Ice-Ocean-Atmosphere Interactions & Processes

The paucity of synoptic-scale observations of ice flow with high spatial and temporal resolution limits our ability to rigorously test ice flow models. Sub-annual timescale changes in glacier flow dynamics can now be observed within reasonable study periods and provide important constraints on the mechanics of glacier systems.  Understanding the mechanics of glacier ice and the ice-bed interface is essential for reliable projections of future potential states of major ice sheets. 

Here we extend to the time domain methods for inferring surface velocity fields in 3 spatial dimensions using synthetic aperture radar (SAR) and optical data. We apply this original methodology to spatially and temporally dense SAR data collected over Rutford Ice Stream, West Antarctica, in order to generate some of the first-ever truly 4D velocity fields. Rutford provides an ideal proof-of-concept location because of its well-documented, large amplitude tidal-timescale ice flow variations and relatively stable multiannual flow speeds. SAR data were collected over 9 months, beginning in August 2013, using COSMO-SkyMed. All 4 satellites collected data nearly every pass in ascending and descending orbits and along two separate beams (which define incidence angles). This observational scheme resulted in 32 unique tracks and more than 1500 usable two-component (along-track and line-of-sight) displacement fields, calculated using speckle tracking methods. We use all of these displacement fields to infer the secular velocity field and the amplitude and phase values for a family of 3 sinusoidal time functions, all in 3 spatial dimensions. The periods of the sinusoidal functions—M2 (12.42 hours), O1 (25.82 hours), and Msf (14.77 days)—correspond to the primary observable periods in vertical and horizontal ice flow variability on Rutford. 

The resulting 4D displacement fields elucidate the spatial characteristics of the response of ice flow to ocean tidal forcing. Consistent with GPS data collected inland and immediately seaward of the grounding zone, which separates grounded ice from floating ice, we observe strong modulations (~ 20%) in horizontal ice flow rates over fortnightly timescales. The primary period of these modulations corresponds to Msf, the beat frequency of the lunar and solar semi-diurnal tides, which have the largest vertical amplitudes of all tidal constituents. Our results indicate that amplitudes of tidal response are stronger over the ice shelf than over grounded ice by up to a factor of 3 and that the ice shelf leads the grounded ice in response to tidal forcing. Horizontal Msf-period flow variability is transmitted upstream at a mean rate of approximately 28 km/day and is almost completely damped beyond 80 km upstream of the grounding zone. Using ice flow models we show that ice in the margins of Rutford for tens of kilometers upstream of the grounding zone is more viscous than previously supposed from laboratory and field observations in other regions. Enhanced viscosity may allow changes in ice shelf buttressing to be transmitted upstream directly through the ice column, which is not possible with less viscous ice. 

Arctic sea ice has been reduced drastically while Antarctic sea ice has been stable or even slightly increasing.  These different behaviors in Arctic and Antarctic sea ice constitute a challenging paradox to be explained by the cryosphere science community in particular as well as the climate science community in general.  The polar sea ice paradox is a key issue in climate science to be convincingly resolved through appropriate accounting of controlling factors.  Our new understanding of Arctic and Antarctic sea ice trends in a changing climate can therefore advance Earth system models to achieve projections with a significantly reduced uncertainty. 

We first present a review of a number of hypotheses that have been proposed to explain the polar paradox, including: (a) Stratospheric ozone depletion may affect atmospheric circulation and wind patterns such as the Southern Annular Mode, and thereby sustaining the Antarctic sea ice cover; (b) reduction of salinity and density in the near-surface layer may weaken the convective mixing of cold and warmer waters, and thus maintaining regions of no warming around the Antarctic; (c) decrease in sea ice growth may reduce salt rejection and upper-ocean density to enhance thermohalocline stratification, and thus supporting Antarctic sea ice production; (d) melt water from Antarctic ice shelves and icebergs collects in a cool and fresh surface layer pre-conditioning and shielding the surface ocean from the warmer deeper waters, and thus leading to an expansion of Antarctic sea ice; (e) wind effects may positively contribute to Antarctic sea ice growth; and (f) Antarctica lacks of additional heat sources such as warm river discharge to melt sea ice as is the case in the Arctic.

Beyond the above that contribute to explaining how there can be more ice growth and less ice loss in the Antarctic compared to the Arctic, we have identified factors that consistently protect and persistently maintain the stability of Antarctic sea ice.  Moreover, these Antarctic factors are operating opposite to factors that help accelerate sea ice loss in the Arctic.  To understand the differences, we compare and contrast Arctic and Antarctic sea ice properties and trends from current observations with data from satellite scatterometer (e.g., QuikSCAT, Oceansat-2), radiometer (e.g., AMSR-E, SSMIS), and operational sea ice products derived from these and other satellite sensors (multispectral/optical, active/passive microwave sensors).  Based on these results, we identify current gaps in sea ice observations and develop necessary algorithms using satellite data in conjunction with available field observations.  Results on the sea ice cover and its dynamics from the new satellite algorithms will be presented.  We combine these results with ancillary data to characterize the ocean and atmosphere in the Polar Regions and present a new explanation to address the Arctic and Antarctic sea ice paradox.  Finally, we propose necessary future observations from NASA, ESA, and other international space agencies together with new field campaigns designed to quantify these different factors.

Sea surface height (SSH) is poorly observed in the Arctic due to limitations of conventional observation techniques. We utilise data from the Envisat and CryoSat-2 missions to present the first monthly estimates of Arctic Ocean SSH from satellite radar altimetry and combine this with GRACE ocean mass to estimate steric height. SSH estimates from altimetry agree well with tide gauges and estimates of geopotential height from Ice-Tethered Profilers. The large seasonal cycle of Arctic SSH (amplitude ~5cm) is dominated by seasonal freshwater fluxes and peaks in October-November. Overall, the annual mean steric height in our study region increases between 2003-2012 before falling to ca. 2003 levels between 2012-2014. The total secular change in SSH between 2003-2014 in our study region is then dominated by a net increase in ocean mass. The well-documented doming of SSH in the Beaufort Sea dominates non-seasonal SSH variability and is revealed by Empirical Orthogonal Function analysis to be concurrent with SSH reductions in the Siberian Arctic. Ocean storage flux estimates from altimetry agree well with high-resolution modelled results, demonstrating the potential for altimetry to elucidate the Arctic hydrological cycle.

For several years, radar altimeter measurements from low resolution mode altimetry missions have been processed in Arctic sea ice regions. Generally, waveforms from leads are identified and retracked with dedicated retrackers that are most of the time empirical or sometimes based on a Gaussian model. These retrackers differ from physical ocean retrackers (based on the Brown model for example) historically implemented in ground segments. However, care must be taken to avoid introducing estimation discontinuities between open ocean (processed by Brown retracker) and Arctic Ocean (processed by empirical or Gaussian retrackers).

Thanks to the CNES PEACHI project and the ESA Sea Level CCI project, Ka-band AltiKa waveforms and Ku-Band Envisat/RA-2 waveforms have been respectively processed in order to retrieve sea level height estimates in the Arctic Ocean regardless of the sea ice presence. Peaky waveforms coming from leads or polynyas have been identified thanks to waveform classification methods and a new retracking algorithm has been developed based on an ocean model including the mean square slope of the surface in its formulation. This new analytical adaptive 4-parameter solution retracks efficiently ocean waveforms as well as peaky waveforms from leads, accounting for the major instrumental characteristics (point target response, antenna gain pattern …) and ensuring the continuity between open ocean and ice covered regions.

We propose to show the excellent performances obtained with this retracker on Envisat/RA-2 and Saral/AltiKa data allowing us to provide unprecedented SLA maps over regions of major importance for Mean Sea Level closure budget studies.

Since the break-up of the Antarctic Peninsula Larsen A and B ice shelves in 1995 and 2002 respectively, attention has focussed on the possible role of surface melt on ice-shelf stability. In regions where surface ponding occurs, hydrofracture mechanisms are thought to have triggered widespread catastrophic breakup. However, this picture is in conflict with the concept that enhanced melting and refreezing of percolating meltwater may warm the ice and improve its fracture toughness, making the ice shelf in theory more resistant to break-up. Whichever effect is found to be dominant it is imperative to accurately map past melt, to monitor current conditions, and to relate melt to meteorological forcing events.

Just south of the former Larsen A and B Ice Shelves, Larsen C Ice Shelf (LCIS) is the largest on the Antarctic Peninsula and the fourth largest in Antarctica. Here melt ponds have been forming in the northern inlets since at least 2001. Surface melt at these locations is strongly linked to the occurrence of föhn winds descending from the Antarctic Peninsula mountains. These winds bring about episodes of positive surface air temperatures and clear skies with enhanced shortwave solar radiation.

Whilst melt ponds are visible in high-resolution optical satellite imagery when cloud and solar illumination conditions permit, more regular year-round detection of surface melt depends on suitable microwave imagery. The presence of water in the snowpack results in a significant reduction in backscatter when imaged by scatterometer or synthetic aperture radar (SAR) instruments, allowing melt to be readily detected. Data acquired in wide-swath mode by the ASAR instrument on board Envisat allowed the melt patterns and interannual variability on LCIS to be mapped in unprecedented detail. The results demonstrated that regions of high annual melt duration were highly correlated with previously published patterns of firn densification, and with the likely spatial influence of the föhn winds.

Föhn events can be identified by their propensity to push sea ice offshore and open leads between the shelf edge and the sea ice; open water which is easily identifiable in microwave imagery. In SAR imagery we find a correspondence between open water leads and episodes of above average melt intensity in the northern inlets and use this correspondence to determine the likely past and future impact of changes in the synoptic-scale wind fields in the vicinity of the peninsula.

We will also investigate the potential for high-resolution backscatter data acquired by the recently launched Sentinel 1a SAR instrument to extend the ASAR record and allow trends in surface melt over the LCIS to be quantified.