Search

The Paris Agreement aims to limit global mean warming in the twenty-first century to less than 2 degrees Celsius above preindustrial levels, and to promote further efforts to limit warming to 1.5 degrees Celsius1. The amount of greenhouse gas emissions in coming decades will be consequential for global mean sea level (GMSL) on century and longer timescales through a combination of ocean thermal expansion and loss of land ice2. The Antarctic Ice Sheet (AIS) is Earth’s largest land ice reservoir (equivalent to 57.9 metres of GMSL)3, and its ice loss is accelerating4. Extensive regions of the AIS are grounded below sea level and susceptible to dynamical instabilities5–8that are capable of producing very rapid retreat8. Yet the potential for the implementation of the Paris Agreement temperature targets to slow or stop the onset of these instabilities has not been directly tested with physics-based models. Here we use an observationally calibrated ice sheet–shelf model to show that with global warming limited to 2 degrees Celsius or less, Antarctic ice loss will continue at a pace similar to today’s throughout the twenty-first century. However, scenarios more consistent with current policies (allowing 3 degrees Celsius of warming) give an abrupt jump in the pace of Antarctic ice loss after around 2060, contributing about 0.5 centimetres GMSL rise per year by 2100—an order of magnitude faster than today4. More fossil-fuel-intensive scenarios9result in even greater acceleration. Ice-sheet retreat initiated by the thinning and loss of buttressing ice shelves continues for centuries, regardless of bedrock and sea-level feedback mechanisms10–12or geoengineered carbon dioxide reduction. These results demonstrate the possibility that rapid and unstoppable sea-level rise from Antarctica will be triggered if Paris Agreement targets are exceeded.

While twentieth century sea-level rise was dominated by thermal expansion of ocean water, mass loss from glaciers and ice sheets is now a larger annual contributor. There is uncertainty on how ice sheets will respond to further warming, however, reducing confidence in twenty-first century sea-level projections. In 2019, to address the uncertainty, the Intergovernmental Panel on Climate Change (IPCC) reported that sea-level rise from the 1950s levels would likely be within 0.61–1.10 m if warming exceeds 4°C by 2100. The IPCC acknowledged greater sea-level increases were possible through mechanisms not fully incorporated in models used in the assessment. In this perspective, we discuss challenges faced in projecting sea-level change and discuss why the IPCC's sea-level range for 2100 under strong warming is focused at the low end of possible outcomes. We argue outcomes above this range are far more probable than below it and discuss how decision makers may benefit from reframing IPCC's terminology to avoid unintentionally masking worst-case scenarios.

Inference of CO2surface fluxes using atmospheric CO2observations in atmospheric inversions depends critically on accurate representation of atmospheric transport. Here we characterize regional‐scale CO2transport uncertainties due to uncertainties in meteorological fields using a mesoscale atmospheric model and an ensemble of simulations with flow‐dependent transport errors. During a 1‐month summer period over North America, transport uncertainties yield an ensemble spread in instantaneous CO2at 100 m above ground level comparable to the CO2uncertainties resulting from 48% relative uncertainty in 3‐hourly natural CO2fluxes. Temporal averaging reduces transport uncertainties but increases the influence of CO2uncertainties from the lateral boundaries. The influence of CO2background uncertainties is especially large for column‐averaged CO2. These results suggest that transport errors and CO2background errors limit regional atmospheric inversions at two distinct timescales and that the error characteristics of transport and background errors should guide the design of regional inversion systems. Accurate estimates of regional‐scale CO2surface fluxes are essential to improve our understanding of the carbon cycle and to verify human CO2emission inventories. CO2surface fluxes can be inferred from atmospheric CO2measurements through inversion methods, which use atmospheric transport models to relate CO2concentration to fluxes. However, previous studies have shown that inversion results can be sensitive to errors in the simulated atmospheric transport. To better understand how to account for such transport errors, we characterize the uncertainties in simulated CO2concentration due to uncertainties in atmospheric transport by running an ensemble of perturbed transport simulations in a regional atmospheric model. Our results show that CO2uncertainties due to transport uncertainties are about half the magnitude as uncertainties due to erroneous CO2surface fluxes while displaying similar spatial and temporal patterns. Transport uncertainties are reduced when CO2is time averaged, but at the same time the influence of uncertainties in the CO2background concentration is increased at longer timescales. Thus, the flux signals in regional inversions are degraded by transport errors and CO2background errors at different timescales, and it is imperative to properly account for these errors to obtain reliable regional‐scale CO2flux estimates. We quantify CO2transport uncertainties due to uncertainties in meteorological fields at the regional scale and submonthly timescalesAtmospheric CO2estimates that are strongly influenced by CO2surface fluxes tend to also have large transport uncertaintiesThe signal‐to‐noise ratio in in situ CO2is limited by transport errors at subweekly timescales and CO2background errors at longer scales

We present a high‐resolution shear wave velocity model of Greenland's lithosphere from regional and teleseismic Rayleigh waves recorded by the Greenland Ice Sheet Monitoring Network supplemented with observations from several temporary seismic deployments. To construct Rayleigh wave group velocity maps, we integrated signals from regional and teleseismic earthquakes with several years of ambient seismic noise and used the dispersion to constrain crustal and upper‐mantle seismic shear wave velocity structure. Specifically, we used a Markov Chain Monte Carlo technique to estimate 3‐D shear wave velocities beneath Greenland to a depth of 200 km. Our model reveals four prominent anomalies: a deep high‐velocity feature extending from southwestern to northwestern Greenland that may be the signature of a thick cratonic keel, a corridor of relatively low upper‐mantle velocity across central Greenland that could be associated with lithospheric modification from the passage of the Iceland plume beneath Greenland or interpreted as a tectonic boundary between cratonic blocks, an upper‐crustal southwest‐northeast trending boundary separating Greenland into two regions of contrasting tectonic and crustal properties, and a midcrustal low‐velocity anomaly beneath northeastern Greenland. The nature of this midcrustal anomaly is of particular interest given that it underlies the onset of the Northeast Greenland Ice Stream and raises interesting questions regarding how deeper processes may impact the ice stream dynamics and the evolution of the Greenland Ice Sheet. We present a high‐resolution shear wave velocity model of Greenland's lithosphereOur model provides new constraints on the location of the Iceland plume track beneath GreenlandWe find evidence for a midcrustal low‐velocity anomaly beneath the onset of the Northeast Greenland Ice Stream

The complex flow fields of the Antarctic and Greenland ice sheets deform layers deposited as snow at the ice sheet surface, leaving a record of the regional flow history and/or local transitions in basal boundary conditions within the geometry of ice sheet layers. Ice‐penetrating radar reveals these layers, but radar data interpretations are limited by the challenges of quantitatively and reproducibly comparing observations with model output. We present a conceptual framework that relates along‐track reflector slope to gradients in the steady state velocity field of ice sheets. This method makes effective use of englacial reflectors in regions where it is challenging to image continuous layers and avoids the error propagation inherent to tracer‐transport methods, developing the potential for formal radar data assimilation in future modeling studies. We apply our method to radar data collected at the grounding line of Whillans Ice Stream, where enhanced bed friction produces characteristic reflector slopes reproducible using a higher‐order ice flow model. Radar imagery collected to measure ice thickness in Antarctica and Greenland provides a vast amount of information about the internal structure of the ice sheets. Most notably, these data show that the ice sheets are not simply flat layers of ice deposited as snow at the surface. They are complex, folded masses that record the history of ice flow in the geometry of their internal layers. This study establishes a new interpretation framework that allows us to identify unique characteristics of the ice structures and relate them to their formation mechanism, a task that has been historically difficult to perform. Developing methods like this one, which help to describe and interpret the internal structure of the ice sheets in an automated and quantitative way, improves our understanding of the processes that controlled historic ice sheet behavior, allowing us to make better predictions of future ice sheet behavior. Englacial reflector slopes provide information about lateral and vertical gradients in the vertical velocity field of ice sheetsThe presented algorithm, based on the Radon transform, reliably extracts slope information in data with noisy and discontinuous reflectorsNonhorizontal reflector slopes can be used to uniquely identify boundary condition variability not easily detected by other methods

Amplitude analysis of reflection seismic data reveals the presence of highly variable bed conditions under the main sticky spot and adjacent regions of the Kamb Ice Stream (formerly ice stream C). The sticky spot, which is a zone of bed that imparts high basal resistance to ice flow, is situated on a local topographic high composed of consolidated sediments or sedimentary rock. Any meltwater draining from upglacier along the base of the ice is routed around the sticky spot. The ice over the sticky spot includes, in at least some places, a seismically detectable basal layer containing a low concentration of debris, which locally thickens to 40 m over a topographic low in the bed. The ice‐contact basal material ranges from dilated and highly porous to more‐compacted and stiff, and perhaps locally frozen. The softer material is preferentially in topographic lows, but there is not a one‐to‐one correspondence between basal character and basal topography. We speculate that the 40 m thick frozen‐on debris layer formed by glaciohydraulic supercooling of lake‐drainage events along a basal channel during the former, active phase of the ice stream. We also speculate that loss of lubricating water, perhaps from piracy upstream, contributed to the slowdown of the ice stream, with drag from the sticky spot playing an important role, and with the basal heterogeneity greatly increasing after the slowdown of the ice stream. Hydropotential suggests that upstream water would route around the sticky spot which lies on a local topographic highBed conditions vary from soft deformable till at topographic lows to more consolidated stiffer tillObservation of a debris‐rich frozen‐on ice 40 m above a basal channel

The Greenland Ice Sheet (GIS) contains the equivalent of 7.4 metres of global sea-level rise. Its stability in our warming climate is therefore a pressing concern. However, the sparse proxy evidence of the palaeo-stability of the GIS means that its history is controversial (compare refs 2 and 3 to ref. 4). Here we show that Greenland was deglaciated for extended periods during the Pleistocene epoch (from 2.6 million years ago to 11,700 years ago), based on new measurements of cosmic-ray-produced beryllium and aluminium isotopes (10Be and 26Al) in a bedrock core from beneath an ice core near the GIS summit. Models indicate that when this bedrock site is ice-free, any remaining ice is concentrated in the eastern Greenland highlands and the GIS is reduced to less than ten per cent of its current volume. Our results narrow the spectrum of possible GIS histories: the longest period of stability of the present ice sheet that is consistent with the measurements is 1.1 million years, assuming that this was preceded by more than 280,000 years of ice-free conditions. Other scenarios, in which Greenland was ice-free during any or all Pleistocene interglacials, may be more realistic. Our observations are incompatible with most existing model simulations that present a continuously existing Pleistocene GIS. Future simulations of the GIS should take into account that Greenland was nearly ice-free for extended periods under Pleistocene climate forcing.

We present a comprehensive ice‐penetrating radar survey of a subglacial embayment and adjacent peninsula along the grounding zone of Whillans Ice Stream, West Antarctica. Through basal waveform and reflectivity analysis, we identify four distinct basal interfaces: (1) an ice‐water‐saturated till interface inland of grounding; (2) a complex interface in the grounding zone with variations in reflectivity and waveforms caused by reflections from fluting, sediment deposits, and crevasses; (3) an interface of anomalously low‐reflectivity downstream of grounding in unambiguously floating areas of the embayment due to basal roughness and entrained debris; and (4) a high‐reflectivity ice‐seawater interface that occurs immediately seaward of grounding at the subglacial peninsula and several kilometers seaward of grounding in the embayment, occurring after basal debris and grounding zone flutes have melted off the ice bottom. Sediment deposition via basal debris melt‐out occurs in both locations. The higher basal melt rate at the peninsula contributes to greater grounding line stability by enabling faster construction of a stabilizing sediment wedge. In the embayment, the low slopes of the ice bottom and bed prevent development of a strong thermohaline circulation leading to a lower basal melt rate and less rapid sediment deposition. Thus, grounding lines in subglacial embayments are more likely to lack stabilizing sediment deposits and are more prone to external forcing, whether from the ocean, the subglacial water system, or large‐scale ice dynamics. Our conclusions indicate that subglacial peninsulas and embayments should be treated differently in ice sheet‐ocean models if these models are to accurately simulate grounding line response to external forcing. Radar basal reflectivity in a grounding zone embayment is lower than that expected from an ice‐seawater interfaceRadar amplitude and waveform modeling suggests that basal reflectivity is reduced by entrained sediment and basal roughnessA reflectivity increase downflow of the grounding line suggests that sediment deposition in grounding zones occurs via basal debris melt‐out

For common offset radar data, there is no clear way to disentangle path effects from reflector characteristics, so efforts to determine the physical properties at the bed using reflection amplitude are inherently limited by the constraints on englacial attenuation. In this study, we identify the theoretical considerations required for interpreting bistatic radar surveys and use data collected on the Northeast Greenland Ice Stream and Kamb Ice Stream to compute local attenuation profiles. We found that failing to correct for angle‐dependent controls on return power (including antenna directivity, the reflection coefficient, and refractive focusing) can bias the computed attenuation rates as much as 30 dB/km for reflectors at 1 km depth. Because the radiation characteristics are the dominant source of uncertainty in our data, we recommend either a simplified survey design for the future (where the antennae are decoupled from the ice surface) or additional data collection to constrain the near‐field permittivity and its effect on the radiation pattern. Depth‐averaged attenuation rates computed using common midpoint methods for deep reflectors yield values >10 dB/km higher than attenuation rates computed using common offset techniques with the same data. We attribute these anomalously high attenuation rates to additional wavenumber‐dependent (and therefore, angle‐dependent) interferences between subwavelength reflectors. Common midpoint (CMP) radar surveys can be used to constrain englacial attenuationThe source and survey configurations introduce significant uncertainties in attenuation rates, especially at wide‐angles in shallow iceRadar CMP surveys tend to overestimate electromagnetic attenuation in glaciers, due to subwavelength reflector interferences and roughness

Ice‐penetrating radar and GPS observations reveal a perennial firn aquifer (PFA) on a Svalbard ice field, similar to those recently discovered in southeastern Greenland. A bright, widespread radar reflector separates relatively dry and water‐saturated firn. This surface, the phreatic firn water table, is deeper beneath local surface elevation maxima, shallower in surface lows, and steeper where the surface is steep. The reflector crosscuts snow stratigraphy; we use the apparent deflection of accumulation layers due to the higher dielectric permittivity below the water table to infer that the firn pore space becomes progressively more saturated as depth increases. Our observations indicate that PFAs respond rapidly (subannually) to surface forcing, and are capable of providing significant input to the englacial hydrology system. We discovered a perennial firn aquifer on an Arctic glacierThe firn aquifer becomes more water saturated with increasing depthDepth to firn water table varies by several meters in response to surface melt

Rapid retreat of the Larsen A and B ice shelves has provided important clues about the ice shelf destabilization processes. The Larsen C Ice Shelf, the largest remaining ice shelf on the Antarctic Peninsula, may also be vulnerable to future collapse in a warming climate. Here, we utilize multisource satellite images collected over 1963–2020 to derive multidecadal time series of ice front, flow velocities, and critical rift features over Larsen C, with the aim of understanding the controls on its retreat. We complement these observations with modeling experiments using the Ice‐sheet and Sea‐level System Model to examine how front geometry conditions and mechanical weakening due to rifts affect ice shelf dynamics. Over the past six decades, Larsen C lost over 20% of its area, dominated by rift‐induced tabular iceberg calving. The Bawden Ice Rise and Gipps Ice Rise are critical areas for rift formation, through their impact on the longitudinal deviatoric stress field. Mechanical weakening around Gipps Ice Rise is found to be an important control on localized flow acceleration and the propagation of two rifts that caused a major calving event in 2017. Capturing the time‐varying effects of rifts on ice rigidity in ice shelf models is essential for making realistic predictions of ice shelf flow dynamics and instability. In the context of the Larsen A and Larsen B collapses, we infer a chronology of destabilization processes for embayment‐confined ice shelves, which provides a useful framework for understanding the historical and future destabilization of Antarctic ice shelves. The Antarctic Ice Sheet is the largest source of uncertainty in projecting future sea‐level rise. This is due to our limited understanding of drivers and mechanisms triggering tipping points in ice‐sheet instability, and knowledge gaps regarding the retreat and disintegration of ice shelves. Understanding processes leading to ice shelf destabilization is critical to improving estimates of future Antarctic mass loss because of their important role in stabilizing ice flow. We studied Larsen C Ice Shelf changes using satellite data collected over 1963–2020, and conducted modeling experiments to elucidate the observed linkages between front retreat, flow acceleration, and rifts. We find the development of rifts near ice rises to be an important control on Larsen C front calving and flow acceleration in the past six decades. Rifts can affect each other by causing mechanical weakening and modifying stress fields. To predict future dynamical changes, it is necessary to account for this feedback and capture how ice rigidity changes over time in response to rift growth. If Larsen C retreats to the extent that the compressive arch ceases to exist, it will resemble the precollapsed Larsen B ice shelf, producing widespread flow accelerations in response to the backstress loss from ice rises. Multidecadal satellite images and ice shelf modeling experiments were used to examine dynamic changes of Larsen C during 1963–2020Rift development near ice rises is an important control on ice shelf retreat and flow acceleration before the compressive arch is reachedCapturing the time‐varying effects of rifts on ice rigidity is needed to make realistic simulations of future ice shelf change Multidecadal satellite images and ice shelf modeling experiments were used to examine dynamic changes of Larsen C during 1963–2020 Rift development near ice rises is an important control on ice shelf retreat and flow acceleration before the compressive arch is reached Capturing the time‐varying effects of rifts on ice rigidity is needed to make realistic simulations of future ice shelf change

The ability to monitor transient motion along faults is critical to improving our ability to understand many natural phenomena such as landslides and earthquakes. Here, we usedata from a GPS and seismometer network that were deployed to monitor the regularly repeating glacial earthquakes of Whillans Ice Stream, West Antarctica to show that a unique pattern of precursory slip precedes complete rupture along the bed of the ice stream. Additionally, we show that rupture can be independently tracked by increased levels of microseismic activity, including harmonic tremor, that are coincident with the onset of slip at any location, thus providing a remote means of monitoring stress and rupture propagation during the glacial earthquakes. Whillans Ice Stream glacial earthquakes begin with a nucleation phaseSliding may be tracked by microseismic emissionsHarmonic tremor accompanies glacial earthquakes

Palaeoclimatic data document a sensitive climate system subject to large and perhaps difficult–to–predict abrupt changes. These data suggest that neither the sensitivity nor the variability of the climate are fully captured in some climate–change projections, such as the Intergovernmental Panel on Climate Change (IPCC) Summary for Policymakers. Because larger, faster and less–expected climate changes can cause more problems for economies and ecosystems, the palaeoclimatic data suggest the hypothesis that the future may be more challenging than anticipated in ongoing policy making. Large changes have occurred repeatedly with little net forcing. Increasing carbon dioxide concentration appears to have globalized deglacial warming, with climate sensitivity near the upper end of values from general circulation models (GCMs) used to project human–enhanced greenhouse warming; data from the warm Cretaceous period suggest a similarly high climate sensitivity to CO2. Abrupt climate changes of the most recent glacial–interglacial cycle occurred during warm as well as cold times, linked especially to changing North Atlantic freshwater fluxes. GCMs typically project greenhouse–gas–induced North Atlantic freshening and circulation changes with notable but not extreme consequences; however, such models often underestimate the magnitude, speed or extent of past changes. Targeted research to assess model uncertainties would help to test these hypotheses.

Using Arctic sea ice concentration derived from passive microwave satellite observations in autumn and early winter over the 1979–2014 period, the Arctic region was objectively classified into several smaller regions based on the interannual sea ice variability through self-organizing map analyses. The trend in regional sea ice extent (RSIE) in each region was removed using an adaptive, nonlinear, and nonstationary method called Ensemble Empirical Mode Decomposition, which captures well the accelerating decline of Arctic RSIEs in recent decades. Although the linear trend in RSIE is negative in all regions in both seasons, there are marked differences in RSIE trends and variability between regions, with the largest negative trends found during autumn in the Beaufort Sea, the Barents-Kara Seas, and the Laptev-East Siberian Seas. Winter weather patterns associated with the nonlinearly detrended RSIEs show distinct features for different regions and tend to be better correlated with the autumn than early winter RSIE anomalies. Sea ice losses in the Beaufort Sea and the Barents-Kara Seas are both associated with a cooling of Eurasia, but in the former case the circulation anomaly is reminiscent of a Rossby wave train, whereas in the latter case the pattern projects onto the negative phase of the Arctic Oscillation. These results highlight the nonuniform changes in Arctic sea ice and suggest that regional sea ice variations may play a crucial role for the winter weather patterns. The changes in Arctic regional sea ice extents (RSIEs) are quantified for the period 1979-2014Arctic RSIE exhibits marked differences in trends and interannual variability in different regionsRSIEs in different regions are associated with distinct patterns of winter atmospheric variability

We present subannual observations (2009–2014) of a major West Antarctic glacier (Pine Island Glacier) and the neighboring ocean. Ongoing glacier retreat and accelerated ice flow were likely triggered a few decades ago by increased ocean‐induced thinning, which may have initiated marine ice sheet instability. Following a subsequent 60% drop in ocean heat content from early 2012 to late 2013, ice flow slowed, but by < 4%, with flow recovering as the ocean warmed to prior temperatures. During this cold‐ocean period, the evolving glacier‐bed/ice shelf system was also in a geometry favorable to stabilization. However, despite a minor, temporary decrease in ice discharge, the basin‐wide thinning signal did not change. Thus, as predicted by theory, once marine ice sheet instability is underway, a single transient high‐amplitude ocean cooling has only a relatively minor effect on ice flow. The long‐term effects of ocean temperature variability on ice flow, however, are not yet known. Pine Island Glacier speed is correlated with ocean temperatureGrounded ice speed slowed by only ~1% despite ~60% drop in ocean heat contentIce speed recovered after the cold‐ocean anomaly ended

Using new data, we build upon the nearly two‐decade long record of observations from Jakobshavn Isbrae to investigate the processes driving its dynamic evolution. While winter flow speed has not increased substantially over the last three winters, there remains a strong seasonal variation in flow speed that coincides with a cycle of summer thinning and winter thickening. We relate changes in glacier speed to geometry through variations in basal traction and horizontal stresses, using ice‐flow models constrained by satellite and airborne observations. These results suggest that the bed provides little flow resistance along the main trough within about 20 km of the terminus. While the loss of buttressing from the retreat of grounded and floating ice likely contributed to the initial speedup, other processes are of comparable significance at seasonal to decadal time scales. From analysis of the models, we hypothesize that thinning‐induced change in basal effective pressure is the dominant process influencing near‐terminus behavior, while diffusive processes drive the upstream response. The apparent need for the terminus to thin to near flotation before it can calve may limit the rate at which retreat occurs. Our analysis of the processes controlling the speed suggests little potential for further large acceleration. Thinning and elevated speeds may continue at rates similar to present, however, putting the glacier on course to retreat to the head of its deep trough in about a century, at which point it likely would stabilize with a thinner terminus. The speedup of Jakobshavn Isbrae resulted from several processesThe speedup did not result from buttressing aloneThe glacier likely will not speed up much more over the next century

The apparent long‐term persistence and short‐term variability of the Siple Coast ice streams of West Antarctica are tied to regional thermal conditions and local basal lubrication. Numerical simulations indicate that the flux of latent heat in a throughgoing hydrologic system fed by melt beneath thick inland ice maintains the lubrication of fast‐moving modern and ice‐age ice streams despite their tendency to freeze to the bed, and would allow additional thinning and grounding‐line retreat.