Ice-sheet mass balance and climate change. - Hanna et al (2013)

Review: Ice-Sheet Mass Balance and Climate Change

Overview of ice-sheet mass balance and climate change since the 2007 IPCC AR4.
Greenland is losing ice mass at an increasing pace.
Current Antarctic ice loss is likely less than some recent estimates.
Uncertainty remains regarding East Antarctica's ice mass change over the past 20 years.
Significant uncertainties persist in ice-mass change estimates for West Antarctica and the Antarctic Peninsula.

Introduction

This review synthesizes advances in monitoring and modeling ice-sheet mass balance since the IPCC AR4.
Mass balance is defined as the net result of mass gains (snow accumulation) and mass losses (meltwater runoff and dynamical discharge).
Surface mass balance (SMB) includes mass gains/losses at the ice-sheet surface, excluding dynamical mass loss.
Satellite geodetic techniques (altimetry, interferometry, gravimetry) have been refined to determine ice-sheet mass balance.
The review covers mass-balance estimates for the Antarctic Ice Sheet (AIS) and the Greenland Ice Sheet (GIS).
New glacial isostatic adjustment (GIA) models have led to significant downwards revision in GIA, affecting gravimetric and altimetric estimates of Antarctic mass loss.
Ice-sheet models now simulate the coupling between ice sheets, ice streams, and ice shelves more accurately.
Improved model representation of interactions between the ice sheet, bed, atmosphere, and ocean.
Brief discussion of sea-level rise (SLR) contributions from glaciers/ice caps, thermal expansion, and terrestrial water storage changes.
Improved observations and predictions of ice-sheet response to climate change are urgently needed for mitigation and adaptation models of SLR.

Recent Changes in Ice-Sheet Mass Balance

One of the primary goals in Earth science is relating the mass-balance state of ice sheets to observed SLR.
Mass-balance state provides an unambiguous quantification of the ice-sheet system response to climate change.
Mass-change estimates are derived from three categories of techniques:
- Space altimetric techniques: Measuring ice-sheet surface height using radar or laser altimetry.
- Space gravimetric techniques: Measuring Earth's gravity field changes using GRACE to derive ice-sheet mass changes.
- Mass budget technique: Comparing net ice accumulation with discharge across the grounding line.
Each technique relies on unique observational data and has different sensitivities to errors and biases.
Mass budget studies use modeled snowfall fields from atmospheric reanalysis data to estimate mass input, sensitive to errors in mean accumulation rate.
Altimetry studies use the same fields to estimate the effective density of volume changes, less sensitive to accumulation rate fluctuations.
GRACE and altimetry studies require accurate removal of GIA-related vertical bedrock motion.
GIA correction is crucial because vertical motion can be misinterpreted as ice-mass change, especially in GRACE data for Antarctica.

Box 1: Recent Developments in GIA Models

Glacial isostatic adjustment (GIA) is the solid Earth's response to past ice/ocean mass redistributions.
GIA causes regional vertical rebound of the Earth's surface.
GIA models correct measurements of present-day ice-mass change and long-term modeling.
Assimilation of glacial geological constraints and geodetic constraints on rebound reduces GIA uncertainty.
Key challenges remain:
- Poorly known ice extent and thickness changes during the past millennium.
- Lateral variations in Earth structure beneath Antarctica.
- Limitations of data used to tune GIA models, leading to probabilistic approaches.

Comparison of Mass-Balance Estimates

Published estimates of ice-sheet mass change rates show a large spread of values.
Some spread is due to technical differences and varying measurement epochs.
Recent estimates are beginning to provide a more coherent picture for both Antarctica and Greenland.
For Greenland, increasing mass loss trend is clear (SMB decrease and ice-to-ocean discharge increase).
Some large mass loss estimates for Antarctica have been discarded.

Reduced Uncertainties

Recent assessments and GPS observations indicate Antarctic GIA-related bedrock motion peaks at 5-6 \, \text{mm yr}^{-1}.
Resulting GIA models deliver less than half the mass corrections compared to previous models.
GRACE data processing has become more consistent as the time series lengthens.
Estimates using the latest models show moderate, if increasing, decadal mass losses for Antarctica.
The IMBIE project compiled average sets of mass-balance estimates for common time periods.
Technical change: IMBIE mass budget estimates use radar altimetry data to show near-zero rates of mass change in unsurveyed areas.
Including recent GIA estimates for Antarctica brought GRACE estimates closer to altimetry estimates.
IMBIE estimates are simple averages; discordance among methods remains not fully understood.
Figure 1 shows reduced disparity of recent mass-balance results among techniques.
Systematic differences exist between techniques: mass budget gives the most negative estimate, laser altimetry the most positive, and GRACE in between.
IMBIE radar altimetry estimates cover only the sub-peninsular part of Antarctica, consistent with GRACE.
Techniques agree in sign and roughly in magnitude for Greenland, with basin-scale spatial fidelity.
Greenland had small contributions to SLR in the 1990s (-51 \pm 65 \, \text{Gt yr}^{-1}), but recently (2005-10) losing mass at -263 \pm 30 \, \text{Gt yr}^{-1}.
Note: 362.5 \, \text{Gt yr}^{-1} \approx 1 \, \text{mm yr}^{-1} sea-level equivalent.
Antarctica's situation is less clear, with one estimate showing a significant positive mass balance.
Unweighted average indicates Antarctica's mass loss is between -45 and -120 \, \text{Gt yr}^{-1}, with losses in West Antarctica partially offset by SMB gains in East Antarctica.
For Greenland, an independent group compared laser altimetry, mass budget and GRACE estimates over the 2003-09 ICESat period.
Mass budget estimate gave maximum loss rates at -260 \pm 53 \, \text{Gt yr}^{-1}, and GRACE the minimum, at -238 \pm 29 \, \text{Gt yr}^{-1}.
Basin-by-basin agreement between mass budget method and other techniques validates partitioning mass-balance change between discharge and SMB components.
In northern Greenland, mass change was atmospheric in origin, while in the southern part, it was ice dynamics.
New, reconciled IMBIE GRACE estimates of whole Antarctic mass balance are now largely in agreement.
In the Antarctic Peninsula and West Antarctica, the IMBIE estimates from laser altimetry and GRACE are in good agreement, in contrast to East Antarctica.
For East Antarctica, a mass gain of +101 \, \text{Gt yr}^{-1} for 2003-08 has been proposed on the basis of laser altimetry, which is larger than the IMBIE GRACE estimate of +35 \, \text{Gt yr}^{-1}.

Box 2: Grounding Lines and Buttressing

Marine ice sheets (e.g., West AIS) rest on bedrock below sea level, fringed by floating ice shelves.
The grounding line is the contact of the ice sheet with the ocean where the ice mass starts to float.
Ice is discharged across the grounding line into ice shelves, from where icebergs break off (calving).
Grounding-line migration results from the local balance between ice mass and displaced ocean water.
The grounding line advances if floating ice becomes thick enough to ground or retreats if grounded ice thins enough to float.
Simulating grounding-line migration requires including horizontal stress gradients across the grounding zone and high spatial resolution.
Ice discharge generally increases with ice thickness at the grounding line.
For a bed sloping down towards the interior, this may lead to unstable grounding-line retreat (marine ice-sheet instability).
Grounding line is partially stabilized by ice shelves confined laterally or stabilized by locally grounded features (pinning points).
These geometries transmit a back-force ('buttressing') towards the grounded ice sheet.
Thinning of ice shelves reduces drag, leading to increased ice flow across the grounding line and grounding-line retreat.
These mechanisms depend on precise knowledge of the ice-ocean contact geometry.

Recent Advances in Ice-Sheet Modeling

Significant improvements in ice-sheet modeling since IPCC AR4.
Motivated by understanding continuing changes and making realistic projections for the next few centuries.
Primary improvements concern mechanical approximations made to the ice flow equations.
First-generation models were based on the shallow ice approximation (resistance to flow is vertical shear-stress gradients).
More recent models include horizontal stress gradients, classified into four categories:
- Ice shelf/stream models (shallow-shelf approximation): Include horizontal stress gradients, neglect vertical shear stresses.
- Hybrid models: Combine shallow-ice approximation and shallow-shelf approximation.
- Higher-order models: Treat the vertical dimension more rigorously, hydrostatic assumption.
- Full Stokes models: Solve the equations of motion without neglecting any terms.
Spatial resolution of models has improved.
Unstructured grids or adaptive mesh refinement are used to treat the difficulty of resolving ice streams and grounding-line migration.
Satellite and ground-based observations enable improvements.
Quantification of surface velocities and velocity change from satellite interferometry.
Surface elevation change through satellite and airborne campaigns (IceBridge).
High-resolution bedrock and ice thickness measurements.
Ice-sheet model behavior is highly dependent on initial and boundary conditions.
Inverse methods are used to infer the basal drag map, providing agreement between observed and simulated surface velocities.
All these refinements enable models to reproduce present-day observed ice-sheet flow speeds.

Grounding Lines, Sliding and Calving

Warming-induced ice-shelf loss has caused major glaciers and ice streams of Antarctica to speed up.
Oceanic/atmospheric warming leads to ice-shelf thinning or disintegration, which may lead to loss of buttressing, grounding-line retreat, and glacier speed-up.
Observations from the Antarctic Peninsula and the Amundsen Sea Embayment support these mechanisms.
Major theoretical advances show that grounding lines retreat unstably on an upward-sloping bed in the absence of buttressing.
Analytical solutions are available to test and verify marine ice-sheet models.
Models should incorporate horizontal stress transmission across the grounding line and resolve it at a sufficiently high spatial resolution.
GIA influences ice-sheet behavior; Earth's deformation in response to ocean loading affects grounding-line positions.
Ice flow across the grounding line is controlled by inland basal hydrological conditions and processes governing basal sliding and sediment deformation.
Surface melt water reaches the bed and affects basal lubrication.
Basal sliding depends largely on empirical parameterizations based on observations of seasonal variations in ice flow.
Recent developments in understanding calving follow either process approaches or stochastic modeling and fracture theory.
Models will likely continue to rely on empirically based parameterizations of calving.

Future Ice-Sheet Changes

For significantly warmer climates, both the GIS and AIS are projected to lose mass.
General circulation models (GCMs) generally project a small increase of snowfall over both ice sheets.
Mass loss from increasing surface melt will be dominant over the GIS.
Rising temperatures will mainly affect mass loss through increased surface melt in summer.

Positive Feedbacks

Polar amplification of global warming resulting from the decrease of sea-ice extent over the Arctic Ocean.
Positive snow albedo feedback over the ice sheet itself associated with the expansion of the bare ice zone.
Positive elevation feedbacks associated with the thinning of the ice sheet.
Dynamical changes of the GIS due to enhanced lubrication, calving, and ocean warming still remain difficult to predict.
Higher-order ice flow modeling leads to a minimum additional SLR of 6 \pm 2 mm by 2100, with an upper bound of 45 mm when recurring forcing is applied.
Increased ice shelf melt rates of 2 \, \text{m yr}^{-1} lead to 27 mm SLR by 2100 (and 135 mm from a high melt rate of 20 \, \text{m yr}^{-1}).
For Antarctica, the contribution to SLR is predicted to increase logarithmically with rising global temperatures but with little change, and even perhaps a negative contribution, in the next 100-200 years.
Polar amplification resulting from reduced sea-ice coverage seems to be smaller than for the Arctic.
Little surface melt currently occurs, and rising temperatures are not expected to enhance it significantly in the next 100 years.
An increase in snowfall is expected to be more significant, leading to an increase in SMB.
The response of ice-sheet dynamics is twofold, due to increased accumulation and to higher ocean temperatures.
Two models produce ice-sheet thickening over East Antarctica and increased ice flux at the grounding line due to higher snowfall.
A continental-scale Antarctic ice-sheet model assessment is lacking.
Process-based modeling of parts of the West AIS results in a SLR contribution of 27 mm by 2100 for a modest grounding-line retreat of 25 km.
An alternative method leads to a SLR contribution of 130 mm by 2100.

Other Contributions to SLR

The global average rate of SLR over the past few decades is about 2-3 \, \text{mm yr}^{-1}.
Estimates of the global contribution from glaciers and ice caps (GICs) to SLR in the IPCC AR4 were underestimates.
Satellite gravimetry has recently been used to estimate the global contribution of GICs to SLR.
A consensus estimate combining GRACE, laser altimetry, and the extrapolation-based method has very recently enabled reconciliation of the disparate global estimates of wastage from GICs.
The consensus value is 0.71 \pm 0.08 \, \text{mm yr}^{-1} during 2003-09.
Ocean thermal expansion (OTE) is a major component of the SLR observed during the late twentieth century.
A recent sea level budget indicates that OTE contributed ~40% of the observed SLR since 1970 and ~30% since 1993.
Multi-decadal rates for OTE in the upper 700 m are 0.71 \pm 0.10 \, \text{mm yr}^{-1} for 1970-2011 and 0.85 \pm 0.20 \, \text{mm yr}^{-1} for 1993-2011.
Recent estimates for total terrestrial water storage changes during 1993-2008 give values ranging from -0.08 \pm 0.19 \, \text{mm yr}^{-1} to 0.10 \pm 0.20 \, \text{mm yr}^{-1}.
Table 1 summarizes the recent and current contributions to SLR.
OTE appears as the main current contributor to SLR, closely followed by the large ice sheets and the GICs.

Box 3: Glaciological Versus Geodetic Method

GIC mass-balance estimates by the glaciological method are based on extrapolation of in situ measurements.
Estimates by the geodetic method are based on repeated mapping of glacier surface elevations to estimate volume changes.

Conclusions and Outlook

During the past 20 years, the AIS and the GIS have been losing mass.
There are still disagreements between the numbers that come from the mass-balance retrieval techniques, particularly for East Antarctica.
Gravimetry and laser altimetry will have GRACE and ICESat-2 follow-on missions.
Antarctica as a whole is losing mass slowly (assessed to be contributing 0.2 mm yr-1 sea-level equivalent by IMBIE), Greenland, the Antarctic Peninsula and parts of West Antarctica are together losing mass at a moderate (~1 mm yr-1 sea-level equivalent) rate today (~70% of this mass loss is from Greenland) and rates for each are becoming increasingly negative.
For the past decade, the collective sea-level contribution from the ice sheets is similar to those from each of GICs and oceanic thermal expansion.
Although the West AIS is most probably going to continue to contribute to SLR (although the amount is poorly constrained), the sign of the contribution of the East AIS over the next century is uncertain.
Improved knowledge of key ice-sheet thresholds would support climate policy decisions.

Important Challenges

A need for upscaling parameterizations to allow low-resolution models to represent crucial processes better.
The nonlinearity of basal drag and its dependency on basal hydrology remains a concern.
Time-dependent evolution of basal drag is not yet fully implemented in operational models.
A further vital step will be to couple improved ice-sheet models with atmosphere/ocean models and GIA models.