fresh regime). In the Ross Sea, multidecadal freshening indicates a potential regime shift by mid century, if the freshening trend is sustained [Jacobs et al., 2022] – although recent recoveries in shelf salinity makes future freshening less certain. In the eastern Weddell Sea, wind changes could allow more warm water to reach the ice shelves, as seen in recent years [Lauber et al., 2023], though recent simulations suggest that the western Weddell Sea is less prone to a regime shift, at least this century [Naughten et al., 2021]. Nevertheless, improved constraints on melt rates and their oceanic drivers is important in such regions in order to better predict and detect when regime shifts will occur. 5.1.5 Areas of particularly high melt rates or potential glacial instability Some of the warmest waters on the Antarctic continental shelf are found in the Amundsen and Bellingshausen Seas in West Antarctica, due to transport of Circumpolar Deep Water and relatively low levels of surface transformation [Jacobs et al., 1996, Martinson and McKee, 2012, Petty et al., 2013]. Whereas ice shelves in the Bellingshausen drain relatively small glaciers, those of the Amundsen buttress large, fast-flowing ice streams with large, deepened catchments, with significant potential for influencing sea level rise and ocean circulation. For instance, it is thought that Thwaites Glacier could exhibit Marine Ice Sheet Instability under continued or increased ocean temperatures [Joughin et al., 2014]. Around much of East Antarctica, there is High Salinity Shelf Water on the continental shelf, and the associated ice shelves are relatively insulated from warm intrusions; however, several ice shelves draining the Aurora Basin (Moscow Ice Shelf, Totten Ice Shelf) are exposed to waters approaching temperatures in the Amundsen. Totten exhibits high melt rates and modelling studies suggest its loss rates could greatly increase in the coming century [Pelle et al., 2020]. 24 5.2 Importance of diverse Ice-Ocean boundary conditions Our ability to constrain and understand ice shelf melt rates at continental scales, as well as improve the representation of ice shelf melt in climate models, is made all the more challenging by the diverse range of ice-ocean boundary conditions found beneath Antarctic ice shelves. Understanding how the physics of melting varies across these regimes is a first-order challenge to the community. Basal melting is controlled by the rate at which heat (and salt) are transferred across the meter-scale ice-ocean boundary layer toward the ice base. Beyond the diffusive boundary layer found right at the ice-ocean interface where molecular diffusion dominates, turbulent processes drive the critical heat and salt fluxes. The source and strength of this turbulence vary widely and remain poorly understood. Beneath sloping or vertical ice, the input of fresh basal melt water is theorised to create a turbulent buoyant plume that flows up the ice base and entrains heat and salt from below through the generation of shear-driven turbulence [Jenkins, 2011]. This “convection-driven” regime is expected to be particularly relevant for some of the highest rates of observed basal melting associated with deep ice shelf grounding lines. Very few in-situ observations of this regime exist, however, and much of our knowledge is derived from small-scale modelling and laboratory experiments that do not resolve the geophysical scales of an Antarctic ice shelf [Kerr and McConnochie, 2015, Gayen et al., 2016]. Lessons from convection-driven melting regimes found in Greenlandic and Alaskan outlet glaciers [Zhao et al., 2024, Weiss et al., 2025] may inform future regimes in Antarctic ice shelves. Beneath flat ice, externally driven ocean tidal currents or mean flow can also drive vigorous shear-generated turbulence. In locations where the turbulence is sufficiently energetic, the input of stratification from basal melting is rapidly mixed throughout the boundary layer, generating a well-mixed layer beneath the ice [Davis and Nicholls, 2019, Stevens et al., 2020]. In this regime, stratification plays no dynamical role, and the basal melt rate is dependent upon the friction velocity (i.e. turbulence intensity) and thermal driving (i.e. difference between the temperature of the mixed layer and the in-situ freezing point). In quiescent flow regimes with weak current shear beneath flat ice, the input of buoyant melt water is capable of stratifying the boundary layer, e.g., [Sheehan and Heywood, 2024], and thus it becomes an important dynamical term in the turbulence budget [Davis et al., 2023]. The presence of this stratification suppresses turbulence and reduces the melt rate compared with the strong shear-driven environment [Vreugdenhil and Taylor, 2019, Rosevear et al., 2022]. In strongly-stratified environments, the differing molecular diffusivities of heat and salt also give rise to double-diffusive convection, which in the absence of turbulence, can drive an elevated heat flux to the ice base compared to a zero turbulence or diffusion only environment [Middleton et al., 2021, Middleton et al., 2022, Rosevear et al., 2021]. Irrespective of the different turbulent processes and regimes that drive basal melting, the scales of turbulence remain far smaller than those that can be re25 Chapter 7 A Framework for reconciling estimates of ice-shelf melt 7.1 Framework Aims, Themes, and Values This framework is designed to support the reconciliation of different methods used to estimate ice-shelf melt, with the aim of improving the consistency, reliability, and scientific value of melt rate estimates. Reconciliation, in this context, does not imply that all methods must yield identical results. Rather, it emphasizes understanding and explaining the discrepancies between approaches, quantifying uncertainties, and defining the conditions under which each method performs best. The core values of the framework are transparency, collaboration, and methodological rigor. These will be realized through standardized protocols and quality control mechanisms that ensure consistency in data collection, processing, and reporting. Each method must be accompanied by detailed metadata, including data formats, methodological descriptions, and quality assessment procedures. Uncertainty quantification is central—each method must include clear estimates of uncertainty, as well as an articulation of the spatial and temporal resolution it can realistically resolve. By identifying the sources of uncertainty and limitations for each method—be they observational, instrumental, or interpretative—the framework allows for more informed comparisons. Furthermore, the framework promotes open science principles. Data and methods should be accessible, interoperable, and well-documented, fostering a culture of reproducibility and collective progress. Finally, it emphasizes the strategic prioritization of improving individual methods’ performance before focusing on reconciliation across them. 7.2 What Can We Do Now? Leveraging Existing Data and Networks Even before the collection of new data or launching new campaigns, institutional and organisational mechanisms can be put into place to improve the intercom32 parison of melt rate calculation. A number of standards and best practices can be put into place. Concerning oceanographic-based estimates of cavity-wide melt: •Oceanographic measurements including temperature, salinity, pressure, and current velocity are vital. Best practices for sensor calibration can be put in place. Similarly best practices for lab standards for tracers such as dissolved oxygen can be used. •The calculations and physical parameters used for melt water concentration and melt water transport can be standardized. For the latter case, this is especially important because in general this is an underconstrained problem. Differing standards can be put in place for both the use of ADCP currents and the use of geostrophic currents, and shared codes should be made available. For satellite processing: •There are several sources of disagreement associated with satellite-derived melt-rate calculations. Quantifying the overall error will help us assess the level of disagreement between different studies. This should be done by tracking each term in Equation 3.5 (Lagrangian elevation change; ice divergence; firn thickness and compaction rate; and surface mass balance), and conducting a formal error propagation analysis. •Where possible, standards should be set for (i) processing methods and (ii) upstream data sources. For instance, more than one satellite velocity product exists, and there is also variability in how such products are used to generate divergence fields and Lagrangian thinning. Upstream data products should be codified and standardised; and processing codes shared in public repositories. •Similarly, firn and surface mass balance model choices should be standardised. Known uncertainties within these models should be made more visible in publications. Finally, for ApRES, it is noted that most processing has followed similar principles, but that there is not a shared open-source version-controlled code for generating melt estimates – and this should be put in place. More standardized validations for various melt products can be carried out – and a protocol can be described. Dotson ice shelf, for instance, has the longest, most continuous (and self-consistent) series of oceanographic-based melt estimates of all Antarctic ice shelves. A comparison is carried out by [Adusumilli et al., 2020] but not others. [Vaˇnkov´a and Nicholls, 2022], for instance, made comparisons between ApRES and the Adusumilli product over the Filchner-Ronne Ice Shelf. Long-term means compare favorably in general (though notably not where Adusumilli shows refreezing) – but the temporal variability of melt in the remotely sensed product is much greater than that shown by ApRES. Building on this, a protocol of validation of remote-sensing products can be devised with the (ever increasing 33 set of) validation data shared publicly, as a better way to evaluate the utility of existing and new remotely sensed products [Cook et al., 2022]. To some extent this standard of validation may take the form of a formal intercomparison, and may require additional resources (see Chapter 7.3). 7.3 What Could We Do? A Collaborative and Integrative Approach Looking forward, a collaborative and integrated approach is required to fully reconcile melt estimates. A general consensus reached at the RECOIL workshop was that the continent wide and long term observations of ice shelf melt needed for effective sustained monitoring of melt can only be achieved practicably via remote sensing (satellite) methods. The (critical) importance of other methods discussed in this document was widely agreed upon, particularly for the development of process understanding and implementation in models. However, it was recognised that only remote sensing could economically provide the large scale and long term data required for the ongoing monitoring needed to project future ice sheet melt and consequent sea level rise and ocean circulation impacts. 7.3.1 Observations In order to address the current mismatch between in situ and remote sensing, it was felt that coordinated field campaigns and satellite analysis, specifically for cross validation, were needed. Ideally these would be modelled after successful international collaborations such as the International Thwaites Glacier Collaboration (ITGC), with cross community engagement between remote sensing experts and both observational and modelling glaciologists and oceanographers. One compelling concept discussed is the establishment of an ‘integrative ice shelf laboratory’ – a year round multi-method observational campaign on a suitable candidate ice shelf. This would nominally consist of a number of in situ ‘super sites’ which would maintain year round observations of ice shelf properties on, under and within the shelf, specifically geared for comparison with remote sensing instruments. The number and nature of these sites would depend on the size and dynamics of the ice shelf chosen, and the required spatial resolution. The Nansen ice shelf was proposed as a potential candidate for such a campaign, but regardless of the specific location the essential idea is to establish a platform for the testing of, and identification of biases in, methodologies and the refinement of reconciliation protocols between remote and in situ approaches. In the data intercomparison space, a similar collaborative approach was suggested. Such an effort may be structured following the model intercomparison project (MIP) system, whereby various platforms are systematically designed and compared using an agreed upon set of metrics, and the Ice Sheet Mass Balance Intercomparison Exercise (IMBIE) or the equivalent for glacier (GlaMBIE). This would allow for robust assessments of consistency and to help identify systematic biases. 34 Figure 7.1: A potential vision of an “ice shelf laboratory”. A network of instrument stations can enable detailed monitoring of melt rates at discrete locations, as well as weather and GNSS data to validate climate and altimetry data, respectively. Autonomous vehicles and moorings can establish near-ice conditions with deployed ships measuring shelf-wide transports over comparable time windows. Dedicated funding mechanisms and data frameworks are needed to support such integrative, multinational efforts. Specific examples where such funding should be targeted include the upcoming Antarctica InSync (2027-29) and International Polar Year (2032-33), where multiple nations will be making concerted efforts to address these and other observational issues in a coordinated way. Funding should be targeted at the process-based understanding required to support model improvements and development, then at the harmonisation/calibration field work needed to integrate these datasets together, and finally at the long term observations needed to constrain models and guide their evolution in a changing climate. This should go hand in hand with the development of new in situ instrumental and satellite remote sensing systems, targeted at reducing biases between observational methodologies and supporting modelling improvements. 7.3.2 Modeling There is no way around the fact that even under the best circumstances (such as the pooling of all available observational assets for equipping a regional “supersite” or “ice shelf laboratory”) observational sampling will remain sparse in space and time. This calls for exploring approaches that complement field work with simulation-based approaches. Chapter 4 has touched upon the merit of formal parameter and state estimation approaches to synthesize the diverse and heterogeneous data streams into a full time-evolving, three-dimensional picture of the state of the regional system being considered. Access to such state estimates would enable detailed analyses of “closed property budgets” (i.e., accounting for all source and sink terms in conservation equations without spurious terms) and watermass transformation calculations that are free of unaccounted residuals such 35 as occur in observation-only approaches. Such a framework may also be used to explore formal parameter inversion, targeting head-on the use of the observational to invert for “optimal melt rates” (or coefficients in the melt rate parameterization) that lead to mutual consistency between observed and simulated states. An exciting prospect would be to develop and deploy a strongly coupled adjoint-based ocean-ice shelf-ice stream (and eventually atmospheric) estimation framework, wherein observational data of one component state (the ice shelf, say) will also formally constrain the state of the other component (the ocean). Such a coupled estimation system has not so yet been conducted but is within reach, given adjoint-based systems of the separate components now exist, and the utility of which have been demonstrated. A coupled estimation framework would be configured for one (or several) ice shelf laboratories and used to support the observational campaign in various ways. We note that to do this well requires research in its own right as it would deploy modern computational algorithms that still need to find their way from the computational science literature into real-world applications. From a predictive modeling perspective, another potential benefit is that the simulation and estimation framework used in this way would also find its way – as a well-calibrated, data-constrained component – into next-generation coupled Earth system models used in climate projections. 7.3.3 Final thoughts The rapid change becoming apparent in several regions around Antarctica makes the RECOIL workshop and its recommendations particularly timely. The melt in regions such as the hotspots in the West Antarctic, as well as the Totten, Cook and Denman glaciers in the East is critical to monitor and project, given the potential for future sea level rise and possibility of tipping points in ice shelf stability. However, without significant efforts in expanding our observational capacity, harmonising the various approaches, and developing faithful predictive capabilities, we risk much greater uncertainty in future sea level rise, ocean circulation and climatic impacts. 36 Bibliography [Adusumilli et al., 2020] Adusumilli, S., Fricker, H. A., Medley, B., Padman, L., and Siegfried, M. R. (2020). Interannual variations in meltwater input to the southern ocean from antarctic ice shelves. Nature geoscience, 13(9):616–620. [Adusumilli et al., 2018] Adusumilli, S., Fricker, H. A., Siegfried, M. R., Padman, L., Paolo, F. S., and Ligtenberg, S. R. (2018). Variable basal melt rates of antarctic peninsula ice shelves, 1994–2016. Geophysical Research Letters, 45(9):4086–4095. [Alley et al., 2018] Alley, K. E., Scambos, T. A., Anderson, R. S., Rajaram, H., Pope, A., and Haran, T. M. (2018). Continent-wide estimates of antarctic strain rates from landsat 8-derived velocity grids. Journal of Glaciology, 64(244):321– 332. [Alley et al., 2016] Alley, K. E., Scambos, T. A., Siegfried, M. R., and Fricker, H. A. (2016). Impacts of warm water on antarctic ice shelf stability through basal channel formation. Nature Geoscience, 9(4):290–293. [Anselin et al., 2023] Anselin, J., Reed, B., Jenkins, A., and Green, J. (2023). Ice shelf basal melt sensitivity to tide-induced mixing based on the theory of subglacial plumes. Journal of Geophysical Research: Oceans, 128(4):e2022JC019156. [Arthern et al., 2010] Arthern, R. J., Vaughan, D. G., Rankin, A. M., Mulvaney, R., and Thomas, E. R. (2010). In situ measurements of antarctic snow compaction compared with predictions of models. Journal of Geophysical Research: Earth Surface, 115(F3). [Arthern and Williams, 2017] Arthern, R. J. and Williams, C. R. (2017). The sensitivity of west antarctica to the submarine melting feedback. Geophysical Research Letters, 44(5):2352–2359. [Badgeley et al., 2025] Badgeley, J. A., Morlighem, M., and Seroussi, H. (2025). Increased sea-level contribution from northwestern Greenland for models that reproduce observations. Proceedings of the National Academy of Sciences, 122(25):e2411904122. [Bassis et al., 2024] Bassis, J. N., Crawford, A., Kachuck, S. B., Benn, D. I., Walker, C., Millstein, J., Duddu, R., ˚ Astr¨om, J., Fricker, H. A., and Luckman, A. (2024). Stability of ice shelves and ice cliffs in a changing climate. Annual Review of Earth and Planetary Sciences, 52(1):221–247. 37 [Benn et al., 2022] Benn, D. I., Luckman, A., ˚ Astr¨om, J. A., Crawford, A. J., Cornford, S. L., Bevan, S. L., Zwinger, T., Gladstone, R., Alley, K., Pettit, E., et al. (2022). Rapid fragmentation of thwaites eastern ice shelf. The Cryosphere, 16(6):2545–2564. [Berger et al., 2017] Berger, S., Drews, R., Helm, V., Sun, S., and Pattyn, F. (2017). Detecting high spatial variability of ice shelf basal mass balance, roi baudouin ice shelf, antarctica. The Cryosphere, 11(6):2675–2690. [Biddle et al., 2019] Biddle, L. C., Loose, B., and Heywood, K. J. (2019). Upper ocean distribution of glacial meltwater in the amundsen sea, antarctica. Journal of Geophysical Research: Oceans, 124(10):6854–6870. [Bronselaer et al., 2018] Bronselaer, B., Winton, M., Griffies, S. M., Hurlin, W. J., Rodgers, K. B., Sergienko, O. V., Stouffer, R. J., and Russell, J. L. (2018). Change in future climate due to Antarctic meltwater. Nature, 348(7734):327 – 58. [Case and Kingslake, 2022] Case, E. and Kingslake, J. (2022). Phase-sensitive radar as a tool for measuring firn compaction. Journal of Glaciology, 68(267):139–152. [Chartrand et al., 2024] Chartrand, A. M., Howat, I. M., Joughin, I. R., and Smith, B. E. (2024). Thwaites glacier thins and retreats fastest where ice-shelf channels intersect its grounding zone. The Cryosphere, 18(11):4971–4992. [Cook et al., 2022] Cook, S., Nicholls, K. W., Vaˇnkov´a, I., Thompson, S. S., and Galton-Fenzi, B. K. (2022). Data initiatives for ocean-driven melt of antarctic ice shelves. Annals of Glaciology, 63(87-89):27–32. [Cornford et al., 2015] Cornford, S. L., Martin, D., Payne, A., Ng, E., Le Brocq, A., Gladstone, R. M., Edwards, T. L., Shannon, S. R., Agosta, C., van den Broeke, M. R., et al. (2015). Century-scale simulations of the response of the west antarctic ice sheet to a warming climate. The Cryosphere, 9(4):1579–1600. [Davis and Nicholls, 2019] Davis, P. E. and Nicholls, K. W. (2019). Turbulence observations beneath larsen c ice shelf, antarctica. Journal of Geophysical Research: Oceans, 124(8):5529–5550. [Davis et al., 2023] Davis, P. E., Nicholls, K. W., Holland, D. M., Schmidt, B. E., Washam, P., Riverman, K. L., Arthern, R. J., Vaˇnkov´a, I., Eayrs, C., Smith, J. A., et al. (2023). Suppressed basal melting in the eastern thwaites glacier grounding zone. Nature, 614(7948):479–485. [Davison et al., 2023] Davison, B. J., Hogg, A. E., Gourmelen, N., Jakob, L., Wuite, J., Nagler, T., Greene, C. A., Andreasen, J., and Engdahl, M. E. (2023). Annual mass budget of antarctic ice shelves from 1997 to 2021. Science Advances, 9(41):eadi0186. 38 [Depoorter et al., 2013] Depoorter, M. A., Bamber, J. L., Griggs, J. A., Lenaerts, J. T., Ligtenberg, S. R., van den Broeke, M. R., and Moholdt, G. (2013). Calving fluxes and basal melt rates of antarctic ice shelves. Nature, 502(7469):89–92. [Ding et al., 2011] Ding, Q., Steig, E. J., Battisti, D. S., and K¨uttel, M. (2011). Winter warming in West Antarctica caused by central tropical Pacific warming. Nature Geoscience, 4(6):398–403. [Dutrieux et al., 2014] Dutrieux, P., De Rydt, J., Jenkins, A., Holland, P. R., Ha, H. K., Lee, S. H., Steig, E. J., Ding, Q., Abrahamsen, E. P., and Schr¨oder, M. (2014). Strong sensitivity of pine island ice-shelf melting to climatic variability. Science, 343(6167):174–178. [Dutrieux et al., 2013] Dutrieux, P., Vaughan, D. G., Corr, H. F., Jenkins, A., Holland, P. R., Joughin, I., and Fleming, A. (2013). Pine island glacier ice shelf melt distributed at kilometre scales. The Cryosphere, 7(5):1543–1555. [Gardner et al., 2018] Gardner, A. S., Moholdt, G., Scambos, T., Fahnstock, M., Ligtenberg, S., Van Den Broeke, M., and Nilsson, J. (2018). Increased west antarctic and unchanged east antarctic ice discharge over the last 7 years. The Cryosphere, 12(2):521–547. [Gayen et al., 2016] Gayen, B., Griffiths, R. W., and Kerr, R. C. (2016). Simulation of convection at a vertical ice face dissolving into saline water. Journal of Fluid Mechanics, 798:284–298. [Gerringa et al., 2012] Gerringa, L. J., Alderkamp, A.-C., Laan, P., Thuroczy, C.-E., De Baar, H. J., Mills, M. M., van Dijken, G. L., van Haren, H., and Arrigo, K. R. (2012). Iron from melting glaciers fuels the phytoplankton blooms in amundsen sea (southern ocean): Iron biogeochemistry. Deep Sea Research Part II: Topical Studies in Oceanography, 71:16–31. [Goldberg et al., 2019] Goldberg, D., Gourmelen, N., Kimura, S., Millan, R., and Snow, K. (2019). How accurately should we model ice shelf melt rates? Geophysical Research Letters, 46(1):189–199. [Goldberg et al., 2009] Goldberg, D., Holland, D., and Schoof, C. (2009). Grounding line movement and ice shelf buttressing in marine ice sheets. Journal of Geophysical Research: Earth Surface, 114(F4). [Goldberg and Heimbach, 2013] Goldberg, D. N. and Heimbach, P. (2013). Parameter and state estimation with a time-dependent adjoint marine ice sheet model. The Cryosphere, 7(6):1659 – 1678. [Goldberg et al., 2015] Goldberg, D. N., Heimbach, P., Joughin, I., and Smith, B. (2015). Committed retreat of Smith, Pope, and Kohler Glaciers over the next 30 years inferred by transient model calibration. The Cryosphere, 9(6):2429– 2446. 39 [Goldberg et al., 2020] Goldberg, D. N., Smith, T. A., Narayanan, S. H., Heimbach, P., and Morlighem, M. (2020). Bathymetric influences on antarctic ice-shelf melt rates. Journal of Geophysical Research: Oceans, 125(11):e2020JC016370. [Gourmelen et al., 2017] Gourmelen, N., Goldberg, D. N., Snow, K., Henley, S. F., Bingham, R. G., Kimura, S., Hogg, A. E., Shepherd, A., Mouginot, J., Lenaerts, J. T., et al. (2017). Channelized melting drives thinning under a rapidly melting antarctic ice shelf. Geophysical Research Letters, 44(19):9796– 9804. [Gourmelen et al., 2025] Gourmelen, N., Jakob, L., Holland, P., Dutrieux, P., Goldberg, D., Bevan, S., Luckman, A., and Malczyk, G. (2025). The influence of subglacial lake discharge on thwaites glacier ice-shelf melting and groundingline retreat. Nature Communications, 16(1):2272. [Gwyther et al., 2023] Gwyther, D. E., Dow, C. F., Jendersie, S., Gourmelen, N., and Galton-Fenzi, B. K. (2023). Subglacial freshwater drainage increases simulated basal melt of the totten ice shelf. Geophysical Research Letters, 50(12):e2023GL103765. [Harrison et al., 2022] Harrison, L. C., Holland, P. R., Heywood, K. J., Nicholls, K. W., and Brisbourne, A. M. (2022). Sensitivity of melting, freezing and marine ice beneath larsen c ice shelf to changes in ocean forcing. Geophysical Research Letters, 49(4):e2021GL096914. [Hawley and Waddington, 2011] Hawley, R. L. and Waddington, E. D. (2011). In situ measurements of firn compaction profiles using borehole optical stratigraphy. Journal of Glaciology, 57(202):289–294. [Heimbach et al., 2019] Heimbach, P., Fukumori, I., Hill, C. N., Ponte, R. M., Stammer, D., Wunsch, C., Campin, J.-M., Cornuelle, B., Fenty, I., Forget, G., K¨ohl, A., Mazloff, M., Menemenlis, D., Nguyen, A. T., Piecuch, C., Trossman, D., Verdy, A., Wang, O., and Zhang, H. (2019). Putting It All Together: Adding Value to the Global Ocean and Climate Observing Systems With Complete Self-Consistent Ocean State and Parameter Estimates. Frontiers in Marine Science, 6:769 – 10. [Hellmer et al., 2012] Hellmer, H. H., Kauker, F., Timmermann, R., Determann, J., and Rae, J. (2012). Twenty-first-century warming of a large antarctic iceshelf cavity by a redirected coastal current. Nature, 485(7397):225–228. [Herraiz-Borreguero et al., 2016] Herraiz-Borreguero, L., Lannuzel, D., van der Merwe, P., Treverrow, A., and Pedro, J. (2016). Large flux of iron from the amery ice shelf marine ice to prydz bay, east antarctica. Journal of Geophysical Research: Oceans, 121(8):6009–6020. [Heywood et al., 2016] Heywood, K. J., Biddle, L. C., Boehme, L., Dutrieux, P., Fedak, M., Jenkins, A., Jones, R. W., Kaiser, J., Mallett, H., Garabato, A. C. N., et al. (2016). Between the devil and the deep blue sea: The role of the 40 amundsen sea continental shelf in exchanges between ocean and ice shelves. Oceanography, 29(4):118–129. [Hirano et al., 2017] Hirano, D., Tamura, T., Ushio, S., Ohshima, K. I., Simizu, D., Ono, K., Noguchi, T., and Aoki, S. (2017). Observations of ice tongue-ocean interaction at shirase glacier. [Holland et al., 2023] Holland, P. R., Bevan, S. L., and Luckman, A. J. (2023). Strong ocean melting feedback during the recent retreat of thwaites glacier. Geophysical Research Letters, 50(8):e2023GL103088. [Huhn et al., 2018] Huhn, O., Hattermann, T., Davis, P. E., Dunker, E., Hellmer, H. H., Nicholls, K. W., Østerhus, S., Rhein, M., Schr¨oder, M., and S¨ultenfuß, J. (2018). Basal melt and freezing rates from first noble gas samples beneath an ice shelf. Geophysical Research Letters, 45(16):8455–8461. [Jacobs et al., 2022] Jacobs, S., Giulivi, C., and Dutrieux, P. (2022). Persistent ross sea freshening from imbalance west antarctic ice shelf melting. Journal of Geophysical Research: Oceans, 127(3):e2021JC017808. [Jacobs et al., 2013] Jacobs, S., Giulivi, C., Dutrieux, P., Rignot, E., Nitsche, F., and Mouginot, J. (2013). Getz ice shelf melting response to changes in ocean forcing. Journal of Geophysical Research: Oceans, 118(9):4152–4168. [Jacobs et al., 1996] Jacobs, S. S., Hellmer, H. H., and Jenkins, A. (1996). Antarctic ice sheet melting in the southeast pacific. Geophysical Research Letters, 23(9):957–960. [Jacobs et al., 2011] Jacobs, S. S., Jenkins, A., Giulivi, C. F., and Dutrieux, P. (2011). Stronger ocean circulation and increased melting under pine island glacier ice shelf. Nature geoscience, 4(8):519–523. [Jenkins, 2011] Jenkins, A. (2011). Convection-Driven Melting near the Grounding Lines of Ice Shelves and Tidewater Glaciers. Journal of Physical Oceanography, 41(12):2279 – 2294. [Jenkins and Doake, 1991] Jenkins, A. and Doake, C. S. (1991). Ice-ocean interaction on ronne ice shelf, antarctica. Journal of Geophysical Research: Oceans, 96(C1):791–813. [Jenkins and Jacobs, 2008] Jenkins, A. and Jacobs, S. (2008). Circulation and melting beneath george vi ice shelf, antarctica. Journal of Geophysical Research: Oceans, 113(C4). [Jenkins et al., 2010] Jenkins, A., Nicholls, K. W., and Corr, H. F. (2010). Observation and parameterization of ablation at the base of ronne ice shelf, antarctica. Journal of Physical Oceanography, 40(10):2298–2312. [Jenkins et al., 2018] Jenkins, A., Shoosmith, D., Dutrieux, P., Jacobs, S., Kim, T. W., Lee, S. H., Ha, H. K., and Stammerjohn, S. (2018). West antarctic ice sheet retreat in the amundsen sea driven by decadal oceanic variability. Nature Geoscience, 11(10):733–738. 41 [Thompson et al., 2018] Thompson, A. F., Stewart, A. L., Spence, P., and Heywood, K. J. (2018). The antarctic slope current in a changing climate. Reviews of Geophysics, 56(4):741–770. [Timmermann and Hellmer, 2013] Timmermann, R. and Hellmer, H. H. (2013). Southern ocean warming and increased ice shelf basal melting in the twentyfirst and twenty-second centuries based on coupled ice-ocean finite-element modelling. Ocean Dynamics, 63(9):1011–1026. [Vaˇnkov´a et al., 2021] Vaˇnkov´a, I., Cook, S., Winberry, J. P., Nicholls, K. W., and Galton-Fenzi, B. K. (2021). Deriving melt rates at a complex ice shelf base using in situ radar: Application to totten ice shelf. Geophysical Research Letters, 48(7):e2021GL092692. [Vaˇnkov´a and Nicholls, 2022] Vaˇnkov´a, I. and Nicholls, K. W. (2022). Ocean variability beneath the filchner-ronne ice shelf inferred from basal melt rate time series. Journal of Geophysical Research: Oceans, 127(10):e2022JC018879. [Vaˇnkov´a et al., 2020] Vaˇnkov´a, I., Nicholls, K. W., Corr, H. F., Makinson, K., and Brennan, P. V. (2020). Observations of tidal melt and vertical strain at the filchner-ronne ice shelf, antarctica. Journal of Geophysical Research: Earth Surface, 125(1):e2019JF005280. [Vaˇnkov´a et al., 2023] Vaˇnkov´a, I., Winberry, J. P., Cook, S., Nicholls, K. W., Greene, C. A., and Galton-Fenzi, B. K. (2023). High spatial melt rate variability near the totten glacier grounding zone explained by new bathymetry inversion. Geophysical Research Letters, 50(10):e2023GL102960. [Vaughan et al., 2012] Vaughan, D. G., Corr, H. F., Bindschadler, R. A., Dutrieux, P., Gudmundsson, G. H., Jenkins, A., Newman, T., Vornberger, P., and Wingham, D. J. (2012). Subglacial melt channels and fracture in the floating part of pine island glacier, antarctica. Journal of Geophysical Research: Earth Surface, 117(F3). [Verdy and Mazloff, 2017] Verdy, A. and Mazloff, M. R. (2017). A data assimilating model for estimating s outhern o cean biogeochemistry. Journal of Geophysical Research: Oceans, 122(9):6968–6988. [Vieli et al., 2006] Vieli, A., Payne, A. J., Du, Z., and Shepherd, A. (2006). Numerical modelling and data assimilation of the larsen b ice shelf, antarctic peninsula. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 364(1844):1815–1839. [Vreugdenhil and Taylor, 2019] Vreugdenhil, C. A. and Taylor, J. R. (2019). Stratification effects in the turbulent boundary layer beneath a melting ice shelf: Insights from resolved large-eddy simulations. Journal of Physical Oceanography, 49(7):1905–1925. [W˚ahlin et al., 2024] W˚ahlin, A., Alley, K. E., Begeman, C., Hegrenæs, Ø., Yuan, X., Graham, A. G., Hogan, K., Davis, P. E., Dotto, T. S., Eayrs, C., et al. 48 (2024). Swirls and scoops: Ice base melt revealed by multibeam imagery of an antarctic ice shelf. Science Advances, 10(31):eadn9188. [W˚ahlin et al., 2021] W˚ahlin, A., Graham, A., Hogan, K., Queste, B., Boehme, L., Larter, R., Pettit, E., Wellner, J., and Heywood, K. (2021). Pathways and modification of warm water flowing beneath thwaites ice shelf, west antarctica. Science Advances, 7(15):eabd7254. [W˚ahlin et al., 2020] W˚ahlin, A. K., Steiger, N., Darelius, E., Assmann, K. M., Glessmer, M. S., Ha, H. K., Herraiz-Borreguero, L., Heuz´e, C., Jenkins, A., Kim, T.-W., et al. (2020). Ice front blocking of ocean heat transport to an antarctic ice shelf. Nature, 578(7796):568–571. [Walker et al., 2013] Walker, C., Bassis, J., Fricker, H., and Czerwinski, R. (2013). Structural and environmental controls on antarctic ice shelf rift propagation inferred from satellite monitoring. Journal of Geophysical Research: Earth Surface, 118(4):2354–2364. [Walker et al., 2015] Walker, C. C., Bassis, J. N., Fricker, H. A., and Czerwinski, R. J. (2015). Observations of interannual and spatial variability in rift propagation in the amery ice shelf, antarctica, 2002–14. Journal of Glaciology, 61(226):243–252. [Walker et al., 2008] Walker, R., Dupont, T., Parizek, B., and Alley, R. (2008). Effects of basal-melting distribution on the retreat of ice-shelf grounding lines. Geophysical Research Letters, 35(17). [Wang et al., 2023] Wang, S., Liu, H., Alley, R. B., Jezek, K., Alexander, P., Alley, K. E., Huang, Z., and Wang, L. (2023). Multidecadal pre-and postcollapse dynamics of the northern larsen ice shelf. Earth and Planetary Science Letters, 609:118077. [Warburton et al., 2020] Warburton, K., Hewitt, D., and Neufeld, J. (2020). Tidal grounding-line migration modulated by subglacial hydrology. Geophysical Research Letters, 47(17):e2020GL089088. [Weertman, 1974] Weertman, J. (1974). Stability of the junction of an ice sheet and an ice shelf. Journal of Glaciology, 13(67):3–11. [Wei et al., 2020] Wei, W., Blankenship, D. D., Greenbaum, J. S., Gourmelen, N., Dow, C. F., Richter, T. G., Greene, C. A., Young, D. A., Lee, S., Kim, T.-W., et al. (2020). Getz ice shelf melt enhanced by freshwater discharge from beneath the west antarctic ice sheet. The Cryosphere, 14(4):1399–1408. [Weiss et al., 2025] Weiss, K. J., Nash, J. D., Wengrove, M. E., Osman, N., Cohen, N. F., Jackson, R. H., Pettit, E. C., Zhao, K. X., Nahorniak, J. S., and Sutherland, D. A. (2025). Direct Measurement of Glacier Ice Melt: Boundary Layer Details Are Critical for Submarine Melt Prediction at Near-Vertical Ice Faces. Geophysical Research Letters, 52(11). 49 [Wild et al., 2021] Wild, C. T., Alley, K. E., Muto, A., Truffer, M., Scambos, T. A., and Pettit, E. C. (2021). Weakening of the pinning point buttressing thwaites glacier, west antarctica. The Cryosphere Discussions, 2021:1–28. [Wild et al., 2024] Wild, C. T., Kachuck, S. B., Luckman, A., Alley, K. E., Sharp, M. A., Smith, H., Tyler, S. W., Kratt, C., Dotto, T. S., Price, D., et al. (2024). Rift propagation signals the last act of the thwaites eastern ice shelf despite low basal melt rates. Journal of Glaciology, 70:e21. [Wilson et al., 2017] Wilson, N., Straneo, F., and Heimbach, P. (2017). Satellitederived submarine melt rates and mass balance (2011–2015) for greenland’s largest remaining ice tongues. The Cryosphere, 11(6):2773–2782. [Wunsch et al., 2009] Wunsch, C., Heimbach, P., Ponte, R. M., Fukumori, I., and MEMBERS, E.-G. C. (2009). The global general circulation of the ocean estimated by the ecco-consortium. Oceanography, 22(2):88–103. [Wunsch et al., 2023] Wunsch, C., Williamson, S., and Heimbach, P. (2023). Potential artifacts in conservation laws and invariants inferred from sequential state estimation. Ocean Science, 19(4):1253–1275. [Yoon et al., 2022] Yoon, S.-T., Lee, W. S., Nam, S., Lee, C.-K., Yun, S., Heywood, K., Boehme, L., Zheng, Y., Lee, I., Choi, Y., et al. (2022). Ice front retreat reconfigures meltwater-driven gyres modulating ocean heat delivery to an antarctic ice shelf. Nature communications, 13(1):306. [Zeising et al., 2021] Zeising, O., Steinhage, D., Nicholls, K. W., Corr, H. F., Stewart, C. L., and Humbert, A. (2021). Basal melt of the southern filchner ice shelf, antarctica. The Cryosphere Discussions, 2021:1–19. [Zhao et al., 2024] Zhao, K. X., Skyllingstad, E. D., and Nash, J. D. (2024). Improved Parameterizations of Vertical Ice-Ocean Boundary Layers and Melt Rates. Geophysical Research Letters, 51(4). [Zheng et al., 2021] Zheng, Y., Heywood, K. J., Webber, B. G., Stevens, D. P., Biddle, L. C., Boehme, L., and Loose, B. (2021). Winter seal-based observations reveal glacial meltwater surfacing in the southeastern amundsen sea. Communications Earth & Environment, 2(1):40. [Zinck et al., 2023] Zinck, A.-S. P., Wouters, B., Lambert, E., and Lhermitte, S. (2023). Unveiling spatial variability within the dotson melt channel through high-resolution basal melt rates from the reference elevation model of antarctica. The Cryosphere, 17(9):3785–3801. 50