A new Greenland subglacial topographic DEM derived from MCORDS data by application of mathematical morphological algorithms


Corresponding author: Ute Herzfeld

Corresponding author e-mail: uch5678@gmail.com

Subglacial topography is an essential boundary in any ice-dynamic model. Because the subglacial topography of the Greenland ice sheet is complex and regionally variable, but density and quality of radar observations vary widely throughout Greenland, the mathematical approach to estimation of a DEM matters. Especially the representation of outlet glacier troughs presents a challenge, yet the presence of the deep troughs causes the acceleration of outlet glaciers in a dynamic model and influences the distribution of basal water. In this paper we present an approach to derive a new subglacial topographic DEM and the resultant DEM. The DEM is based on MCORDS radar data collected by the University of Kansas Center for Remote Sensing of Ice Sheets (CReSIS) since 1993, as part of NSF projects and in recent years as part of NASA’s Operation IceBridge. The approach combines several algorithm steps, including data aggregation, geostatistical analysis, identification of regions of outlet glacier troughs, and topological and mathematical morphological methods for modeling of outlet glacier troughs. The set of algorithms is implemented as a fully automated software that allows for dynamic additions of data from recent and future collection campaigns.


Determination of anisotropic ice fabric using seismic data

Anja DIEZ, Olaf EISEN, Coen HOFSTEDE, Ilka WEIKUSAT, Ulrich POLOM, Thomas BOHLEN, Astrid LAMBRECHT, Christoph MAYER, Heinrich MILLER

Corresponding author: Anja Diez

Corresponding author e-mail: anja.diez@awi.de

Knowledge about crystal anisotropy is mainly provided by crystal-orientation fabric (COF) data from ice cores. To gain a broader understanding about the distribution of crystal anisotropy in ice sheets and glaciers we use seismic measurements. Two effects are important: (i) sudden changes in COF lead to englacial reflections and (ii) the anisotropic fabric induces an angle dependency on the seismic velocities and thus also recorded travel times. We present a framework to connect COF data with the elasticity tensor and thus determine seismic velocities and reflection coefficients for cone and girdle fabrics from ice-core data. These results are compared with VSP measurements from Antarctica to validate the overall approach. Normal spread reflection seismic data are used to show the large influence of crystal anisotropy on the normal moveout (NMO) velocity of P-waves. These anisotropic NMO velocities can be determined from the elasticity tensor with help of the Thomsen parameters for P-waves as well as SH-waves. The discrepancy introduced by an isotropic assumption using stacking velocities as depth conversion velocities is in the range of 7–8 % for P-waves, but only about 1% for SH-waves. Hence, using velocities for the depth conversion that were derived during the stacking process is no longer valid for compressional waves. However, with knowledge about the depth of reflections from other independent datasets, such as radar data or borehole depth, it is possible to determine the Thomsen parameters. We demonstrate that the analysis of normal spread reflection seismic data in combination with radar data gives a tool for determining the anisotropic ice fabric of glaciers and ice sheets. This is an important contribution to constrain results from the upcoming generation of anisotropic ice-flow models by remotely sensed data.


The bedrock topography of Starbuck Glacier, Antarctic Peninsula, as derived from ground-penetrating radar


Corresponding author: Daniel Farinotti

Corresponding author e-mail: daniel.farinotti@gfz-potsdam.de

Knowledge of the continent-wide bedrock topography of Antarctica has recently been improved by the publication of the BEDMAP2 compilation. However, at the regional scale, there remain large uncertainties. This is especially true for the Antarctic Peninsula, in which the complex topography and its small-scale variability hamper a reliable extrapolation of the few ice thickness measurements available so far. The most obvious way of alleviating this problem is to increase the spatial coverage of direct measurements. In this contribution, we present measurements from ground-penetrating radar (GPR) for Starbuck Glacier (65.62° S, 62.30° W), which we use for deriving a complete bedrock topography of the glacier. After the break-up of the Larsen B Ice Shelf in 2002, growing interest has been shown in the glaciers flowing into the ice shelves fringing the Peninsula, since flow speed increases by up to 600% were observed following the removal of the buttressing effect of the floating ice. Since the remaining part of the Larsen B Ice Shelf is anticipated to break apart in the next few years, Starbuck Glacier and the other glaciers flowing into this sector give a unique opportunity for addressing questions about the interaction between inland glaciers and ice shelves. However, for doing so, the ice volume involved in the current acceleration needs to be better constrained. The GPR measurements on Starbuck Glacier were collected during the 2012/13 field season by using the British Antarctic Survey Deep-Look Radio Echo Sounder (DELORES). For the survey, the system was configured to work with 20 kW peak power and a central frequency of 3 MHz. In total, 175 km of GPR data were collected, revealing ice thicknesses up to 900 m, i.e. bedrock elevations as low as 400 m below sea level. Cross-over points of individual GPR profiles were used to cross-validate the results and overcome ambiguities where necessary.


Morphology of basal crevasses at the grounding zone of Whillans Ice Stream, West Antarctica


Corresponding author: Robert Jacobel

Corresponding author e-mail: jacobel@stolaf.edu

The transition from limited- or no-slip conditions at the base of grounded ice to free-slip conditions beneath floating ice occurs across the few-kilometers-wide grounding zone of ice sheets. This transition is either an elastic flexural transition from bedrock to hydrostatically supported elevations (often tidally influenced), or a transition from thicker to thinner ice over a flat bed, or some combination of these processes. In either case, ice must flow across a changing stress field, often resulting in brittle deformation, which is manifested as basal crevassing at the ice-sheet base and tidal strand cracking on the ice-sheet surface. Thus the position and morphology of basal crevasses reveal important information about the stress state across this transition. We acquired gridded ground-based radar surveys at two locations of the Whillans Ice Stream grounding zone, one over a subglacial peninsula where the transition to floatation is abrupt and the second over a subglacial embayment where several dynamic subglacial lakes drain to the ocean, likely resulting in episodic high sediment and water flux across the grounding line. Our surveys indicate a complex pattern of basal crevasses: some are related to basal topography, but others more likely are associated with ice flexure across the basal channel carrying water and sediment to the ocean. Owing to the high reflectivity of sea water and the relatively shallow ice thickness, we image off-nadir crevasses where the radar energy is first reflected from the ice–water interface and then from the crevasse, forming a double image together with the direct reflection. For a basal crevasse with shallow dip, this geometry effectively enables imaging the crevasse from both upper and lower faces simultaneously, producing curious results similar to a clothing-store mirror. Similarly, we see returns from some crevasses that are subsequently reflected to the receiver from the ice–water interface, producing a crevasse signature with a reversed phase echo due to the second reflection. In several cases, these crevasse echoes mimic the geometry of a sub-ice ‘wedge’ dipping into the sediment, while in reality the radar never penetrates below the basal interface. Our results indicate that basal crevasses offer a rich but unexploited dataset for diagnosing stress state and salient processes, such as subglacial stress change over drainage channels, across grounding zones, and that special care is needed when interpreting subglacial returns in radar data.


The implications of reflector geometry on radar data acquisition


Corresponding author: Nicholas Holschuh

Corresponding author e-mail: ndh147@psu.edu

The structure of internal layers in ice sheets is used to interpret ice-sheet flow dynamics. The goal of radio-echo sounding is to accurately reproduce that layer geometry. Radar data from Thwaites Glacier and the northeast Greenland ice stream (NEGIS) show that layers whose dip angle exceeds a threshold do not produce a coherent signal in the data. This is likely due to destructive interference in trace stacking and off-nadir backscatter. Reduction of signal amplitude due to destructive interference in stacking is a function of radar center frequency, reflector dip angle and stacked trace spacing. As the stacked trace spacing increases over a dipping horizon, the phase difference between component pre-stack traces increases, resulting in a less coherent stack. Airborne data are more prone to this signal loss given the higher velocity acquisition platform. In addition to destructive interference in stacking, dipping reflectors sample off-nadir portions of the antenna radiation pattern, reducing the signal recorded by the receiver. Imaging reflectors from wide angles also results in longer englacial travel times and thus additional englacial attenuation relative to horizontal reflectors at comparable depths. Both of these effects lead to further reduction in reflection amplitude. Here we use signal amplitudes to interpolate the slope field of the internal layers and reconstruct layer geometries in radar data from Thwaites Glacier and NEGIS. Our results show that it is possible to infer layer angle with reasonable uncertainty for most dip angles and thereby also provide useful data on current/past stress state and the basal properties responsible for internal layer folding even when layers are not directly imaged.


Helicopter-borne ice-penetrating radar system used for mapping ice thickness at Glaciar Pichillancahue-Turbio on Volcán Villarrica in southern Chile

Rodrigo ZAMORA, Jonathan OBERRREUTER, Sebastian CISTERNAS, Guisella GACITUA, Jose URIBE, Andres RIVERA

Corresponding author: Rodrigo Zamora

Corresponding author e-mail: rzamora@cecs.cl

Several glaciers in southern Chile are located on active volcanoes. There is not enough information about the water equivalent volume of these glaciers and their possible changes in relation to the volcanic activity. Volcán Villarrica (39°25′ S, 71°56′ W; 2847 m a.s.l.) in the Chilean Lake District is an ice-capped volcano characterized by a large frequency of eruptions in historical times, permanent degassing and periodic explosions. Given the volcanic activity in the Southern Volcanic Zone of the Andes (SVZ), lahars are a significant hazard for the surrounding environment. Knowledge of ice thickness at Villarrica glaciers is very important in order to estimate the volume of the water equivalent stored on this volcanic mountain. Few ice thickness measurements are available for this volcano and all of them have been done using ground radar systems. We describe a helicopter-borne radar system used to survey the ice thickness of Glaciar Pichillancahue-Turbio, the main glacier of Volcán Villarrica. Radar data were collected in April 2012 using an impulse radar installed onboard a helicopter which operates at a central frequency of 20 MHz. The antenna is an aluminium structure weighing ~350 kg. This structure is hanging at 20 m below the helicopter and is connected to the helicopter cabin by an optical fiber cable. A 3000 V transmitter working at a 5 kHz PRF is connected to the antenna. The radar receiver is equipped with a dual frequency GPS for real-time positioning of the measurements. The survey was conducted at a flying speed of 40 knots and an altitude of 50 m above the glacier surface. The survey was conducted along pre-designed tracks covering near 14 km2 of the glacier (75% of the total glacier area). The resulting maximum ice thickness was 190 m.


Radiostratigraphy of the Greenland ice sheet

Joseph A. MacGREGOR, Mark A. FAHNESTOCK, Ginny A. CATANIA, John D. PADEN, Sivaprasad GOGINENI, Susan C. RYBARSKI, S. Keith YOUNG, Alexandria N. MABREY, Benjamin M. WAGMAN

Corresponding author: Joseph A. MacGregor

Corresponding author e-mail: joemac@ig.utexas.edu

Two decades of airborne ice-penetrating radar surveys of the Greenland ice sheet collected by the University of Kansas have revealed numerous widespread englacial reflectors. Where these reflectors intersect dated ice cores, they are isochronal. To trace this radiostratigraphy efficiently, we developed and applied several semi-automatic methods that predict reflector slope, including measurement of the along-track phase gradient and image processing. A comprehensive radiostratigraphy of the Greenland ice sheet reveals a common depth pattern of reflectors across hundreds of thousands of line kilometers: numerous strong reflectors in the Holocene, a reflector hiatus between ~15–35 ka, a distinct reflector triplet between ~35–50 ka and intermittent deeper/older reflections. In numerous locations in the northern sector of the ice sheet, the radiostratigraphy does not drape smoothly over the observed bed topography and is instead deflected upward, distorting the depth–age relationship of the ice column by up to several hundred meters. This phenomenon is likely due to spatially varying basal conditions. This radiostratigraphy is a new and strong constraint on ice-flow models attempting to reproduce the past and present dynamics of the Greenland ice sheet.


Radar/seismic imaging of a subglacial estuary at the grounding zone of Whillans Ice Stream, West Antarctica


Corresponding author: Knut Christianson

Corresponding author e-mail: christik@stolaf.edu

The most common view of subglacial water flow across ice-sheet grounding zones is akin to a subaerial waterfall, where water enters the ocean due to the steep gradient in subglacial hydropotential with essentially no marine influence inland. However, our radio-echo sounding and active-source seismic surveys image a grounding zone more analogous to an estuary or tidal lagoon at the downstream end of the hydrologic system that links the active subglacial lakes beneath Whillans Ice Stream to the ocean beneath the Ross Ice Shelf. Kinematic GPS and radar data indicate a hydropotential trough upstream of grounding that continues until the ice goes afloat. Immediately upstream of floatation, irregular basal ringing that persists well below the basal interface is consistent with reflections from on- and off-nadir water-saturated sediments, or, more simply, a till delta. Seismic data also indicate prograding sedimentation as the ice goes afloat and show that the hydropotential trough is linked to the ocean by a large subglacial channel, which has an apparent width of 1 km and maximum depth of 7 m. Pressure differences along the trough axis are within a range that can be overcome by tidally induced processes, making interaction of subglacial and ocean water likely. A shallow water column in the embayment (never thicker than 12 m) and low radar basal reflectivity also imply a well-mixed tidal estuary, with complex interaction of subglacial and ocean water and sediment. Our results highlight the need for joint radar/seismic surveys to properly assess the nature and spatial extent of basal conditions in grounding zones, and the need to consider complex interactions of subglacial and ocean water, sediment and tidal processes across a few-kilometer-wide grounding zone in ice-sheet models.


New interpretations of radar profiles with so-called ‘refrozen ice’ in the Antarctic and Greenland ice sheets


Corresponding author: Alexey Markov

Corresponding author e-mail: am100@inbox.ru

Recently, abnormal build-ups were found near the base of ice sheets on the radar profiles in the region of Gamburtsev Subglacial Mountains, East Antarctica, and the northeastern part of the Greenland ice sheet. In some places, the height of these build-ups is up to half of the total ice-sheet thickness. Initially, their origin was explained as refrozen ice. We considered three other options for the formation of these phenomena. Option 1 – diapir. By analogy with geological structures the abnormal build-ups can be identified as ice diapirs. Such interpretation almost completely explains not only their shape but also the genesis. Option 2 – relic ice domes. The abnormal build-ups can be also interpreted as relic ice domes, the cross-sectional shape of which matches the shape of the cross section of the wing. The structure of the flow of ice through the relic dome is similar to that of the flow of air or water masses through the wing. This configuration causes non-laminar forms of ice flow, similar to flow breakdown behind the wing. Option 3 – change of the pressure field and the structure of ice inside the flow. It is well known that radar profiles can display not the flow structure but the structure of the pressure field. In this case the abnormal build-ups can bound the subsurface structure of recrystallization ice, which appeared during the ice flow through the area having special pressure and temperature conditions. Thus, the radar profiles can display the total field of lithostatic and dynamic pressure and recrystallization ice (structure, anisotropy, the orientation of the crystal c-axes, crystal size, etc.) inside the ice sheet.


Relationship between layer boundaries with different dynamics parameters and surfaces of contrast reflection of radar signal in the Antarctic ice sheet

Alexey MARKOV, Pavel ТALALAY, Sergey POPOV

Corresponding author: Alexey Markov

Corresponding author e-mail: am100@inbox.ru

As a result of surface radar profiling of the Antarctic ice sheet along the profile Vostok, Vostok-1, Pionerskaya and Mirny stations, local maxima and minima of reflected radar signal amplitude were determined through the whole section. It was found that the change of relative amplitude of a reflected radar signal with depth defines the Antarctic ice sheet as a layered sub-horizontal structure. A high correlation (0.74) between the change of ice-flow velocity with depth and the change of relative amplitude of the reflected radar signal with depth has been recognized. The boundaries of layers with different flow characteristics in points of inflexion correspond to local functions extremes of relative amplitude of reflected radar signal. This phenomenon was observed in the whole cross section (at Vostok, Vostok-1, Pionerskaya and Mirny stations up to depths of 1920, 450, 450 and 360 m, respectively) over the 1402 km long profile. This allows us to conclude that the layered structure of flow is identified on the radar profiles of the Antarctic ice sheet.


A new radar for measuring the total ice thickness and snow/firn accumulation in Antarctica

José URIBE, Rodrigo ZAMORA, Guisella GACITÜA, David ULLOA, Andrés RIVERA

Corresponding author: José Uribe

Corresponding author e-mail: juribeparada@cecs.cl

In order to measure the total thickness of the ice and the surface snow accumulation in Antarctica, we have designed and built two types of radars as a single system. This new equipment was built at relative low cost compared with other similar systems. Also, both radars can operate synchronously during the same survey without interference between them. The system has: (1) A pulse compression radar to measure the ice thickness, operating at a central frequency of 150 MHz, 200 W of peak power and bandwidth of 20 MHz giving a theoretical maximum penetration range of 4000 m and a thickness accuracy of ~5 m of ice. (2) A frequency-modulated continuous-wave (FM-CW) radar to measure the snow accumulation and the internal snow/firn layers at high resolution. This radar operates at a central frequency of 725 MHz with a bandwidth of 350 MHz and power of 21 dBm. The system uses two separated log periodic antennas for the transmitter and receiver. The first test was conducted in December 2010 at Union Glacier (79°46′ S, 83°24′ W) in West Antarctica, where tractor pulling modules were used where the system was installed. The measurements were geopositioned by GPS receivers and processed in situ by specially designed software. The post-processing was done in Valdivia using the commercial software REFLEX. This survey was supported by the private company Antarctic Logistics and Expeditions (ALE) with which an 80 km traverse along Union and other nearby glaciers (Schanz, Schneider and Balish) was carried out. A maximum ice thickness of 1600 m was detected and a maximum of 80 m of snow accumulation was found. The main aim of these measurements was to map the subglacial topography of a poorly known area where Union Glacier is located. The location is the main hub for airborne and ground explorations of inner West Antarctica, thanks to ALE summer facilities available there including a blue-ice runway and nearby skyway. The data collected with both radar systems are promising, enabling us to prepare a system improvement that allows mounting of the radar onboard airplane platforms to enable long-range glaciological surveys to be performed. In this presentation we will describe the electronic designs of both radars, their main features and some of the results obtained during the first campaign.


Spatial variation of englacial attenuation rate across the grounding zone of Whillans Ice Stream, West Antarctica


Corresponding author: Brian D. Craig

Corresponding author e-mail: craigb@stolaf.edu

Radio-echo sounding is used to infer basal properties via basal returned power. Generally, a bright bed is interpreted to indicate the presence of water whereas a dim bed indicates dry conditions. Interpretations of basal returned power are, however, complicated by englacial attenuation, incoherent backscatter, spatiotemporal aliasing and off-nadir returns. Here we focus on correcting for englacial attenuation, which can be directly addressed through analysis of constant-midpoint profiles (CMPs). We collected two CMPs on Whillans Ice Stream, West Antarctica. The first CMP was acquired over Subglacial Lake Whillans; the second is located ~100 km down-glacier near the likely drainage zone of several subglacial lakes across the grounding line. We estimate depth-averaged attenuation length by fitting the ratio of bed-echo intensity as a function of incidence angle. Furthermore, since our common-offset data frequently include the presence of a long-path multiple, we also calculate attenuation by taking the ratio of the power returned from primary bed-echo to that of the first multiple. These two attenuation estimates agree within uncertainty bounds. We also use the long-path multiple to assess the spatial variability of attenuation. Our results indicate only minor variations in attenuation over our entire area of survey (<5 dB km–1). After correction for attenuation, we calculate basal reflection coefficients. In a subglacial embayment of the ice-stream grounding zone where several subglacial lakes likely drain, our data indicate that a thin brackish layer of subglacial water, ocean water and sediment persists several kilometers seaward of grounding. Beneath a nearby subglacial peninsula, we image a more abrupt transition from grounded to floating ice, indicating less vigorous subglacial drainage and likely more robust ocean-driven melt at the grounding line. These results indicate that basal reflectivity, with proper consideration of englacial attenuation, can be used to assess complex interactions of subglacial and ocean water in ice-sheet grounding zones.


On the errors involved in the estimation of glacier ice volume from ice thickness data


Corresponding author: Javier Lapazaran

Corresponding author e-mail: javier.lapazaran@upm.es

The assessment of glacier thickness is one of the most widespread applications of radioglaciology and is the basis for estimating glacier volume. The accuracy of the measurement of ice thickness, the distribution of profiles over the glacier and the accuracy of the boundary delineation of the glacier are the most important factors determining the error in the evaluation of glacier volume. The aim of this study is to obtain an accurate estimate of the error incurred in the estimation of glacier volume from GPR-retrieved ice thickness data. The errors involved can be split into errors in boundary delineation and errors in computation. The former represents the uncertainty in the definition of the glacier boundary (because of snowpatches covering the terrain surrounding the glacier, debris cover, etc.), while the latter includes all the errors incurred in computing the volume once a boundary has been defined. This study focuses on the computation errors and does not consider the sources of error in glacier boundary delineation. The usual way to estimate the glacier volume is by summation of the products of cell area and average ice thickness for the cell, using a digital ice thickness model obtained by interpolation, at the grid nodes, from the ice thickness measured along the GPR profiles. Therefore, errors in area and errors in thickness both intervene in the estimate of the error in computation. For a given boundary, the inner cells have no error in area and hence the boundary cells are the only ones contributing to the error in area, through the pixellation error. Since the thickness is computed at the measurement points as half of the product of the radiowave velocity (RWV) and the two-way travel time, the error in the ice thickness measurements involves the errors in RWV and the errors in timing. But the ice thickness has to be interpolated, at the gridpoints, from the measured ice thickness (using kriging in our case), which involves an interpolation error. We detail how to evaluate the latter using a function relating the error at a given gridpoint with the distance to the closest GPR profile. We also provide a method to evaluate the degrees of freedom in the measured data, derived from the kriging variogram, which we use for calculating the error in glacier volume by combining the errors in volume at the gridcell level.


The basal environment of Evans Ice Stream, West Antarctica, from radio-echo sounding


Corresponding author: David Ashmore

Corresponding author e-mail: d.ashmore@abdn.ac.uk

Airborne ice-penetrating radar (radio-echo sounding) is the most efficient method for investigating subglacial environments across polar ice sheets. Theoretically, analyses of the shape and amplitude of the basal reflector can yield physical information on subglacial conditions. Most notably, owing to the high relative permittivity of liquid water a high-amplitude reflection indicates a temperate (unfrozen) bed, the diagnosis of which is pertinent for understanding controls on ice dynamics and, in particular, tributary and fast-flow phenomena. During the past decade a substantial quantity of new airborne radar data has been collected over the Antarctic ice sheet. While these data have been compiled into large maps of subglacial topography, their exploitation with respect to characterizing the basal boundary of the ice sheet remains difficult. Perhaps the greatest difficulty is posed by characterizing how the ice itself attenuates the radar signal. In this study we consider this problem using a 150 MHz centre-frequency airborne radar survey of Evans Ice Stream, a major West Antarctic ice stream, collected by the British Antarctic Survey in 2006/07. Using temperature output from a 3-D finite-difference ice-sheet model we derive a spatially varying parameterization of englacial attenuation. We find a clear association with fast-flow regions and a bimodal frequency distribution of returned power, separated by 10–15 dB, consistent with the reflectivity of the subglacial interface being dominated by the presence of subglacial water. In order to develop these results we present a comparison of these data with several geographical properties. We discuss the glaciological and geophysical implications of these observations. This study demonstrates the potential for the exploitation of existing radar datasets using relatively straightforward techniques.


Model meets radar data: about a migrating ice divide in eastern Dronning Maud Land, Antarctica

Reinhard DREWS, Kenichi MATSUOKA, Carlos MARTIN, Denis CALLENS, Frank PATTYN

Corresponding author: Reinhard Drews

Corresponding author e-mail: rdrews@ulb.ac.be

Ice rises are grounded topographic highs in the coastal margin of Antarctica. They originate from a locally elevated bedrock topography and are typically enclosed by fast-flowing ice shelves. Radar data collected near the dome or below the ice divides show that the internal stratigraphy arches upwards due to the non-linear ice rheology, which stiffens ice at low deviatoric stresses. The arch (or Raymond bump) characteristics allow us to deduce the history of the divide position – and with it the history of the flow regime including a potential change in the dynamics of the surrounding ice shelves. We present data from Derwael Ice Rise (70.5° S, 26.5° E) which buttresses and deviates Western Ragnhild Glacier, one of the main ice streams in Dronning Maud Land. Combining different radar systems (400 MHz, 5 MHz) we visualize the bedrock and the internal layering three- dimensionally. The data reveal spatially varying accumulation rates as well as multiple isochrone arches, which appear unrelated to the flat bedrock and exhibit a varying bump-amplitude versus depth function below the current ice divide. More importantly, we also observe relict arches in the flanks, which indicate that the divide most likely migrated to its current position. Using numerical models (higher order and full Stokes) together with the radar stratigraphy and the derived accumulation rates we aim to explain the relict arches as a result of changing boundary conditions induced by a changing geometry of the surrounding Roi Baudoin ice shelf. We hypothesize that the relict arches bear witness to a larger-scale change in ice flow that may encompass variations of Western Ragnhild Glacier. If this holds true, this sector of East Antarctica may be more susceptible to changes than previously assumed.


Applying a composite pattern scheme to clutter cancellation with the airborne POLARIS ice sounder


Corresponding author: Keith Morrison

Corresponding author e-mail: k.morrison@cranfield.ac.uk

The European Space Agency’s (ESA) POLarimetric Airborne Radar Ice Sounder demonstrator (POLARIS) – built, maintained and deployed by the Technical University of Denmark – operates at P-band and features a multi-phase-center antenna for surface clutter suppression. Data for the development and demonstration of surface clutter cancellation methods were acquired in February 2011 during the IceGrav campaign in Antarctica. The POLARIS radar uses a four-element cross-track antenna to control clutter suppression. The 0.96 m element spacing – 1.4 wavelengths – is significantly greater than the spacing required to avoid grating-lobe ambiguities. The ESA funded the present study to investigate and compare different methods for surface clutter cancellation and to implement a software tool to augment the along-track POLARIS processor developed by the ESA, thus improving bedrock detectability. Within this study, Cranfield University carried out an investigation of the performance of a composite pattern approach to the POLARIS clutter suppression. The scheme exploits the principle of pattern multiplication, whereby the required composite four-element array is produced by the convolution of smaller sub-arrays. This has the advantage that it is computationally much simpler to work with two-element arrays and generate nulls in preferred directions, which are retained in the angular response of the final four-element composite array. Other approaches that consider all four elements need to seek solutions to a set of simultaneous equations. The work considered the performance of the scheme with reference to varying input parameters, which included pulse bandwidth, aircraft roll and cross-track terrain slopes. The work started by recognizing that a four-element array can produce three primary nulls (plus higher-order nulls due to the wide element spacing). A general excitation for this particular case was developed to make best use of all the available information at each array element. The optimum solution was found to have two nulls placed towards the clutter angles required, but with the third always placed towards the endfire direction. This positioning of the third null was found to maximize the gain in the nadir direction. Because of grating-lobe effects arising from the large element spacing, placing nulls at the 45° clutter angles caused a null to be also placed at nadir, leading to a singularity at this depth. However, it was found that by degrading nulling performance at this angle the nadir-singularity could be avoided.


Meltwater drainage pathways beneath Institute and Möller Ice Streams, Antarctica


Corresponding author: Hugh Corr

Corresponding author e-mail: hfjc@bas.ac.uk

Subglacial hydrology directly influences the dynamic behaviour of the overlying ice sheet. Meltwater influences the strength of the subglacial sediment and the amount of friction exerted by the bedrock on the ice. Therefore, understanding the distribution and volume of meltwater beneath the ice is essential in resolving ice-flow dynamics. The direct link between ice flow and global sea level means that characterizing and comprehending Antarctic subglacial hydrological networks are of critical importance for predicting how present-day ice sheets may respond to a changing environment. In 2010/11 a high-resolution aerogeophysical survey was acquired over the catchments of Institute and Möller Ice Streams (IIS and MIS), West Antarctica. The data from the airborne ice-sounding radar have provided an unprecedented opportunity to understand the interactions between rugged subglacial topography, the subglacial hydrological system, and the flow regime, structure and physical properties of this sector of West Antarctica. From the revealed bed topography we map the hydraulic potential, which is a function of the elevation potential and the ice overburden pressure. Simulated meltwater pathways that are derived from routing algorithms show distinct regions where the subglacial relief controls the flow rather that the more usual case where the surface relief dominates. However, the two ice streams exhibit the same funnelling characteristic such that the majority of the meltwater that crosses the grounding line is steered into single channels. Despite the unobstructed flow of water to the ocean we find discrete areas where there is evidence of basal accretion. Examination of the basal power-reflection coefficient confirms these interpretations.


Interpolating accumulation rates by means of internal layering detected by airborne radio-echo sounding


Corresponding author: Daniel Steinhage

Corresponding author e-mail: daniel.steinhage@awi.de

Mass-balance studies using the input–output method as well as modelling studies require reliable knowledge of the regional distribution of accumulation rates. Since for the dating of ice, samples have to be taken in the field either in snow pits or by drilling firn, respectively ice cores, there are only few sampling sites and thus little accumulation rate data available from the Greenland and Antarctic ice sheets. Among the time series used for dating of ice cores are measurements of the dielectric properties of the firn and ice cores. Since radar also detects changes in dielectric properties, it is possible to link radar soundings to ice cores. The internal layering thus shows how representative the ice core is and tracing isochrones between two drill sites not only allows us to test the dating of the ice cores based on independent measurements, but also provides an insight into the regional variations of the accumulation rate between the drill sites. The latter can be used to extend considerably the database for compiling datasets of the accumulation rate prior to applying statistical tools for inter- and extrapolating the data for areal datasets. The Alfred-Wegener-Institut (AWI) is operating amongst other instruments a high-frequency frequency-modulated continuous-wave radar system on board of its ski-equipped aircraft. Compared with ground-based measurements, airborne soundings allow us to map large areas in a relatively short time. AWI mapped in 2010 and 2012 the internal structure of the Greenland ice sheet within the 2000 m contour line with a regional focus on the NEEM deep ice-core drill site (77°30′ N, 51°18′ W). The flown profiles also reached the deep ice-core drill site NGRIP and the shallow ice-coring sites of AWI’s North Greenland Traverse. By linking the isochrones detected by the airborne frequency radar to ice cores, new datasets of the regional distribution of accumulation rates will be derived.


Isochrones and snow accumulation patterns using FM-CW radar data on West Antarctica

Guisella GACITÚA, Andrés RIVERA, José Andrés URIBE, Rodrigo ZAMORA

Corresponding author: Guisella Gacitúa

Corresponding author e-mail: ggacitua@cecs.cl

The West Antarctic ice sheet (WAIS) has been considered potentially unstable because its bedrock is well below sea level and its total disintegration could contribute up to 4.3 m to global sea-level rise. Several recent surveys have studied the area with remotely sensed imagery, airborne platforms and ground surveys, this kind of expedition being very difficult due to the remoteness, difficult logistical approach and several meteorological constraints. However, since 2008 the private company Antarctic Logistics & Expeditions (ALE) has been operating at Union Glacier (79°46′ S, 83°24′ W), providing logistical and technical support to scientific activities from this new gateway for the exploration of inner Antarctica. This support allowed CECs to carry out several expeditions from this base camp, including ice thickness, crevasse detection, ice dynamic, mass-balance and snow accumulation surveys. The last oversnow traverse (2010) started from Union Glacier towards the west covering ~80 km throughout four glaciers before reaching the Antarctic plateau. In this traverse, CECs used a synchronized snow accumulation/ice depth radar system. The snow accumulation was registered by FM-CW radar with an output frequency of 50–400 MHz, which was able to collect nearly 80 m of snow depth data. The processed data showed clear isochrones produced by differential annual snow accumulated. Also, a well-defined ice–snow interface was detected in steep valleys where crevasses were precluding some of the surveys. The obtained data were evaluated, confirming the effectiveness and accuracy of the system when measuring snow layer characteristics, thickness, slopes, inflections and continuity. Additionally, we analysed the patterns of accumulation at the joint between two of the main ice masses in order to characterize the ice dynamic effect on the surface snow layer patterns. In this contribution, the FM-CW system will be described and the surveys performed and the results obtained in the region will be shown.


Integrating RES and remotely sensed ice surface data to enhance survey design, mapping and characterization of the ice-sheet bed


Corresponding author: Neil Ross

Corresponding author e-mail: neil.ross@ncl.ac.uk

Although commonly used to identify the locations and spatial extent of subglacial lakes and to map past and present ice-flow regimes, ice surface imagery is typically an oft-overlooked resource when major aerogeophysical surveys are designed, acquired and interpreted. Satellite-derived ice velocity data are now often used in the design of major airborne geophysical campaigns and the gridding of the resultant data (e.g. using mass conservation approaches). We propose and advocate that the careful targeted analysis and application of ice-sheet surface imagery can also provide considerable useful information at both pre- and post-survey stages. Better integration of remote-sensing imagery with radio-echo sounding (RES) data can increase the efficiency of, and enhance the scientific output from, both local and large-scale geophysical surveys of sub-ice conditions. Combining MODIS Mosaic of Antarctica and/or RADARSAT imagery with ground- and airborne RES data, we show that remote-sensing products that characterize the ice-sheet surface contain important, spatially continuous information on bed topography, sub-ice geology and basal conditions. We will outline the opportunities, benefits and limitations of the integration of ice-sheet surface imagery within the design and interpretation of major aerogeophysical campaigns, describing methods through which the information extracted from ice surface imagery can be enhanced and quantitatively analysed. We illustrate the importance of these data with examples from Antarctica and Greenland, discussing the present and past glaciological implications of our findings. Improved methodologies for the analysis of ice-sheet surface data products may unlock the potential for these high-resolution spatially contiguous datasets to enhance gridding of subglacial topography from sparse RES measurements and as input data for numerical ice-sheet models.


IceSAR 2012 campaign result on P-band SAR ice-flow observation


Corresponding author: Chung-Chi Lin

Corresponding author e-mail: Chung-Chi.Lin@esa.int

The IceSAR 2012 campaign using the ESA’s POLARIS instrument was carried out from April to June 2012 over Greenland with a main focus on the so-called K-transect on Russell Glacier near Kangerlussuaq. Two sites were mapped intensively: SHR in the ablation zone and S10 in the percolation zone at about 700 m and 1850 m elevation, respectively. The campaign was organized in support of the BIOMASS satellite mission which was recently selected as the ESA’s 7th Earth Explorer mission following a user consultation meeting held in March 2013. BIOMASS is a P-band synthetic aperture radar (SAR) mission with the primary objective of measuring the world’s forest biomass. One of the secondary objectives of the mission is to observe the world’s ice sheets, in particular for providing observation of ice flows. The POLARIS radar sounder was modified to operate in side-looking imaging configuration in order to simulate the BIOMASS system. The campaign was conducted such as to reproduce all of the acquisition modes of the BIOMASS system, i.e. polarimetry, polarimetric interferometry and tomography. In addition, POLARIS was also operated in the sounding configuration in order to support interpretation of the physical state of the glacier. The campaign was conducted in three sessions separated temporally by, respectively, 17 and 33 days in order to investigate the effect of image decorrelation over time. Tomographic acquisitions were performed by repeating the same flight track ten times with the baselines resulting from the inevitable offset caused by the achievable flight control precision in practice and wind effects. SAR-focused POLARIS data have been processed to velocity maps using two different classes of techniques: (1) offset tracking cross-correlating complex or intensity patches; and (2) differential interferometry eliminating the topographic effect with an external DEM. The derived ice velocities have been compared with in situ ice velocities that are measured with GPS receivers deployed on the ice. The result of the comparisons suggests that above the equilibrium line, where it is difficult to measure ice velocities at higher frequencies such as at L- and C-bands, P-band could offer a valuable complementary capability. In the case of space-based systems, however, uncompensated ionospheric scintillations may inhibit such applications. Regarding the result of the tomographic processing, the SHR data were found to be overall highly coherent independently of the spatial and temporal baselines. This fact suggests the prevalence of a very coherent, dominating scattering mechanism, i.e. superficial scattering. The S10 data show a significantly lower overall coherence than the SHR data. There seems to be evidence of penetration at least up to –50 m beneath the surface in the far range zone.


Ice sheets and sea level – data, models and ways forward


Corresponding author: Richard B. Alley

Corresponding author e-mail: rba6@psu.edu

The ‘unknown unknowns’ of ice-sheet behavior have been shrinking rapidly under the coordinated efforts of surface observations, airborne and satellite remote sensing, and modeling, together with atmospheric, oceanic and geologic investigations around the ice sheets, including paleoclimatic studies. For most ice-sheet regions, it is now possible to place useful limits on likely rates of change, quantify uncertainties and define research plans for reducing those uncertainties. Unfortunately, this optimistic outlook does not apply universally. Sufficient retreat of the Thwaites Glacier grounding zone, for example, could shift a calving front into a region of combined width and water depth larger than any outlet on Earth today, raising physical questions that are not as yet close to being answered and that may prove very difficult to constrain tightly. The community faces the challenge of continuing the highly successful work of reducing uncertainties in well-characterized flow regimes, while identifying and characterizing those physical processes that are not yet well represented in key places. Furthermore, policy-makers would like guidance from plausible scenarios until those physical processes are better represented. The need for coordinated observations and modeling is thus growing, not shrinking.


Performance of a helicopter-borne snow radar system over Alaska glaciers

Alessio GUSMEROLI, Gabriel J. WOLKEN, Anthony ARENDT, Shad O’NEEL, Louis SASS, Christian KIENHOLZ, Chris McNEIL, Daniel McGRATH

Corresponding author: Alessio Gusmeroli

Corresponding author e-mail: agusmeroli@alaska.edu

Snow accumulation measurements have traditionally relied on time-intensive and spatially limited probing methods that may not be appropriate for regional-scale applications. These measurements are necessary to estimate glacier changes, forecast freshwater delivery to riverine and marine ecosystems, improve flood forecasts and to model elastic changes in the Earth’s crust. During the spring of 2012 and 2013, we performed 500 MHz helicopter-borne ground-penetrating radar (GPR) surveys on several glaciers in south-central Alaska, USA. The surveys were designed to obtain spatially distributed measurements of snow accumulation spanning a broad range of continental and maritime climatic zones. Surveys took place prior to surface melting and wetting of snowpack. Airborne surveys were validated using ground-based GPR and common midpoint (CMP) GPR surveys, detailed snow stratigraphy, coring and probing . Preliminary results suggest that the technique works very well in smooth zones of the ablation area where snow-ice reflections are clearly recognizable. Conversely, radar returns are difficult to interpret in regions of rough topography (debris-covered termini and ice falls) and in accumulation areas where the previous summer melt surface is poorly defined. In this presentation we will discuss the performance of this system over various glacier surfaces and at different flying heights and speeds.


Sensitivity of Thwaites Glacier to ice-shelf melting


Corresponding author: Ian Joughin

Corresponding author e-mail: ian@apl.washington.edu

Strong thinning as ice streams have sped up along the Amundsen coast produces ice loss well in excess of that from other regions of Antarctica. Much of the increases in speed appear to be caused by the loss of buttressing as ice shelves have thinned in response to warmer ocean water and subsequent loss of basal traction as the grounding line has retreated. We have developed a finite-element implementation of a prognostic shallow-shelf ice-stream/shelf model, which we have applied to Thwaites Glacier, Antarctica. The model uses an improved bed map from data recently acquired as part of operation IceBridge. We have conducted a number of numerical tests to examine the sensitivity of the glacier to increased melting and surface accumulation. For melt rates comparable with present, the glacier continues to lose mass at roughly its present rate. Strong sub-shelf melt produces stepped retreat of the grounding line by >40 km over 250 years. Examination of the annual thinning rates shows rapid evolution of the spatial distribution of loss over periods of several years (i.e. comparable in length to a typical satellite altimetry mission). In particular, with each episode of grounding-line retreat, a pattern of strong thinning initially develops near the grounding line that then diffuses inland over periods of several years. Only with increased surface accumulation and reduced melting does the glacier stabilize. Thus, it is likely that Thwaites Glacier will continue to lose mass over the next several centuries at a rate largely determined by the amount of warm circumpolar deep water that makes its way to near the grounding line.


Material heterogeneity and rift termination along suture zones in the Larsen C Ice Shelf, Antarctica


Corresponding author: Daniel McGrath

Corresponding author e-mail: daniel.mcgrath@colorado.edu

Rifts are consistently observed to terminate along the lateral edges of suture zones, locations of material heterogeneity that form the bounds of meteoric inflows in ice shelves. The heterogeneity can consist of marine ice, meteoric ice with modified rheological properties due to past shear, or the presence of fracture. Ground-penetrating radar observations on the Larsen C Ice Shelf, Antarctica, investigate (1) the termination of a 25 km long rift in the Churchill Peninsula suture zone, which was found to contain ~45 m of accreted marine ice, and (2) the along-flow evolution of a suture zone originating at Cole Peninsula, to examine the ice column structure near the calving front. Basal mass-balance values are determined using a steady-state mass conservation model and are applied to a flowline model to delineate the along-flow evolution of layers within the ice shelf. Near the calving front, the thickening surface wedge of locally accumulated meteoric ice, which likely has limited lateral variation in its mechanical properties, accounts for ~60% of the total ice thickness. Thus, it is likely the lower ~40% of the ice column, and the material heterogeneities present there, are responsible for delaying the propagation of rifts, and therefore tabular calving events, as demonstrated in the >40 year time series leading up to the 2004/05 calving event for Larsen C. This likely represents a highly sensitive aspect of ice shelves as changes in the oceanic forcing may lead to the erosion of these features.


Leveraging ice thickness measurements with physically based interpolation: a case study of Jakobshavn Isbræ, Greenland


Corresponding author: Andy Aschwanden

Corresponding author e-mail: aaschwanden@alaska.edu

Ice-sheet models are most highly sensitive to uncertainties in ice thickness, stressing the need for accurate estimates of ice thickness. Recent efforts by CReSIS and Operation IceBridge have vastly improved our knowledge of subglacial topography. For ice-sheet modeling purposes, point and along-flight-line measurements of ice thickness must be gridded to fill gaps where no measurements exist. A variety of interpolation methods exist, including: (1) bilinear interpolation, (2) methods using statistical properties of the data such as kriging and (3) physically based interpolation methods. Physically based interpolation methods such as mass conserving beds are governed by the physical principle of conservation of mass and utilize other observations, such as surface velocity and accumulation, and thus result in more accurate estimates of ice thickness where no measurements are available. In a case study of Jakobshavn Isbræ, Greenland, we compare bilinear interpolation, kriging and mass conserving beds. Our results reveal an asymmetrically shaped Jakobshavn trough when using mass conservation, a feature not recovered by conventional interpolation methods. We test the robustness of mass conserving beds by using measurements of surface velocities and surface elevation for multiple years and discuss the limitations of the method. By using physical principles for interpolation, we leverage existing accurate measurements of ice thickness to provide improved initial conditions for ice-sheet models.


Flow history of Thwaites Glacier inferred from radar-detected flowlines and flowbands


Corresponding author: T.J. Fudge

Corresponding author e-mail: tjfudge@uw.edu

Patterns in radar-detected internal layers in glaciers and ice streams can often be tracked several hundred kilometers downstream from their origin. Here we use distinctive patterns detected in the onset region of Thwaites Glacier in the Amundsen Sea sector of West Antarctica to delineate flowlines and flowbands. Flowbands in the onset region contain information about flow over the past 700 years, which is the approximate time for ice to flow along the flowband. Our analysis of flow conditions over century scales gives perspective on recent changes observed on Thwaites Glacier. Along the eastern margin, flow measured with GPS between 2009 and 2010 is rotated outward by about 1° compared with the long-term flow direction. However, such small rotation is within the directional uncertainty of the long-term flow (about 3°); it is not clear that this apparent outward rotation is a response to changes at the grounding line. We use two radar-detected flowlines to define a 110 km flowband in the middle tributary. The ratio of fluxes through gates at the downstream and upstream ends of the flowband is calculated from continuity for a range of values for past thinning and accumulation rate along the flowband. For comparison, we use InSAR-derived surface velocities (from 1996) and estimates of accumulation rate, to define the geometry of the present-day flowband and to calculate the present-day thinning rate and flux ratio. The geometry of the modern flowband is closely similar to the long-term average, but the flux ratio is higher than the long-term average. The simplest explanation for the change is that the modern rate of thinning along the flowband (about 0.52 m a–1) is larger than the long-term average. The method does not allow us to determine when in the past 700 years the rate of thinning increased.


The sensitivity of Greenland force balance calculations to mass conserving reconstructions of the bed


Corresponding author: Jesse V. Johnson

Corresponding author e-mail: jesse.v.johnson@gmail.com

Mass conservation provides the basis for partial differential equation constrained optimization of ice thickness. When combined with radar, velocity and accumulation data, the optimized thickness provides a geometry having better mass conservation and producing lower errors than geometries produced by techniques such as kriging. In this paper we compute a mass conserving bed for Greenland. This bed takes into account all known radar data, as well as best estimates of accumulation, surface velocity and surface elevation. To date, mass conserving beds have been computed on ice assumed to be in geometric steady state. Here, we investigate the role that observed surface rates of change play in ice-sheet geometries derived from mass conservation. The impact mass conserving geometries have on modeled ice-sheet dynamics is not well studied. We consider the force budgets resulting from both kriged and mass conserving beds. While the force budget corresponding to kriged products produces a set of contradictions that likely arise from artefacts in the bed, the mass conserving geometry gives smoother and more plausible results. Beyond providing a new realization of Greenland’s basal topography, our findings are significant for investigators interested in the relative importance of lateral drag, basal drag and longitudinal stresses in explaining the observed trends in surface velocities of outlet glaciers.


Snow properties over McMurdo Sound, Antarctica, charcterized by GPR and airborne measurements


Corresponding author: Wolfgang Rack

Corresponding author e-mail: wolfgang.rack@canterbury.ac.nz

In this study we present ground-penetrating radar (GPR) measurements to investigate snow properties on sea ice and ice shelf in McMurdo Sound in the western Ross Sea. In various field seasons from 2008 to 2011 a commercial GPR system was operated at varying nominal carrier frequencies from 500 to 1000 MHz depending on the research question. The aims of the measurements were to: (1) derive snow morphology, accumulation and compaction; (2) obtain snow thickness on sea ice to convert total freeboard from satellite and helicopter altimeters to ice thickness; (3) validate CryoSat-2 freeboard measurements; and (4) calibrate snowdrift models and validate numerical simulations for snow accumulation. The fieldwork was conducted in coordination with helicopter (HEM-bird) and satellite measurements to obtain information on ice thickness, freeboard and surface characteristics. Ice shelf – Stake farms were established in 2008, with snow stakes separated by 100 m in a regular grid of a size comparable with a satellite radar altimeter footprint. Two stake farms on the ice shelf measured 800 m &mult; 800 m. Snow pit, firn core and snow fork measurements were conducted for reference snow morphology, and GPR measurements were repeated along the regular grid. GPR processing included sharpening of internal horizons by applying a deterministic deconvolution algorithm for matching up layers from repeat measurements. The main results of this work are snow compaction with depth down to about 11 m over the period of 1 year and spatial information on average snow accumulation since 2004. Sea ice – GPR measurements were conducted on sea ice across McMurdo Sound together with simple ruler and snow density measurements as a reference. With the sledge-mounted GPR we detected continuous snow depths if greater than about 15 cm. An additional airborne camera on the HEM-bird allowed characterization of the irregularity of the snow cover. The main outcome is a snow thickness map across McMurdo Sound and statistics on the snow variability as input to a freeboard to thickness conversion, and a sea-ice thickness map of McMurdo Sound.


Buried information: constraining bed geometry and material from the Doppler-dependent radar-scattering function


Corresponding author: Dustin M. Schroeder

Corresponding author e-mail: dustin.m.schroeder@utexas.edu

The morphological, lithological and hydrological basal boundary conditions of ice sheets and glaciers can exert strong, even dominating, control on their behavior, evolution and stability. However, the scales at which the physical processes and observable signatures of this control occur are typically smaller than the spatial resolutions achievable using ice-penetrating radar. Further, the strength of calibrated radar bed echo returns is a combination of both the material (i.e. relative permittivity, conductivity) and geometric (i.e. rms height, rms slope, auto-correlation length) properties of the ice–bed interface. This ambiguity in the relative contribution of material and geometric bed properties, along with uncertainty in englacial attenuation from underconstrained ice temperature and chemistry, also makes definitive assessment of basal conditions from echo strengths extremely difficult. To address these challenges in interpreting geometric and material bed properties at glaciologically relevant scales, we present a new algorithmic approach to measuring the radar-scattering function of the ice–bed interface with varying Doppler frequency by performing range-migrated SAR focusing using multiple reference functions spanning different ranges of Doppler frequencies from the bed. We parameterize this scattering function in terms of the relative contribution of angularly narrow specular energy and isotropically scattered diffuse energy. This specularity content of the bed echo is insensitive to englacial attenuation and is a measure of both the angular distribution of returned echo energy and the geometry of the ice–bed interface at the sub-azimuth-resolution scale. We present an application of this technique to a gridded airborne radar survey over the entire catchment of Thwaites Glacier, West Antarctica. We show how the information in the along-track scattering function of the bed can be used to assess the extent and configuration of distributed water across the catchment and detect the transition of the water system from distributed canals to concentrated channels. We also show how this information can be used to constrain the morphology of basal bedforms and infer the distribution of deformable sediments and exposed bedrocks across the catchment. These applications demonstrate the potential to extract rich information from focusable radar-sounding data to constrain the radar-scattering function as well as the material and geometric properties of the bed.


Basal conditions of Northeast Greenland Ice Stream mapped from joint seismic/radar surveys


Corresponding author: John Erich M. Christian

Corresponding author e-mail: christij@stolaf.edu

Northeast Greenland Ice Stream (NEGIS) is the sole long (~700 km) fast-flowing ice stream of the Greenland ice sheet. The ice stream initiates near the summit dome and terminates in three large overdeepened outlet glaciers that calve into the Greenland Sea, which are subject to both the marine ice-sheet instability and oceanic forcing. Understanding the dynamics of NEGIS is therefore crucial to accurately estimating the future mass balance of the Greenland ice sheet. Although rapid flow is likely caused by high geothermal flux near the onset of fast flow, the mechanism to maintain and constrain streaming flow downstream was unknown. Here we present radio-echo sounding (RES) and active-source seismic amplitude-vs-offset (AVO) data that show that the dynamic effects of streaming flow limit ice-stream extent. Bright radar basal reflections are located within the central trunk of the ice-stream. Dim bands (~20 dB reduction in relative basal reflectivity) surround the central bright area and are spatially coincident with the shear margins. These dim bands are surrounded by another set of relatively bright bands of basal reflectivity. AVO surveys reveal that the NEGIS interior is underlain by several meters of soft, wet and dilatant till. Both shear margins show a harder dewatered bed. We use the AVO results to calculate the basal reflection coefficient for the entire RES survey. Our results indicate that dilatant till is common under ice in streaming flow, whereas the margins are dewatered and coincident with surface depressions. These surface troughs create steep gradients in the subglacial hydropotential that generate parallel ‘slippery’ and ‘sticky’ bands beneath the margins. The ‘sticky’ bands limit ice entrainment and subglacial water flow across the margin and thus restrict further widening, producing the long, narrow and relatively stable ice stream.


Tomographic observation and bedmapping of polar glaciers and ice sheets with IceBridge sounding radar

Xiaoqing WU, John PADEN, Ken JEZEK, Eric RIGNOT, Younggyu GIM

Corresponding author: Xiaoqing Wu

Corresponding author e-mail: xiaoqing.wu@jpl.nasa.gov

We produced high-resolution bedmaps of several glaciers in western Greenland and Antarctica from IceBridge mission sounding radar data using the tomographic sounding technique. The bedmaps cover three western Greenland regions (Russell, Umanaq and Jakobshavn Glaciers) and one Antarctic region (Pine Island Glacier). The ground resolution is 50 m and the average ice thickness accuracy is 10–20 m. There are some void areas within the swath of the tracks in the bedmaps where the ice thickness is not known. Tomographic observations of these void areas indicate that the surface and shallow sub-surface pockets, likely filled with water, are highly reflective and greatly weaken the radar signal and reduce the energy reaching, and reflected from, the ice-sheet bottom. We present these interesting observations and the bedmaps, which can soon be accessed by the public through the National Snow and Ice Data Center website.


Using coupled seismic and radar methods to fully characterize the subglacial environment


Corresponding author: Leo E. Peters

Corresponding author e-mail: lep144@psu.edu

Subglacial morphology plays a critical role in ice dynamics as the potential for rapid ice drainage is defined by how readily the ice stream or glacier can slide over the underlying substrate. Active seismic methods provide the best means of capturing basal conditions by using amplitude-vs-offset (AVO) analysis. However, seismic data only provide information on a small part of the bed, with at most tens of line-kilometer coverage in a given field season. Ice-penetrating radar produces detailed images of basal topography to map hydrologic pathways and can provide information on basal conditions on a regional scale. Here we present the details of a joint seismic-radar analysis of the subglacial environment in the onset region of Northwest Greenland Ice Stream (NEGIS), Greenland. High-resolution seismic AVO profiles were collected at five locations across NEGIS, from which we infer a wet, dilatant, meters-thick sediment layer beneath the central part of the ice stream, with loosely consolidated, dewatered sediments beneath the margins of the ice stream. Synthetic radar waveforms were generated at each of these seismic AVO sites, based on the seismically derived basal conditions and a coupled radar-seismic analysis of the ice column. A regional-scale analysis of the subglacial environment was then conducted by calibrating a radar grid across the onset region of NEGIS to the five synthetic radar waveforms. These observations highlight a drape of unconsolidated sediments across the region, with basal meltwater draining into the main channel, creating a dilatant till layer beneath the streaming ice. This type of joint seismic-radar analysis in glaciated regions can capture these important basal parameters at a scale that is


Firn variability derived from a statistical analysis of airborne ice-penetrating radar over the Thwaites Glacier catchment, West Antarctica


Corresponding author: Cyril Grima

Corresponding author e-mail: cyril.grima@gmail.com

A dry firn layer covers most of the Antarctic ice sheet. Firn characteristics are a function of accumulation rate, air temperature and surface winds. As such, they are indicators of ice-sheet accumulation history and mass balance. To date, most of the observational techniques for firn characterization at depths of a meter or more achieve limited geographical coverage (i.e. ice/firn cores, ground-based GPR). During the aerogeophysical campaign of the 2004/05 austral summer the Airborne Geophysical Survey of the Amundsen Sea Embayment, Antarctica (AGASEA) project surveyed a 15 km grid over a 600 km &mult; 400 km area covering the Thwaites Glacier catchment, West Antarctica, with the High-Capability Radar Sounder (HiCARS) system operated by the University of Texas Institute for Geophysics (UTIG) onboard a de Havilland DHC-6 Twin Otter aircraft. The HiCARS system transmits pulses with a 60 MHz (λ = 5 m) central frequency that are chirped over a 15 MHz bandwidth and 8000 W peak power. One resulting data product is a calibrated radar dataset sampled every ~10 m along the survey tracks that have been coherently integrated and range compressed. In this study, we applied a statistical method to the surface echo in order to separate the coherent (specular) and incoherent (scattered) parts of the signal. We use these estimated components with a backscattering model to derive and map the roughness and real part of the surface permittivity. The resulting permittivity values reflect the physical properties of the first 5 m of the firn. We analyze these results in the context of firn density and/or possibly wetness spatial variability. We observe a ~30 km wide vein of high surface permittivities ~100 km inward from the coastline with a northern boundary that matches a prominent slope break for the surface. We discuss the implications of our results for formation climatological context of catchment-wide firn properties in general and the high-permittivity vein in particular.


Constraining the recent sea-level contributions of Pine Island and Thwaites Glaciers, West Antarctica, using CReSIS ultra-wideband airborne radar systems


Corresponding author: Brooke Medley

Corresponding author e-mail: bmed@u.washington.edu

One of the largest sources of uncertainty in quantifying the glacial contribution to sea-level rise originates from our lack of understanding of spatio-temporal snow accumulation rates. Traditional in situ measurements of the accumulation rate (i.e. using firn cores, snow pits and stake farms) are time-consuming and inadequately capture the complex spatial variations in regional accumulation. We use ultra-wideband airborne radar data to track near-surface internal horizons to calculate spatio-temporal accumulation rates over Pine Island and Thwaites Glaciers along the Amundsen coast of West Antarctica. Here, we combine data from both CReSIS snow and accumulation radar systems to generate a spatially complete high-resolution gridded map of mean accumulation rate, thereby constraining the total mass input into these dynamic glaciers over the past 25 years. We furthermore find the snow radar is capable of imaging annual horizons, an improvement over the multi-year resolution available using the accumulation radar system. Based on the annual accumulation rates generated from the snow-radar echograms, we find no significant trend in the accumulation rate over much of Thwaites Glacier. These data indicate that the recent substantial increase in Thwaites ice discharge to the ocean has not been balanced inland by additional snow accumulation. This suggests the Thwaites contribution to sea-level rise has increased over the past few decades as regional accumulation rates have not increased to offset the accelerating discharge of this glacier.


A survey of techniques for detecting layers in polar radar imagery

Jerome E. MITCHELL, David J. CRANDALL, Geoffrey C. FOX, John D. PADEN

Corresponding author: Geoffrey C. Fox

Corresponding author e-mail: gcf@indiana.edu

The Earth’s rise in temperature can cause significant consequences and affect the subsurface dynamics of the polar regions. In an effort to investigate the stratigraphy and basal conditions in Greenland and Antarctica, the Center for Remote Sensing of Ice Sheets (CReSIS) has used instruments capable of recording the rapidly changing polar ice sheets. The acquired data are manually analyzed with subjective and highly time-consuming approaches, especially when considering large amounts of data. As instrument accuracy improves, automatic methods to support objective extraction of results matching a 1 cm (snow) to 50 cm (bed) improvement are needed. The development of automatic techniques for the analysis of data acquired from Earth’s polar regions will allow for reliable tools, which could analyze a large quantity of data in a timely manner for the scientific community. We will present a comprehensive survey of suitable techniques for identifying layers from varying classes of radar imagery, each from snow (over polar firm and sea ice) and accumulation radars as well as the multichannel coherent radar depth sounder. General-purpose image segmentation has been used for a variety of problems, but, more concretely, methods for tracing near-surface internal layers have been studied using snakes and dynamic time warping while bedrock and surface layers have benefited from both snakes and level sets approaches. Progress towards identifying ice interfaces in snow radar over sea ice using a classification technique, called support vector machines, has also been studied. There still exists, however, an opportunity for more innovative techniques. By discussing current techniques in detail, new approaches can be readily envisioned. Our data domain is in radar remote sensing of polar ice sheets, but understanding automatic approaches could be easily adopted for finding layers in other remotely sensed data.


UWB MIMO SAR for imaging the ice–bed interface of polar ice sheets over a wide swath


Corresponding author: Theresa Stumpf

Corresponding author e-mail: ts9@ku.edu

Understanding how the Earth’s ice sheets will respond to warming is a practical socio-economic concern because of their potential to augment global sea levels by the end of the 21st century. One of the challenges associated with rising sea levels is the increased incidences of flooding due to tidal surge events; policy-makers need reliable sea-level rise projections so that they may make informed decisions about the development of infrastructures for mitigating the risks associated with future surges. The sea-level rise projections for the end of the 21st century that were summarized in the Fourth Assessment Report of the Intergovernmental Panel on Climate Change were characterized by large uncertainties because they did not account for rapid dynamic changes being observed in the cryosphere. Consequently, their contributions to projected sea levels are likely to be much greater once these dynamic processes have been taken into account. Bed topography is an important input parameter in ice-sheet models used to project future sea levels, but observations of bed geometry are lacking in many key areas. Integrating fine-scale topography in key areas of interest within a coarse-resolution digital elevation model (DEM) of the bed has become a priority for members of the ice-sheet modeling community. Data from multichannel ice-penetrating synthetic aperture radars (SARs) can be tomographically processed to three-dimensionally image basal topography. Recent publications addressing multiple-input multiple-output (MIMO) SAR imaging suggest that a multichannel airborne ice sounder/imager with a large cross-track array may be able to leverage MIMO SAR techniques to image basal topography with fine resolution over a wide swath. The scientific community’s need for detailed basal topography in areas of interest justifies a closer look at the usefulness of MIMO SAR for mapping the bed three-dimensionally. The University of Kansas is developing an ultra-wideband (UWB) ice sounder/imager that will operate from a Basler BT-67 aircraft using a 24-element cross-track array. This paper analyzes the potential benefits of two different MIMO SAR modes for wide-swath ice-sheet bed mapping using this particular platform. In the first mode, the swath is partitioned into multiple sub-swaths. Multiple orthogonal narrow transmit beams and narrow receive beams are used to image individual sub-swaths. In the second mode, wide orthogonal transmit beams are used to increase phase center diversity, facilitating the synthesis of narrow receive beams for imaging. A discussion of waveforms will also be addressed.


Development of a spatial data infrastructure (SDI) for storage, access and analysis of multi-layer datasets


Corresponding author: Kyle W. Purdon

Corresponding author e-mail: kpurdon@ku.edu

As the amount of data collected by the Center for Remote Sensing of Ice Sheets (CReSIS) increases, the need for a flexible and expandable geospatial data storage, access and analysis solution was identified. Owing to the range of data products, ever-expanding data volume and the need for a field deployable system, the requirements of the API were that it be expandable, deployable and reliable. We set out to create a system based on the needs of CReSIS, but flexible enough to be used by any organization with multi-layer datasets. Using a Red Hat Linux operating system deployed on an Oracle VM VirtualBox, we created a solution comprised of an Apache web server, GeoServer geospatial data server, PostgreSQL database with the PostGIS geospatial extension, Python server side scripts, and an OpenLayers javascript data access portal served through the web. The CReSIS Matlab image browser, which includes layer tracking and analysis tools, was modified to interface to this system. We have established APIs so that the system can be interfaced from other software packages. The Web Map Service (WMS) and Web Feature Service (WFS) of the GeoServer allow standard (OGC compliant) GIS packages to interface without custom software development. This system is also being developed as a free and open source (FOSS) project maintained by CReSIS so that the international science community can use the system and contribute to its development. We will present the system, currently modeled as the CReSIS Data Portal (CDP), at the meeting. The demonstration will include layer tracking using the CReSIS image browser and layer retrieval and visualization using the web-based access portal. Multiple relevant datasets, including CReSIS ice surface and ice bottom layers, NASA ATM lidar ice surface, University of Texas LayerMap layers and the CReSIS snow radar layers for land and sea ice, will be demonstrated using this system.


Improving the geothermal heat flux estimates in the Greenland ice sheet using radar-detected basal water


Corresponding author: Soroush Rezvanbehbahani

Corresponding author e-mail: soroushr@ku.edu

The presence of water at the base of an ice sheet is a critical boundary condition for ice-sheet and glacier flow modeling. Basal water can decrease the effective pressure, reduce basal friction and consequently cause enhanced flow of ice. Three main heat sources govern basal ice temperature and thus the formation of subglacial water: geothermal heat from Earth’s interior, heat generated from internal deformation and heat from the friction of ice with the glacier bed. For the interior regions of the ice sheets, where ice velocities are small, strain heating and frictional heating can be combined into a basal heat flux augmenting the geothermal heat flux. Although spatial variations in geothermal heat flux are shown to have significant effect on the thermodynamic state of basal ice, it is one of the largest unknowns in ice-sheet modeling. Recent observations from the Center for Remote Sensing of Ice Sheets (CReSIS) airborne radar have delineated regions in Greenland (north of 74° N) where subglacial water is likely to exist. These observations, however, do not correspond well with regions of high geothermal heat flux inferred from tectonic, seismic and magnetic data for the Greenland ice sheet (GrIS). This study seeks to improve geothermal heat flux estimates for the GrIS by comparing modeled basal ice temperatures with observations of basal water from radar-sounding data. We employ a steady-state solution based on the Robin model, adjusted for heat generated through ice deformation. We estimate basal temperature using three spatially varying geothermal heat flux models. Our model assumes that the heat generated due to internal deformation of the ice column is entirely exerted on the base of the ice sheet. The magnitude of the calculated ice-deformational heat is added to the geothermal heat and incorporated into the Robin solution. We conduct a sensitivity analysis on the effect of geothermal heat flux and ice thickness variations on the basal temperature of the ice sheet and compare the results with radar-sounding observations. In the end, the available geothermal heat flux estimates will be adjusted to correspond with the remotely sensed observations of basal water.


UAV-mounted GPR for remote area radioglaciology


Corresponding author: Adrian B. McCallum

Corresponding author e-mail: amccallu@usc.edu.au

UAV technology is rapidly developing and increased payloads are available from more affordable platforms. Similarly, the development of Software Defined Radio (SDR) technology provides opportunities for configuration of lightweight and low-cost ground-penetrating radar (GPR) systems. This discussion paper examines the viability of a UAV-mounted SDR GPR system and presents preliminary field data from remote non-glaciated terrain. The rapid development of utilized technologies suggests that remote sub-surface imaging of previously inaccessible areas will soon be feasible using UAVs.


Determining winter mass balance and the previous year snowline position of the Juneau Icefield, Alaska, using high-frequency ground-penetrating radar


Corresponding author: Seth Campbell

Corresponding author e-mail: seth.campbell@umit.maine.edu

The Juneau Icefield in southeast Alaska covers ~4000 km2 and feeds numerous outlet glaciers. Its location in a coastal mountain range implies a strong coastal to inland precipitation gradient, but other than a few yearly snow-pit measurements collected in this region of Alaska, the spatial distribution of snowfall is not well understood. In July 2012, over 200 km of 400 MHz ground-penetrating radar (GPR) profiles covering an elevation range of ~1000 m were collected across the icefield. The goal was to determine whether GPR methods could be used to map winter mass-balance spatial patterns of glaciers in maritime climates. Radar methods for accumulation mapping are well developed for the polar ice sheets, but deriving winter balance estimates by tracking radar stratigraphy in temperate glacial settings is not as well established. Profiles were collected along the center line and cross sections of the main, northwest and southwest branches of Taku Glacier as well as Mathes, Llewellyn and Demorest Glaciers. Associated measurements of stratigraphy from 16 shallow snow pits were used to help interpret the radar-detected stratigraphy. Radar-detected stratigraphy was visible down to 25 m at high elevations; penetration was reduced at lower elevations, likely because of volume scattering from free water within the firn and ice. Ice lenses and annual layers measured in snow pits correlate well with continuous stratigraphy imaged in GPR profiles. Results show that the point measurements of stratigraphy can be extrapolated over most of the accumulation zone of the icefield to map variations of mass balance with elevation [b(z)]. Radar-detected stratigraphy in the northwest and southwest branches of Taku Glacier show a strong b(z) gradient from the coast towards the mainland; the main trunk of Taku Glacier, which originates from the Mathes–Llewellyn ice divide, shows a similar decrease in b(z) from the divide to the equilibrium-line altitude (ELA). The observations are consistent with mass-balance gradients expected in accumulation zones; the notably high mass balance in the upper catchments of the southwest and northwest branches likely result from their close proximity to the ocean. GPR profiles in the vicinity of the ELA typically show highly reflective ablation horizons beneath the seasonal winter accumulation. We interpret these strong reflectors to be firn-ice transitions from previous years; sequential mapping of these transitions may provide an annual time series of end-of-season snowlines that can be used to infer the balance state of glaciers.


Detection of layer slopes and mapping of discrete reflectors in radio echograms

Christian PANTON

Corresponding author: Christian Panton

Corresponding author e-mail: panton@nbi.ku.dk

An automated method for detecting layers in radio echograms is presented. The method is designed to work with multiple radio-echo sounding (RES) data products. The method has been tested successfully with both CReSIS depth sounders and snow radar. To accurately detect layers, first approximate layer positions are calculated by integrating the local layer slope, then the positions are refined using an iterative optimization. Local layer slope is inferred by computing the deformation between adjacent radar traces using dynamic time warping. The layers are detected using an active contour model or snake, which can be constrained to conserve both echogram features and smooth layers. With this technique it is possible to detect internal layers over distances of several hundred kilometers. The method is tested between the Greenland deep ice cores where the age–depth relation is known.


Effects of firn on determining bed topography of polar ice sheets using radar


Corresponding author: Kenny Matsuoka

Corresponding author e-mail: matsuoka@npolar.no

Radiowave propagation speed in ice sheets is a fundamental parameter to derive ice thickness and thus bed elevation from two-way travel time measured with ice-penetrating radar. The propagation speed primarily depends on the firn density/thickness and secondarily on ice temperature and alignments of ice crystals. Here, we examine firn effects on the propagation speed. First, we present the depth profiles of the propagation speed estimated using different refraction-index models and conclude that variations between models are minor. Second, using the simplest refraction-index model, sensitivities of the depth-averaged propagation speeds are examined in terms of ice thickness and fraction of the air in the ice column. Finally, we apply this method to firn thickness estimates over the Antarctic ice sheet.


Deformation and folds of the basal ice under the Greenland ice sheet

Dorthe DAHL-JENSEN, Sivaprasad GOGINENI, Christian PANTON

Corresponding author: Dorthe Dahl-Jensen

Corresponding author e-mail: ddj@gfy.ku.dk

Improvement of the depth sounding radio-echo sounding over the Greenland ice sheet has made it possible to map the near basal layers that have not been ‘seen’ earlier due to the very high demand of attenuation needed to reach through more than 3000 m of ice. The basal 10% of the ice thickness reveals very disturbed layering in the central and north parts of the Greenland ice sheet. The onset of the disturbances very often seem to coincide with the ice from the climatic inception from the Eemian period to the Last Glacial period around 115 000 years before present. Studies of the ice rheology and deformation tests of the ice reveal big changes of ice crystal size and orientation at this boundary and very different deformation properties in the ice from the warm and cold climatic periods. Based on the ice rheology of the interglacial ice and the glacial ice, constitutive equations are used to evaluate the deformation property differences.


Layer tracing guided by a statistical method

Hugh CORR, Carlos MARTIN

Corresponding author: Hugh Corr

Corresponding author e-mail: hfjc@bas.ac.uk

Radar-echo sounding (RES) of ice sheets shows the presence of layers at depth that are understood to represent former surfaces of the ice sheet. When submerged by subsequent snowfall the result is an isochrone stratigraphy with an architecture that is due to the combined influences of mass balance and ice flow. Therefore, the pattern of the internal layering provides information on the past behaviour of an ice mass. The tracing of layers over large distances has been used to investigate changes in ice flow, calculate past accumulation rates and for the constraint of layer ages from ice cores. The following of a single isocrone in RES datasets is often subjective and therefore prone to error. However, recent numerical models of ice dynamics have been able to incorporate the local layer dip obtained from discrete sections of reflectors. To remove off-axis scattering modern RES instruments are by necessity stable and coherent. Using these characteristics we present a novel method that directly employs the echo-phase to accurately extract layer dips (slopes) and, uniquely, the dip error. Furthermore, by assuming the dip error has a normal distribution we apply a statistical method to trace internal reflecting horizons over large distances. The method not only gives a measure of the layer depth but also a probability distribution of the error estimate. We apply the method to extend the age–depth record from deep boreholes in Antarctica and Greenland and provide explanations for disparities obtained from a simple age–depth model (Dansgaard–Johnsen).


Sounding of subglacial nunatak ridges in West Antarctica using a UHF radar

Cameron LEWIS, Howard CONWAY, John STONE, Perry SPECTOR, John PADEN, Prasad GOGINENI

Corresponding author: Cameron Lewis

Corresponding author e-mail: cameronlewis@ku.edu

We developed an ultra-high-frequency (UHF) radar that operates over the frequency range of 600–900 MHz for surface-based measurements with a virtual antenna array of 16 elements. We used this radar to collect data on ice-covered nunataks in the Pirrit Hills of West Antarctica during the 2012/13 Antarctic summer season. These data were collected to generate fine-resolution 3-D bedrock topography maps for identifying the position and depth of the subglacial ridges of the Harter and John nunataks in preparation for future drilling campaigns. Data were collected over a grid for two areas: one approximately 0.6 km &mult; 0.3 km and the other approximately 1.2 km &mult; 0.5 km. We completed preliminary processing of the collected data, which included coherent integration, pulse compression and geo-synchronization. The processed data were used to generate radar echograms that revealed bedrock features at depths between 100 and 400 m below the surface. We also observed the presence of significant range hyperbolae indicating that a long aperture can be synthesized to obtain fine along-track resolution. Virtual phase center (tomographic) techniques will be applied to obtain fine resolution in the cross-track direction for producing 3-D maps of the bedrock topography. The combination of the radar system and data-processing techniques demonstrates the ability to image the ice–bedrock interface with fine resolution. This radar can also be used for fine-resolution imaging of the ice–water interface of ice shelves. We have successfully sounded 500–600 m thick ice shelves with an airborne version of the radar operating with a single receiver. Time-separated in situ measurements can provide single point analysis of ice-shelf basal melt rates; however, continuous wide-area coverage is needed to accurately evaluate the integrity of an ice shelf. Application of the virtual antenna array to the airborne version of the radar using the aforementioned data-processing techniques, with time-separated data collection campaigns, helps fill this data gap.


Bed topography under Greenland outlet glaciers revealed by Operation IceBridge data


Corresponding author: Mathieu Morlighem

Corresponding author e-mail: mathieu.morlighem@uci.edu

Bed topography, together with ice thickness, is an essential characteristic of glaciers and ice sheets for many glaciological applications, but is challenging to measure remotely. This is especially true for ice-sheet coastal sectors and mountain glaciers, where the ice surface is significantly broken up, the ice substrate is near or at the melting point, and scattering and signal absorption are high. At present, maps of bed elevation and ice thickness are traditionally obtained using airborne radar-sounding profiler data interpolated onto regular grids using geostatistical tools such as kriging. While this approach provides continuous maps of bedrock elevation, it generates products that are not consistent with ice-flow dynamics and are impractical for high-resolution ice-flow simulations. A different approach consists of using the mass conservation equation by combining OIB data with high-resolution ice motion data from interferometric SAR (ALOS PALSAR, RADARSAT-1 and Envisat ASAR). The results reveal overdeepening in the glacier fjords that is not apparent in current maps, and deep subglacial valleys that channelize ice flow to the coast. These features potentially have a significant impact on ice-flow modeling and are mapped for the first time around Greenland using a combination of OIB and InSAR data. These results provide improvements in the bed map of Greenland and guidelines for future deployments. This work was performed at the University of California Irvine and the California Institute of Technology’s Jet Propulsion Laboratory under a contract with the National Aeronautics and Space Administration, Cryospheric Sciences Program, grant NNX12AB86G.


Challenges and limitations in designing and integrating airborne radar antenna arrays used for remote sensing

Emily J. ARNOLD, Richard D. HALE, Jie-Bang YAN

Corresponding author: Emily J. Arnold

Corresponding author e-mail: earnold10@gmail.com

Airborne radars used for sounding and imaging fast-flowing glaciers and ice-sheet margins require advanced signal-processing capabilities and antenna arrays with very low spatial sidelobes to minimize surface clutter that can potentially mask weak bed echoes. Large cross-track antenna arrays with high gains are required to image the bedrock deep below the ice sheets. To facilitate the largest array possible, the antenna elements are often mounted externally to the airframe of the aircraft. By attaching the antenna array to the airframe, the electrical performance is influenced by its motion as in-flight loads cause the structure to. Airframe deformation will translate to the array and cause relative phase errors between elements which greatly degrade the advanced array-beam formation required to suppress surface clutter. One difficulty in designing a support structure (often called a fairing or radome) to house the externally mounted array is that structural and electromagnetic requirements often conflict. Common aerospace materials such as metal and carbon are both conductive and electrically lossy, and the proximity of these materials to antenna elements can greatly reduce their electrical performance. It is often difficult to avoid using these materials as strength and stiffness requirements cannot always be met with electrically transparent materials such as S-2 glass or quartz. As such, structural and electromagnetic design and simulations must be performed concurrently, requiring many iterations. Another challenge of integrating an airborne array onto a platform is the ground plane and aperture size requirements and limitations. Appropriate ground plane offsets from antenna elements increase the size and loads of the structure and have the potential to create a convergent nozzle that can choke the flow causing shock waves to form. Care must be given to the aerodynamics of the fairing. The presence of the external fairing will increase the total drag of the platform and reduce range and endurance. In addition the array must be adequately spaced from propellers and wing tips so as not to subject the structure to the adverse effects of prop wash and tip vortices. The turbulent flow from these features will increase the drag of the fairing as well as the structural fatigue. This paper will highlight several theoretical and experimental studies that address aircraft integration effects on radar used for remote-sensing of ice sheets. These studies demonstrate problems and solutions to aircraft integration effects that degrade the high gain, low sidelobe antenna array performance required for sounding the most difficult regions of ice sheets.


Radar sounding of temperate ice masses


Corresponding author: J. Mouginot

Corresponding author e-mail: jmougino@uci.edu

Glaciers in many parts of the world, including Antarctica, are changing rapidly in response to climate change. Yet for many of these glaciers we do not have good ice thickness data. This problem is exacerbated in regions where ice is temperate, i.e. at the melting point throughout the ice column, which is the case for Alaska and Patagonia, but also in Iceland, New Zealand, the European Alps and some of the coastal regions of Greenland. We developed a low-frequency (2.5 MHz) radar for the sounding of temperate ice masses that circumvents the limitations of higher-frequency radars. The Warm Ice Sounding Explorer (WISE) was flown in Alaska in October 2008 and March 2012, in Greenland in May 2008 and March 2010 and in Antarctica in January and December 2009. Here we present the radar design and the data collected from the southeastern coast of Alaska in 2008 and 2012, where it yields the first comprehensive GPS-tagged measurements of ice thickness over major glaciers and icefields (Hubbard, Columbia, Bering, Malaspina). The results show bed returns to depths of up to 1150 m, including on the never-sounded deep Bagley Icefield in the Mount St Wrangell/St Elias range. The radar reveals that Bagley Icefield and Bering Glacier exhibit deep bed pockets that are prone to subglacial meltwater ponding, which probably plays a central role in the surge nature of Bering Glacier. On Columbia Glacier, we demonstrate that the main branch of the glacier is below sea level until 40 km above the reference position of the glacier prior to its retreat, but a western branch remains below sea level for another 20 km upstream. Application of this radar sounder to other temperate glaciers in Alaska or temperate ice masses elsewhere such as southern Greenland will improve our knowledge of ice thickness in these regions, which in turn is critical to understand glacier dynamics, its contribution to sea-level change and projections of its evolution in a changing climate.


Flow dynamics of Byrd Glacier, East Antarctica


Corresponding author: C.J. Van der Veen

Corresponding author e-mail: cjvdv@ku.edu

Byrd Glacier is a fast-moving outlet glacier transecting the Transantarctic Mountains, funneling an estimated 20.6???? Gt a–1 of ice originating on the East Antarctic plateau into the Ross Ice Shelf through a fjord that is ~100 km long and ~20 km wide. The glacier has been the subject of glaciological investigations since the early 1960s, including a comprehensive assessment of balance of forces on the lower trunk by Whillans and others (1989) using surface elevations and ice velocities derived from repeat photogrammetry in the late 1970s. That study, as well as subsequent more recent studies, was limited by lack of detailed information on the bed topography under the glacier. In 2011–2012 the Center for Remote Sensing of Ice Sheets (University of Kansas) conducted extensive airborne radar sounding for mapping the bed under Byrd Glacier, thereby allowing re-evaluation of results from earlier studies and, in particular, to investigate the relation between ‘sticky spots’ and basal relief. The present study aims to investigate flow dynamics and essentially represents an update of the study of Whillans and others (1989). Force-balance calculations reveal large variations in the along-flow component of driving stress that are muted by gradients in longitudinal stress such that basal drag is less variable spatially. The most pronounced sticky spot is located and the downstream end of a basal overdeepening, while smaller regions of high basal drag are co-located with a bed ridge transverse to the flow direction. On the large scale, gradients in longitudinal stress play a small role in balancing the driving stress, and flow resistance is partitioned between basal and lateral drags. Confirming earlier results, there is a significant component of driving stress in the across-flow direction resulting in non-zero basal drag in the direction perpendicular to ice flow. This is an unrealistic result and we propose that there are spatial variations in ice strength similar to those found on other glaciers.


Crevasse patterns in the catchment of Byrd Glacier, East Antarctica


Corresponding author: Logan C. Byers

Corresponding author e-mail: loganbyers@ku.edu

Complex patterns of surface crevasses are observed in extensive fields dispersed throughout the catchment of Byrd Glacier, Antarctica, using WorldView multispectral visible-band satellite imagery at 0.6 m horizontal resolution. Thorough mapping and orientation analysis reveals that some fields are composed of numerous intersecting and interacting crevasse sets, which experience differential deformation across their domains. A comparison of observed crevasse field occurrence to surface elevation demonstrates that fields are located primarily in regions where the ice surface dips highly downstream and uncrevassed areas tend to occur at positive or less negative downstream slopes. The spatial extent of some of these regions without observable crevasses is equivalent to shallow surface basins that maintain a zone of surficial snow deposition. Analysis of RADARSAT imagery with a horizontal resolution of 25 m and a penetration depth of ~8 m demonstrates that crevasses persist at depth as they are advected through shallow surface basins. As ice advects out of the basin, crevasses continue to be covered but may display some sagging of newly deposited snow. The continued presence of crevasses from high in the catchment past the grounding line may indicate that infilling with snow provides a resistance to closure but has little affect on opening. Infilling would maintain crevasses as structural weaknesses within the ice as advection and differential velocities subject the ice to varied stress states. Surface crevasses have previously been used as indicators and predictors for numerous dynamical properties of glaciers. The findings of this study therefore have implications for automated feature tracking in both visible and long wavelength imagery, the sensing and determination of complex glacier flow paths and controls on glacier flow, and accumulation rates and histories in ice-sheet conditions. This work also challenges the assumption that glacier ice is a structurally homogeneous material and demonstrates a connection exists between observable surface structures and basal conditions that dictate ice motion.


Ice thickness and density of the Pine Island Glacier floating ice shelf


Corresponding author: Kiya Riverman

Corresponding author e-mail: klw367@psu.edu

Pine Island Glacier (PIG) in West Antarctica flows into an ice shelf that has been thinning since at least 1990. This has resulted in a 34% increase in the flow speed of Pine Island Glacier from 1996 to 2006 due to reduced buttressing forces on grounded ice. With the potential for this glacier to contribute dramatically to future sea-level rise, there is strong interest in modeling the current and future melt dynamics of the PIG floating ice shelf using coupled ocean–ice models. These models require ice thickness and density data. We present high-resolution gridded ice density and thickness data for use in future modeling work. We have used a digital elevation model (DEM) generated from stereographic pairs of images from high-resolution (WorldView) imagery. We determine variations in ice density and degree of flotation from precise ice thickness (from seismic and radar data) and elevation measurements. In January 2013, ice thickness measurements were collected from ground-based reflection seismology, ice-penetrating radar and hot-water drilling. By inverting for ice density and degree of flotation and interpolating across the shelf, we have used these data to generate an ice thickness map from the WorldView DEM.


North East Greenland Ice Stream basal conditions from reflection seismology and ice-penetrating radar


Corresponding author: Kiya Riverman

Corresponding author e-mail: klw367@psu.edu

North East Greenland Ice Stream (NEGIS) is the largest ice stream on Greenland, draining ~8.4% of the ice-sheet area. Because of the unique potential for this ice stream to influence inland ice dynamics through its large catchment area, it is important to include the potential changes in NEGIS flow speeds in future mass-balance studies of Greenland. We present the results of the first ground-based geophysical survey of NE Greenland in order to describe the basal conditions that support rapid flow and maintain a stable flow field on this unusual ice stream. Reflection seismic data indicating the presence or absence of lubricating sediments and sediment thickness are compared with ice-penetrating radar bed reflection results. Variations in radar bed amplitude appear to be related to changes in sediment properties, indicating ‘sticky’ high-basal-resistance regions under the ice-stream margins.


Insight into the existence and implications of ‘surface waves’ on Byrd Glacier


Corresponding author: Leigh Stearns

Corresponding author e-mail: stearns@ku.edu

Byrd Glacier has one of the largest catchment basins in Antarctica and drains ~20.5 km3 of ice into the Ross Ice Shelf annually. Despite various studies since the late 1970s focusing on flow dynamics of Byrd Glacier, there is still little consensus about its ice–bed coupling and ice-flow relation to bed topography. In this study, we will utilize new bed and surface topography data, in conjunction with high-resolution velocity maps, to model the importance of bed topography and its impact on glacier flow. Our results yield insight into the dynamical flow regime and stability of Byrd Glacier in response to different external forcings. Reusch and Hughes (2003) hypothesized that as Byrd Glacier transitions from sheet flow to stream flow, the ice surface undergoes changes in surface slope (‘surface waves’) that appear to be unrelated to bed topography. The implication is that these surface waves reflect variations in the coupling between ice and the bed and that they may move as individual ice columns and migrate through the glacier. According to this hypothesis, surface waves represent regions of high longitudinal tensile stresses on the ice surface and the dominant resistance for the flow of Byrd Glacier is due to these longitudinal stress gradients. This theory contrasts several studies that conclude the driving stress of Byrd Glacier is primarily resisted by isolated regions of high basal drag (‘sticky spots’). This research investigates the surface wave theory, which has not been tested previously due to the lack of high-resolution bed topography data. From November 2011 to January 2012, the Center for Remote Sensing of Ice Sheets (University of Kansas) collected bed topography and ice thickness data over ~55 000 km2 of Byrd Glacier and its catchment. This dataset, in combination with satellite-derived velocity data, will be used to explore the origin and evolution of surface waves and their relationship to bed topography and longitudinal stress patterns. A correlation of the surface slopes (waves) and the bed topography slope will be performed to determine if wave location is independent of bed relief. Wave migration requires ice–bed decoupling, which could be an indication of Byrd Glacier having a thawed bed in contrast to a frozen one.


Investigation into the cause and effect of flow variability on Helheim Glacier, Greenland


Corresponding author: Leigh A. Stearns

Corresponding author e-mail: stearns@ku.edu

Helheim Glacier in southeast Greenland has undergone rapid and complex changes in its flow dynamics in the past two decades. Interannual observations reveal large changes in ice velocity, surface elevation and terminus position. Recent satellite and ground-based observations also reveal substantial seasonal and daily flow variability. It is unclear, however, what drives glacier variability at different timescales, and how perturbations of different magnitudes propagate along the length of the glacier. In this study, we explore changes in velocity patterns that occur over three different time periods: interannual, seasonal and daily. Force-balance calculations allow us to explore the relative importance of resistive forces and how these vary at different epochs. A flowline perturbation model is used to investigate the response of Helheim Glacier to changes in back pressure at the calving front and to changes in subglacial drainage and water pressure. Velocity data comes from repeat satellite (ASTER, Landsat, InSAR, TerraSarX) and ground-based (ground-based interferometry, GPS) observations; surface elevation changes are derived using ICESat and ATM data (Csatho and others, 2013); bed topography is from CReSIS radar depth sounding.


Spatial variability of interior ice-sheet accumulation determined with an FM-CW radar and connections to the NAO

David BRAATEN, Prasad GOGINENI, Claude LAIRD, Susanne BUCHARDT, Hilary BARBOUR, Anthony HOCK

Corresponding author: David Braaten

Corresponding author e-mail: braaten@ku.edu

Snow accumulation is a key component of ice-sheet mass balance, and formulating a consistent picture of regional mass-balance trends is important to providing good estimates of the contributions of the ice sheets to sea-level rise. In addition, identifying teleconnections between ice-sheet accumulation and atmospheric indices (e.g. the North Atlantic Oscillation (NAO)) may lead to alternative methods of estimating future accumulation trends using global and regional climate models. A regional time series of annual snow accumulation in central Greenland has been obtained from continuous radar measurements using an ultra-fine-resolution FM-CW surface radar system developed by CReSIS. This radar was operated along a 375 km ice divide traverse between NGRIP and NEEM in 2007 and the data were processed to reveal annual internal layers in the ice. Using snow density profiles from NGRIP and NEEM, the differential radar signal propagation speed through the ice was determined and used to determine annual layer depth. The density profiles were also used to determine water equivalent accumulation for each year. The radar-determined annual accumulation was compared with independent assessments of annual layer depths and accumulation from the ice-core sites. Annual layers over this region were tracked and the water equivalent snow accumulation determined for each accumulation year between 1957 and 2006. When the entire traverse is segmented into smaller sub-regions, we find contiguous sub-regions that have a statistically significant correlation with the NAO. Both ice-core data and gridded atmospheric modeling of precipitation over Greenland do not show significant correlations with NAO. The results presented highlight the value of continuous internal layer radar mapping that provides good spatial averaging and a direct observation of annual accumulation.


Physical and environmental impacts on Helheim Glacier flow and calving rate


Corresponding author: Leigh A. Stearns

Corresponding author e-mail: stearns@ku.edu

Helheim Glacier, southeast Greenland, has undergone large changes in flow dynamics in the past decade, resulting in a near doubling of its mass flux into Sermilik Fjord and the ocean. An overdeepening at the glacier bed is observed in both radar echograms from the coherent radar depth sounder (CORDS), as well as bed topography models that consider the mass conservation of ice. When the terminus of the glacier retreats into the overdeepening, the increase in water depth may play a role in controlling the terminus position and calving dynamics. Retreat to this bed depression was observed in the summer of 2004, but no quantitative observations have been made in regards to the impact of increased water depth on retreat rate, calving frequency or calving intensity. We identified the major calving events and compared their magnitude with water depth for 2002 through 2012. Terminus position is determined on a daily basis using MODIS visible imagery and an automatic detection process to calculate the area and position of the terminus. Data from an automatic weather station (AWS) were used to investigate the relationship between surface melt production and calving rate. Fjord characteristics may also have an impact on the frequency of calving events and glacier flow. Icebergs and sea ice combine to form the ice melange, which may cause buttressing at the glacier front. Using MODIS time series of images, the change in presence of ice melange was quantified and compared with calving rate.


Using radar layer data in ice-sheet models


Corresponding author: Richard Hindmarsh

Corresponding author e-mail: rcah@bas.ac.uk

Deep radar layers emanate from surface deposition of electrolytes, and the subsequent motion and deformation of firn and ice determines the geometry of these passive tracers. They contain information about the integrated history of the ice dynamics since their deposition and can in principle be used to infer accumulation rate, melt rate, ice rheology and ice thickness history. This paper will examine how this principle meets with practice, with the aim of seeing how well we can answer the following questions. How should data best be presented to modellers? The geometrical information in radar layers is completely encoded in their slope at any point and these slopes are the raw data from which line picks are made. The implication is that models should be matched against slope data, rather than reconstituted picks, as the picking method has the potential to introduce systematic errors that are difficult to account for in any inversion process. The choice of data style presented to model has implications for the following questions. How accurately can we reconstitute accumulation rate and thickness variations in space and in time? What are the prospects for inverting for ice rheological parameters? Which other techniques can be usefully used with isochrone layes to improve these estimates? The answers to these questions depend both on the dynamics and mechanics of the problem and also on the data errors in the radar layers. The consequences of these for the way data should be presented to modellers will be addressed.


Clutter and cross-track slopes: interpreting airborne and orbital radar-sounding data in glacial environments

John W. HOLT, Scott D. KEMPF, Duncan A. YOUNG

Corresponding author: John W. Holt

Corresponding author e-mail: jack@ig.utexas.edu

Surface clutter can greatly hamper the interpretation of radar-sounding data in areas of strong topographic relief. This is especially relevant for valley glaciers in temperate environments where low-frequency airborne radars are used with a trailing-wire dipole antenna. It is also a significant issue for orbital radar sounders at Mars and will be for any future radar sounders in orbit around Earth. We have developed radar clutter prediction tools that incorporate surface topography, radar trajectory, aircraft attitude and antenna pattern. We have demonstrated the effectiveness of these techniques for glaciers in Antarctica studied with VHF and HF radars operated by the University of Texas Institute for Geophysics. Near-global topographic datasets now such as the ASTER GDEM are adequate for most purposes and facilitate a generalized clutter simulator. Higher-resolution lidar-based DEMs can be nested to include glacier surface lineations and other small-scale topography that produce clutter in some cases. The cross-track migration of putative subsurface echoes onto high-resolution imagery can further aid in identifying potential clutter sources below the resolution of available DEMs. We have implemented similar algorithms for the Shallow Radar (SHARAD) on Mars Reconnaissance Orbiter to study mid-latitude glaciers and polar ice. We have found that in addition to predicting clutter, determining the location of first returns is a critical part of the interpretation process for orbital data. Even for relatively smooth surfaces such as the polar caps of Mars, minor cross-track slopes can cause the first Fresnel zone to deviate from the spacecraft ground track by tens of kilometers, causing significant delays in the position of reflectors from both the surface and subsurface. Our algorithm accurately predicts the first-return location and allows for a geometric correction to be calculated. Combined with clutter analysis, this results in the accurate interpretation and positioning of subsurface interfaces. The methods developed can be applied to both existing airborne radar systems and future Earth-orbiting radar sounders.


Tracing internal radio-echo sounding layers

RES Tracing Community represented by Dorthe DAHL-JENSEN

Corresponding author: Dorthe Dahl-Jensen

Corresponding author e-mail: ddj@gfy.ku.dk

Tracing of the internal radio-echo sounding layers observed in glaciers, ice caps and ice sheets opens vast opportunities for interactions with research related to flow and deformation of ice, finding the oldest ice and mapping the time-dependent accumulation rates. A workshop was held in Copenhagen on 6–10 May 2013 where tracing methods were presented and discussed. In addition, use of the radio-echo sounding data in related research areas was presented and discussed. Internal radio-echo sounding layers can be dated by connecting them to ice cores and putting an age on a layer multiplier sour ability to exploit isochrone characteristics for numerous questions. The state of the ability to trace layers and the visions for the future of the development will be described as well as the applications of the traced layers. (The abstract can be refined after the workshop has been held.)


Southeast Greenland snow accumulation derived from airborne IceBridge accumulation radar, ground-based radar and firn-core data

Clément MIÈGE, Richard R. FORSTER, Jason E. BOX, Lora S. KOENIG, Evan W. BURGESS

Corresponding author: Clément Miège

Corresponding author e-mail: clement.miege@gmail.com

Constraining Greenland ice sheet snow accumulation rates remains important because recent climate warming has almost tripled its mass loss since 1958 and accumulation is the only component that could potentially offset this loss. Accumulation needs to be further quantified over the last decades to observe a possible increasing trend in accumulation synchronous with climate warming. We focus on the southeast sector of the ice sheet that currently contains the highest accumulation rates, contributing about one-third of the entire ice-sheet annual accumulation. This work presents accumulation rates derived from the accumulation radar collected in spring 2010 and 2011 by NASA’s Operation IceBridge mission. This new dataset has the capability to map internal firn layers with high spatial coverage and a 30 cm vertical resolution. In conjunction, two Arctic Circle Traverses in 2010 and 2011 (ACT-10 and ACT-11) collected up to 400 km of ground-based radar data and a total of five firn cores (depths between 25 and 60 m). From the two field campaigns, dated isochrones are produced and compared with the airborne radar internal layering recorded simultaneously over the same transects. We note a good agreement between the ground- and airborne-derived accumulation rates. Topographic undulations influence accumulation rate variability of up to 20% over less than 10 km and is not yet taken into account by regional snow accumulation models. We observe greater accumulation rates and greater interannual variability from the interior toward the southeast coast, in agreement with existing climate models. Over the last decades, no significant increase in accumulation is observed with this radar-derived accumulation data. A notable limitation of this accumulation mapping technique is the presence of surface melt that complicates the identification of internal firn layers and their spatial tracking.


Spatial and temporal variability of snow cover in Himalayan basins

Sunal OJHA

Corresponding author: Sunal Ojha

Corresponding author e-mail: ojha.sunal@i.mbox.nagoya-u.ac.jp

Satellite remote sensing is an effective tool for monitoring snow-covered areaa. However, complex terrain and heterogeneous land cover and the presence of clouds impose challenges to snow-cover mapping. This research analyzes snow cover and glaciers with a perspective of climate change in Himalayan regions using remote-sensing techniques. The remote-sensing snow-cover data from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite from 2000 to 2010 have been used to analyze some climate change indicators. In particular, the variability in the maximum snow extent with elevations, its temporal variability (8 day, monthly, seasonal and annual), its variation trend and its relation with temperature have been analyzed. The snow products used in this study are the maximum snow extent and fractional snow covers, which come in 8 day temporal and 500 m and 0.05° spatial resolutions, respectively. The results showed the tremendous potential of the MODIS snow product for studying the spatial and temporal variability of snow as well as the study of climate change impact in large and inaccessible regions like the Himalaya. The snow area extent (SAE) (%) time series exhibits similar patterns during seven hydrological years, even though there are some deviations in the accumulation and melt periods. The analysis showed relatively well inverse relation between the daily mean temperature and SAE during the melting period. Some important trends of snowfall are also observed. In particular, the decreasing trend in January and increasing trend in late winter and early spring may be interpreted as a signal of a possible seasonal shift. However, more years of data are required to verify this conclusion. Significant coverage of lake ice was found in the lower elevation zone, which is due to the flat terrain in this zone.


Extending East Antarctic ice-core chronology with radar layer stratigraphy

Marie G.P. CAVITTE, Donald D. BLANKENSHIP, Duncan A. YOUNG, Dusty M. SCHROEDER, Martin J. SIEGERT, Emmanuel LE MEUR

Corresponding author: Marie G.P. Cavitte

Corresponding author e-mail: mariecavitte@gmail.com

Airborne radar-sounding surveys collected by the University of Texas Institute of Geophysics (UTIG) with a 60 MHz system are used to trace englacial layering between the two deep East Antarctic ice cores: EPICA Dome C and Vostok. As a result of their isochronal properties, these englacial reflectors are used to connect the two cores continuously. Eleven layers spanning the last two 100 ka glacial cycles have been successfully connected, thereby providing a direct stratigraphic comparison of the two deep age–depth timescales over a 2200 m depth interval and a distance of 500 km. The coherent radar system used allows the identification of a layer depth to a precision much smaller than range resolution owing to strong signal-to-noise ratios of the layer strengths. These radar depth uncertainties can be can be converted to age uncertainties using the ice-core sites integrated in the radar surveys. We show that radar layer dating can therefore serve a useful role in recalibrating ice-core timescales with large age uncertainties. We also give a first-order recalibration of the EPICA Dome C EDC3 timescale using a radar-extended Vostok O2/N2 chronology (Suwa and Bender, 2008). In addition, the radar transects between Vostok and EDC3 show that aeolian stratigraphic reworking has a strong impact on layer depth accuracy, which impacts layers only in the last glacial cycle where ice-core chemistry is very reliable. As ice-core chemistry uncertainties increase in the penultimate glacial cycle, radar layering is apparently undisrupted by aeolian reworking, and the radar-extended EDC3 chronology is both reliable and characterized by smaller uncertainties than those for the existing geochemistry.


Flyby sounding of Europa’s icy shell: radar investigations, analogs and instruments for the Europa Clipper mission


Corresponding author: Donald. D. Blankenship

Corresponding author e-mail: blank@ig.utexas.edu

The Europa Clipper is a NASA mission concept to study Europa, the ice-covered moon of Jupiter, though a series of flyby observations of the Europan surface and subsurface from a spacecraft in Jovian orbit. One of the primary instruments in the strawman scientific payload is a multi-frequency multi-channel ice-penetrating radar (IPR) system. The IPR will play a critical role in achieving the mission’s science objectives for Europa’s ice shell, which are to characterize the distribution of any shallow subsurface water, search for an ice–ocean interface and evaluate a spectrum of ice–ocean exchange hypotheses. The Europa Clipper mission concept presents a range of technical challenges and opportunities for IPR science and engineering. The flyby-centric mission configuration is an opportunity to collect and transmit minimally processed data back to Earth and exploit advanced processing approaches developed for terrestrial airborne datasets. The mission concept also includes using the IPR as a nadir altimeter capable of measuring tides to test ice-shell and ocean hypotheses as well as characterizing roughness across the surface statistically to identify potential follow-on landing sites. Finally, the observation and characterization of subsurface features beneath Europa’s chaotic surface requires discriminating abundant surface clutter from a relatively weak subsurface signal. We will present instrument design and data-processing concepts for addressing these challenges. The development of successful instrumentation and data interpretation techniques for exploring Europa should leverage analogous terrestrial environments and processes. Towards this end, we will discuss a range of terrestrial radio glaciological analogs for hypothesized physical, chemical and biological processes on Europa and present airborne data collected with the University of Texas dual-frequency radar system over a variety of terrestrial targets. These targets include water-filled fractures, brine-rich ice, water lenses, accreted marine ice, and ice surfaces with roughness ranging from firn to crevasse fields and will provide context for understanding and optimizing the observable signature of these processes in future radar data collected at Europa.


Airborne radar sensor package for coincidental multi-frequency measurements of the cryosphere

Fernando RODRIGUEZ-MORALES, Sivaprasad GOGINENI, Carlton LEUSCHEN, Daniel GOMEZ-GARCIA, Benjamin PANZER, Cameron LEWIS, Reid CROWE, John PADEN, Richard HALE

Corresponding author: Fernando Rodriguez-Morales

Corresponding author e-mail: frodrigu@ku.edu

Recent observations performed using spaceborne sensors have revealed the dramatic changes that are taking place on both the Greenland and Antarctic ice sheets. While satellite sensors are a very useful tool for documenting these changes, they do not provide the data necessary to define boundary conditions with the resolution required to improve current large-scale ice-sheet models. Improved ice-sheet models are needed to generate accurate estimates of potential sea-level rise. Drastic variations in sea-ice thickness and extent are important indicators of climate change. Satellite sensors are also capable of measuring ice freeboard to estimate sea-ice thickness, but they do not provide accurate information about snow-cover thickness, which is necessary to minimize the uncertainty in sea-ice thickness estimates. Radar systems operated from an airborne platform can offer many advantages when compared with spaceborne instruments for cryospheric studies. These advantages include reduced operational costs, finer spatial and temporal resolution, etc. We have developed a multi-frequency radar sensor package designed for operation from fixed-wing aircraft. Our instrument package addresses the need for a suitable remote-sensing solution that enables the characterization of ice surface topography, internal ice stratigraphy, bedrock topography and snow-cover thickness over sea ice. The CReSIS instrument package is presently composed of four primary instruments operating over different frequency bands spanning from 160 MHz to 18 GHz. The first instrument operates in the 160–240 MHz range for sounding several-kilometer thick ice and mapping the ice–bedrock interface from low and high altitudes. The second instrument operates in the 600–900 MHz range for mapping annual layer thickness to a depth of a few hundred meters. This radar can also profile the bedrock topography under shallow ice with very high resolution. The third radar is an ultra-wideband (UWB) system operating at Ku-band frequencies (12–18 GHz) for ice surface topography retrieval and mapping of shallow subsurface layers to a depth of 10 m. In some platforms we have integrated a second UWB microwave (2–8 GHz) radar to measure snow-cover thickness over sea ice and map internal layers over land ice to a depth of 50 m. To maintain aerodynamic performance and maximize survey time, we have integrated all the required antennas into the aircraft. In this paper we provide a brief overview of the instrument package we have developed. We present some data examples to demonstrate the utility of putting a variety of sensors with different capabilities on a single airborne platform. We also show how the combined analysis of multi-spectral radar data is useful for characterizing the entire ice column, revealing features such as englacial water layers, and studying the transmission of basal variability to the ice surface.


How well can we determine ice thickness? Examples from Thwaites Glacier

Duncan A. YOUNG, Donald D. BLANKENSHIP, Scott D. KEMPF, Chad A. GREENE

Corresponding author: Duncan A. Young

Corresponding author e-mail: duncan@ig.utexas.edu

Ice-sheet models increasingly require high-resolution ice thickness and topographic data to resolve basal hydrology and internal stress fields. Additionally, new technologies for sampling the bed (e.g. RAID) will require good understanding of bedrock topography. Our primary tool for ice thickness determination has been airborne ice-penetrating radar. A variety of different systems have been fielded over ice sheets, with variations in center frequency, power, range, cross track and azimuth resolutions. Given the expense of fielding airborne campaigns, we need to be able to assess the resolutions and uncertainties that can be retrieved with through both legacy datasets and new systems, to target campaigns appropriately. Bed uncertainty quantification for ice-sheet models requires an evaluation of the spatial distribution of uncertainty in the ice thickness data upon which they rely. Ice thickness uncertainties are dominated by errors caused by cross-track reflectors, which bias thickness measurements low. Cross-track uncertainty is anisotropic, meaning that determinations of cross-over uncertainty do not capture our full knowledge of the bed. The grounding zone of Thwaites Glacier in West Antarctica is an area of fast and changing ice; ice flow is fast and the bed is rough. It has been a target of data acquisition both by the AGASEA program of 2004–05 and Operation IceBridge (OIB) between 2008 and 2012. AGASEA fielded a 60 MHz, 15 MHz bandwidth coherent system on a Twin Otter flying at 60 m s–1. OIB fielded a 195 MHz, 10 MHz system on a DC-8 flying at 130 m s–1. Both systems had broad cross-track beam patterns. The surveys were designed to interleave over the grounding-zone region, with one line reflown for intercomparison purposes. Over deeper ice we also have incoherent data from the 1990s with much less along-track resolution, but tighter line spacing. We evaluate the along-track repeatability and orthogonal cross-overs of these three surveys and construct a model for sensor-based uncertainty as a function of basal roughness and sensor configuration.


Joint seismic- and radar-sounding analysis of the subglacial environment of upper Thwaites Glacier, West Antarctica


Corresponding author: Leo E. Peters

Corresponding author e-mail: lep144@psu.edu

Thwaites Glacier is one of the fastest and largest glaciers draining the West Antarctic ice sheet. While much attention has been given to recent retreat, thinning and acceleration near its grounding line, little is known of the subglacial environment of Thwaites Glacier farther inland and how ice dynamics there might respond to coastal changes. Here we present both ground-based seismic- and radar-sounding surveys from upper Thwaites Glacier, characterizing the subglacial environment and its influence upon ice dynamics. During the 2008–2009 Antarctic field season, we collected 60 km of seismic data and 440 km of radar data ~200 km inland of Thwaites Glacier grounding line. These coincident surveys extend 40 km along flow and 10 km across flow. We find large variability in the subglacial environment, even in this slow-flowing region of the glacier (<200 m a–1), with distinct regions of wet unconsolidated sediments and potentially lithified dewatered sediments at the bed. Some of the brightest bed reflections in the radar data are observed across seismically inferred lithified beds, suggesting that in regions where bed roughness varies significantly, bright radar reflections are not indicative exclusively of wet ice-sheet beds. Modeled basal shear stress, seismically inferred basal conditions and radar-inferred small-scale bed roughness are all correlated. Our observations will allow modelers to better conceptualize the subglacial environment and to predict how Thwaites Glacier will respond to ongoing perturbations in ice flow originating near the grounding line.


Ice layer detection from radar depth sounder data using a novel approach based on the theory of electrostatics

Maryam RAHNEMOONFAR, Geoffrey Charles FOX

Corresponding author: Maryam Rahnemoonfar

Corresponding author e-mail: maryrahn@indiana.edu

In recent years global warming has caused serious damage to our environment. Accelerated loss of ice from Greenland and Antarctica has been observed in recent decades. The melting of polar ice sheets and mountain glaciers has a considerable influence on sea-level rise and altering ocean currents, potentially leading to the flooding of coastal regions and putting millions of people around the world at risk.The radar depth sounder is an important instrument that can provide relevant information about changes to polar ice sheets. Ice thickness can be determined by distinguishing layers of different dielectric constants such as air, ice and rock in radar echograms. Ice layer identification in radar echogram cross-section images facilitates three-dimensional modeling of ice layers and subsurface bedrock. Manual layer identification is very time-consuming and is not practical for regular long-term ice-sheet monitoring. The development of automated techniques is thus fundamental for proper data management. This paper proposes a novel approach to ice layer detection, using electrostatic force. According to Coulomb’s law, the magnitude of the electrostatic force between two charged particles is directly proportional to the scalar multiplication of the magnitudes of charges and inversely proportional to the square of the distances between them. In our proposed method, every pixel is assumed to be an electrically charged particle which has electrostatic interaction with other neighboring particles/pixels. The electrical charge of each particle is represented indirectly by the grayscale intensity of the pixel. In fact because pixel signs are always positive, the electrostatic force between them would be repulsive. To have both repulsive and attractive force between pixels, in the first step, pixels are transformed so that they have small values with positive and negative sign. In the next step, the electrostatic forces a pixel exerts on every other pixel around it are computed using Coulomb’s law. Finally the vector sum of all electrostatic forces is used to calculate the magnitude of signal variation. Borders between ice layers are indicated by high magnitudes of electrostatic forces. Experimental results of testing on publicly available radar echograms of Greenland and Antarctica show promising capabilities for automatically detecting ice layers.


Basal topography of Kronebreen, Svalbard


Corresponding author: Jack Kohler

Corresponding author e-mail: jack@npolar.no

Kronebreen, a tidewater outlet glacier, drains the icefield Holtedahlfonna in the Kongsfjord area of northwest Svalbard. Kronebreen is one of the fastest non-surging glaciers in Svalbard, with average annual speeds around 450 m a–1. As a result of rapid and accelerating flow, the most downstream 10 km of the glacier are heavily crevassed. While velocities have been recorded in a number of ways, it has not been possible to calculate ice fluxes or do ice-flow modeling since the bed has never been successfully mapped. Earlier airborne radar campaigns failed to detect the bed in the lower reaches of the glacier, due both to radar clutter from the extremely rough surface and most likely to water within the glacier. In 2009 and 2010, ice thickness data were successfully obtained using a10 MHz impulse dipole radar suspended beneath a helicopter. Using a helicopter allows both relatively long antennas to be used and more data stacking due to the slower flying speed. Certification for use of the radar on the helicopter was not required since the entire system was mounted to a frame suspended by the helicopter’s long line. The radar was controlled with a laptop in the cabin connected via WiFi and remote desktop to the radar’s computer. These new thickness data are combined with surface elevation maps to determine bed elevations. Together with older ice depth data from the upper icefield and fjord bathymetric data, bed topography was mapped in the area. Analysis of the new data will give a better understanding of Kronebreen’s retreat history and help in making predictions of when and how quickly further retreat may occur.


Wireless control of a portable lightweight dual-frequency radar system

Laurent MINGO, Gwenn FLOWERS, Nathaniel J. WILSON

Corresponding author: Laurent Mingo

Corresponding author e-mail: laurent.m@bluesystem.ca

The speed and ease with which a ground-based ice-penetrating radar survey is conducted can be important determinants of whether a project’s science objectives are met. For applications in which surveys at multiple frequencies are required (e.g. higher frequency for accumulation, lower frequency for bed sounding), it can be highly advantageous to collect all data in one pass. For the purpose of dual-frequency surveys on mountain glaciers and ice caps, where machine travel may be unfeasible, we have adapted a lightweight portable single-frequency ice-penetrating radar to accommodate two frequencies. This system is implemented using off-the-shelf components, including a 1–200 MHz Narod transmitter modified to accept a digital input square wave to drive radar pulsing, a 12-bit 100 MHz bandwidth digitizer for acquiring radar signals up to 250 MS s–1 (125 MS s–1 when two channels are used simultaneously), a GPS for geolocation and an embedded computer. In the modified system we use a second set of resistively loaded dipole antennas and a second Narod impulse transmitter while dual-frequency operation is managed by implementing a radio communication scheme between one radar receiver node and two radar transmitting nodes. This scheme is comparable with commonly used wireless sensor networks technology and is based on the IEEE 802.15.4 standards that define point-to-point and star communication at baud rates up to 250 kbits s–1. In this application, the two remote nodes associated with each transmitter are used for actuating (as opposed to sensing) through the implementation of two microcontroller-based timing engines (one per radar transmitter) that drive each radar transmitter independently of one another. The purpose of this approach is to ensure that the timing engines of the two radar transmitters operate independently, such that the transmitters never pulse at the same time. The embedded processing unit on the radar receiving side is comprised of the same standard components used for the single-frequency radar system, with the addition of a communication node. While operating, the system can stack up to 1024 traces in its current implementation. Preliminary tests were conducted in May 2012 using centre frequencies of 10 MHz and 40 MHz on a polythermal glacier in the St Elias Mountains of southwest Yukon, Canada. Previous work had suggested these frequencies to be useful for bed sounding and for detecting englacial thermal structure. Using the current system components, frequencies up to about 120 MHz could, in principle, be accommodated.


Interpretation of CryoSat-2 waveforms using CReSIS Ku-band altimeter data


Corresponding author: Aqsa Patel

Corresponding author e-mail: aqsa@cresis.ku.edu

Both satellite and airborne radar altimeters have been used to measure surface elevation for generating mass-balance estimates of polar ice sheets and sea ice thickness estimates from ice freeboard measurements. However, owing to the penetration of the altimeter signal into the polar firn there is ambiguity in between the tracking point and the actual surface location, producing an error in mass-balance estimates. In addition, owing to the presence of snow cover on sea ice, the estimation of ice thickness from freeboard measurements is more complicated, as the dominant scattering surface is not always the air–snow interface. Therefore, it is important to study Ku-band waveforms over different snow types and the dominant scattering interface at Ku-band frequencies in order to correctly interpret CryoSat-2 data. To address this problem we developed an ultra-wideband radar altimeter that operates over the frequency range of 12–18 GHz. The Center for Remote Sensing of Ice Sheets (CReSIS) Ku-band altimeter is designed to encompass the CryoSat-2 frequency band from 13.4 to 13.75 GHz. Data from the airborne radar altimeter can be sub-banded during processing to have the same frequency band as CryoSat-2 waveforms. When processed with the full 6 GHz bandwidth, the radar can resolve the air–snow and snow–ice interfaces over sea ice and near-surface internal layers in the Greenland and Antarctic ice sheets. The CReSIS Ku-band altimeter has been operating on NASA’s long-range aircraft as a part of Operation IceBridge and on the Twin Otter during CReSIS missions. This work will present Ku-band waveforms from data taken over land ice and sea ice using full bandwidth and reduced bandwidth processing. The layer tracker results from three different snow facies in Greenland show that a threshold tracker produces an accurate and consistent estimate of the snow surface. Over snow-covered sea ice, we find that the tracker varies between the air–snow and snow–ice interfaces even during the winter months in the Arctic. This suggests that the current sea ice thickness results from satellite radar altimetry, which assume the snow–ice interface always dominates, may be overestimating the thickness.


Development of high-frequency radar depth sounder


Corresponding author: Ali Mahmood

Corresponding author e-mail: ali23@ku.edu

Ice-sheet models require detailed information on the bed topography and basal conditions. However, ice sounding in fast-flowing glaciers and around ice-sheet margins poses major challenges due to surface clutter and volume scattering. These issues arise from ice surface roughness, as well as the composition, attenuation and backscattering from the ice. To overcome these challenges, a dual high-frequency (HF) and very high-frequency (VHF) radar depth sounder has been proposed and is currently under development. The radar operates at both 14 MHz and 35 MHz, with bandwidths of 1.4 MHz and 5 MHz, respectively. Unlike our VHF, P-band and ultra-high-frequency (UHF) sounders, the lower operating frequencies of this radar depth sounder reduces propagation loss through temperate ice and decreases volume scattering, which could help to improve the sensitivity of the system. The realization of a physical antenna array design to achieve a narrow cross-track beam width becomes impractical due to the large antenna sizes required for operation in the HF band. An alternate approach is to fly the radar along adjacent lines with an ~10 m spacing to synthesize a large two-dimensional (2-D) aperture. This provides a narrow beam width in both the along-track and cross-track directions. Covering such a tightly spaced 2-D sampling grid requires an airborne platform with high maneuverability. A small unmanned aerial vehicle (UAV) platform, such as the 40% scaled YAK-54, offers maneuverability advantages over large manned aircraft. Further improvements to ground-track accuracy of the small UAVs are enabled by a nonlinear model predictive controller, real-time system identification algorithms and, ideally, inter-vehicle coordination. The implementation of HF/VHF antennas on a small UAV is one of the major challenges in the design of the radar. For instance, the physical length of a 14 MHz half-wave dipole in free space is 10.7 m, which is twice as long as the entire wingspan of the UAV (5 m). The antenna for the 14 MHz band has been developed using folding and resistive loading. For the 35 MHz band, a bow-tie radiating structure has been developed. Both antennas were designed to meet the radar specification without affecting the implementation constraints of the UAV structure, such as aerodynamic performance and payload limitations to include volume, weight and power. This paper presents an overview of the radar system, antenna design considerations and preliminary results in the development of this system.


Regional ice accumulation patterns and flow channels revealed by airborne radar imaging of Jakobshavn Glacier, Greenland

George TSOFLIAS, Anthony HOCH, Leigh STEARNS, Cornelis VAN DER VEEN

Corresponding author: George Tsoflias

Corresponding author e-mail: tsoflias@ku.edu

Jakobshavn Isbræ in West Greenland is one of the fastest-flowing outlet glaciers in the world, draining ~7% of the Greenland ice sheet. Jakobshavn Glacier has exhibited unprecedented change by doubling its flow velocity over a decade. Understanding the behavior of the glacier has important implications in mass-balance modeling of the Greenland ice sheet and its contribution to sea-level change. In this study we present observations of ice accumulation patterns and ice-flow channels over the Jakobshavn catchment region obtained from a dense grid of airborne radar data acquired by the Center for Remote Sensing of Ice Sheets (CReSIS). The data were acquired using the multi-channel coherent radar depth sounder (MCRDS) flown in an orthogonal grid pattern along north–south and east–west bearings spaced 10 km apart over a survey region covering approximately 320 km &mult; 160 km. The radar data delineated the main Jakobshavn channel to the west and revealed four previously unknown tributary channels originating near the center of the survey grid and extending east toward the ice divide. The bed reflection and multiple internal layers were traced and gridded over the survey area. Varying intervals in the ice column were grouped and analyzed, from consecutive horizons to the entire ice column thickness, and were correlated to ice-core data. Internal ice layer maps show that bed topography is evident in all ice layers and it becomes more prominent with increasing depth. Ice layer thicknesses are greatest in the regions overlying the bed channels, suggesting ice flow into the topographic low channels from adjacent regions. Surface ice velocity observations are shown to correlate well with bed and internal ice layer channels at distances greater than 200 km inland from the glacier terminus. Large-scale ice thickness intervals grouped into shallow (~0–700 m), intermediate (~700–1300 m) and deep (~1300–1650 m) zones reveal variable ice column thickness trends that correlate to accumulation rate changes observed in the GRIP ice core. This work demonstrates that catchment-scale high-resolution dense grid radar imaging of fast-flowing outlet glaciers is critical for delineating accurately bed and ice layer topography, for quantifying their spatial relationships and for gaining insights to the controls on ice flow, past climate variability and accumulation changes.


Using seismic observations to detect deformable ice layers at the base of Jakobshavn Glacier and comparison with radar imaging


Corresponding author: Jose Velez

Corresponding author e-mail: javg@ku.edu

The development of preferred ice crystal orientation within the ice column can have a strong influence on the flow behavior of an ice sheet or glacier. Ice characterized by a preferred orientation (crystals aligned optimally for deformation) is about three times softer than ice with crystals oriented randomly. Typically, crystal fabrics are measured on samples recovered from ice cores. Although ice-core data are accurate and sampled densely in depth, they are 1-D limiting ice-sheet and glacier modeling capabilities. Therefore, there is a need to use remote-imaging techniques to determine the spatial extent of deformable layers and quantify preferred ice crystal orientation. Seismic reflection profiles acquired at Jakobshavn Glacier have shown englacial reflectors occurring in the lower 200–300 m of the 1900 m ice column. It has been postulated that the englacial reflectivity is the result of complex fabric development, which can introduce changes in seismic propagation velocity. Prior investigations have shown that seismic waves in ice travel up to 5% faster along the c-axis than perpendicular to it. Therefore, an ice column with ice crystals oriented preferentially will exhibit seismic wave velocities of propagation that vary as a function of angle of incidence. We employed an existing model that relates the variation of seismic slowness as a function of source–receiver offset to the mean ice crystal orientation of the ice column at Jakobshavn Glacier. Based on the anisotropy analysis we concluded that the upper 1640 m of the ice column consists mostly of isotropic ice with c-axes distributed over a conical region of 70° from vertical. The lower 300 m of the ice column is characterized by ice with preferred ice crystal orientation. Furthermore, warmer temperature ice is suggested by the seismic observations of the lower part of the glacier. Airborne radar profiles collected over the same region show lack of radar reflectivity in the lower 200–300 m of the ice column. The distinctive lack of coherent radar reflections from the basal ice interval at Jakobshavn also suggests highly deformed warmer ice.


Comparison of tomographic methods for ice bottom mapping


Corresponding author: John D. Paden

Corresponding author e-mail: paden@cresis.ku.edu

Basal topography is used in several important applications by the cryosphere community. These include boundary conditions in ice-sheet models, ice-core site selection and mass-balance calculations. Radio-echo sounding has been used for decades to measure the ice bottom topography, but primarily in profile or 1-D mode where only a single topographic point is collected at each radar position and the angle of arrival of this return is assumed to come from nadir even though it is known that strong off-nadir reflections can often dominate the radar echo return. The importance of fine-resolution 2-D ice bottom maps has been identified, especially in grounding-line retreat models and in ice-core site selection. The multichannel coherent radar depth sounder (MCoRDS) was developed for advanced array processing. The multiple channels have been used to produce a topographic swath from a single pass of the radar system. In this work we compare the accuracy of three different tomographic methods for measuring the basal topography underneath glacial ice at two different sites. The methods compared are multiple signal classification (MUSIC), reiterative super resolution (RISR) algorithm and the maximum likelihood estimator (MLE). MUSIC and MLE use the sample covariance matrix (SCM) to estimate the noise covariance, with MUSIC being a simplification of the MLE algorithm with much lower computational cost. The SCM is generated from data snapshots and has been applied under the assumption that neighboring pixels can be used as independent snapshots with similar statistics if the bed surface is relatively smooth. RISR avoids the generation of the SCM and is based on an iterative application of the minimum mean squared error (MMSE) estimator. We have applied each algorithm to two separate grids of data: one collected over Russell Glacier, Greenland, and another over Thwaites Glacier, Antarctica. The algorithms are compared by measuring the RMS elevation difference of each swath at cross-overs.


ICEPOD: an imaging system capable of monitoring supraglacial, englacial and subglacial processes in Greenland and Antarctica


Corresponding author: Nick Frearson

Corresponding author e-mail: nfre@ldeo.columbia.edu

The ICEPOD has been developed and flown by Columbia University in New York. IcCEPO is a geophysical instrumentation package in a POD that is capable of monitoring dynamic supraglacial, englacial and subglacial processes. The instruments include a high-power scanning laser for precise measurements of the ice surface elevation, stereo-photogrammetry from a high-sensitivity infrared camera (<20 mK) and a high-resolution visible-wave camera (2456 &mult; 2058 pixels) to document fine-scale ice temperature changes and surface features, a near-surface ice-penetrating radar (down to 100 m) and an ice depth measuring radar that can be used to study interior and basal processes of ice shelves, glaciers, ice streams and ice sheets down to a depth of >4.5 km. The absolute depth penetration is dependent on ice chemistry, temperature and surface features. All instrument datasets are time-tagged and geo-referenced using precision GPS satellite data. Aircraft orientation is corrected for using inertial measurement technology integrated into the pod. The vision is that this instrumentation will be operated both on routine and on dedicated science flights of the New York Air National Guard (NYANG) in the polar regions. The missions will initially operate out of McMurdo and South Pole stations and throughout Greenland. They will be used to map sea ice and outlet glaciers, such as those surrounding Ross Island and Greenland, through to quantifying large subglacial drainage systems in East Antarctica. A key aspect of the design is that the pod and data acquisition system will be available for use by the science community to install their own instrumentation onto. Proposals for new observations are welcome. The sensor system will become a research facility operated for the science community and data will be maintained at, and provided through, a polar data center. The science requirements and specifications for the primary instruments in the ICEPOD program can be viewed on-line at www.ldeo.columbia.edu/icepod. The system was flight certified in January 2013 and the first flights were conducted over the Greenland ice sheet in the spring and summer of 2013.


Using frequency tracking methods for SAR processing of sloped internal layers


Corresponding author: John Paden

Corresponding author e-mail: paden@cresis.ku.edu

Internal layering and slopes provide additional constraints on ice flow that can be used to test and tune models and isochronous layers can be used to estimate regional paleoaccumulation rate when connected to ice cores. Traditional synthetic aperture radar (SAR) processing works to detect internal layers, but is not optimized for their detection. SAR is essentially the application of a two-dimensional matched filter to the raw radar data where the value of each pixel in the output image is the result of the inner product between the raw radar data and the expected response from a point target at the location represented by the pixel. While internal layers can be viewed as a collection of point targets that interfere with each other to form a continuous reflector, the high level of structure or predictable coherence suggests that a matched filter to the collection of point targets would provide increased signal-to-noise and clutter ratio and therefore provide improved detectability. The problem with this technique is that the orientation of the layers is unknown a priori and must be estimated from the data. The signature from a continuous sloped internal layer is equivalent to a slowly changing frequency tone. Frequency tracking has been the subject of many studies and there is a long legacy of algorithms and applications. Using a frequency tracking concept while constraining the frequency search based on the knowledge that there is a high level of correlation between neighboring internal layers, we have created a new slope map data product and an improved internal layer radar echogram product. For the large antenna array on the P-3, we apply this same concept in the cross-track direction to co-estimate the cross-track and along-track slopes using a two-dimensional frequency tracking approach.


RES investigations of the Dome A region, East Antarctica: three-dimensional mapping of the basal accretion layers

SUN Bo, GUO Jingxue, CUI Xiangbin, WANG Tiantian, CHEN Yun, LUI Xiaojun, ZHAO Bo, LANG Shinan

Corresponding author: Guo Jingxue

Corresponding author e-mail: guojingxue@pric.gov.cn

AGAP project’s airborne radio-echo sounding (RES) data display accretion layering within the ice sheet in the Dome A region. These accretion layers represent the sub-ice water processes and meltwater refreezing onto the base of the ice sheet. Basal accretion ice formation processes are critical to understanding sub-ice hydrological features and their processes in relation to ice-sheet stability. Despite this importance to ice-sheet dynamics, high-density ground-based ice-penetrating radar surveys over the ice-sheet zones are sparse. Our lack of knowledge of the accretion ice three-dimensional geometry feature formation and transitting processes means, however, that the accretion ice formation processes and the transmitting track are uncertain. Here we present the most high-density ground-based ice radar investigations in the region of Dome A, East Antarctica. A new ground-based wideband coherent radar depth sounder system was designed and developed to obtain better delineation of the deeper internal layers and structure in the lowest hundreds of meters onto bedrock, wherein lies the echo-free zone (EFZ) in general. During CHINARE’s 2012/13 Dome A inland traverse, the newly developed ice radar was installed onto a snow vehicle and high-density radar transects were first made through detailed radio-echo sounding. The unprecedented density of radar transects in this region means that the geometry of the accretion ice was detailed and depicted. The three-dimensional mapping feature and its implication for the basal accretion layers are discussed.


Survey of the NASA ASAID basal stress boundary in the vicinity of Elizabeth City State University Bay and West Antarctica Peninsula


Corresponding author: Michael Jefferson

Corresponding author e-mail: michaeljefferso@yahoo.com

Gradual reduction of a small ice shelf in the Pine Island Bay area was discovered and examined using eleven Landsat images spanning 1972 to 2003. Measurements of Ice shelf area made possible by the NASA sponsored development of a circa 2003 Antarctic basal stress boundary data set indicate that it expanded slightly during the first two decades of observations from approximately 6.19 km2 measured on December 7, 1972 to a maximum of about 6.82 km2 observed in 1986. This maximum was followed by a nearly continuous decrease in area and ultimate disappearance of the ice shelf by January 17, 2003. No ice shelf has reappeared since 2003 as observed in subsequent Landsat images. Ten of the eleven Landsat images were co-registered and warped to one of a pair of 2003 geographic reference images before area measurement. Individual study team members made independent measurements of the ice shelf area apparent in each image. The average of these measurements had a standard deviation of 0.14 km2. The small, previously unnamed ice shelf formerly occupied what is now known as the Elizabeth City State University Bay. The specific cause of the disappearance of the ice shelf occupying Elizabeth City State University Bay is unknown, but is probably related to increased basal melting by warmer ocean waters reaching Pine Island Bay. Intrusions of warm ‘circumpolar deep water’ are related to ice shelf and outlet glacier thinning and retreat as reported throughout the Amundsen Sea region. This is the first report of complete ice shelf loss so far south or in the Amundsen Bay region and suggests that the advent of the NASA ASAID basal stress boundary will be a useful tool for performing a similar historical survey of other parts of the West Antarctic coastline that may be subject to similar changes. Presentation will focus on the methodology employed in the original research and expansion of the research scope to determine inaccuracies in current knowledge of the basal stress boundary or any instances of its change revealed by methods of passive and active remote sensing at other Antarctic coastal locales.


Automatic identification of ice layers in radar echograms

David CRANDALL, Geoffrey FOX

Corresponding author: David Crandall

Corresponding author e-mail: djcran@indiana.edu

Recent work has developed reliable ground-penetrating radar systems to study the subsurface structure of the polar ice sheets. These systems can produce vast amounts of data, to the extent that manually reviewing and labeling them can be prohibitively tedious and expensive. In this paper we describe our recent work on automating one such labeling task: automatically finding ice layer boundaries in noisy echogram images. Automatic layer finding in these images is quite challenging, due to unpredictable signal scatter and reflection, limited resolution, and variation in the basal terrain. Our approach applies modern techniques in the fields of computer vision and machine learning to this problem, in particular posing layer-finding as an inference problem on a statistical graphical model. These models can naturally incorporate diverse and noisy sources of information and a priori constraints, including known properties of ice sheets, ground truth from ice cores, and human operator input. To learn the parameters of these models, we take a machine learning approach, using a set of training data labeled by humans to infer parameter values automatically. We have tested the approach on a set of about 800 echograms and measured error quantitatively using a variety of metrics, and shown that our approach performs more accurately than several baselines.


Antarctic Ice-Sheet mass gains exceed losses: by how much and why?


Corresponding author: H. Jay ZWALLY

Corresponding author e-mail: zwally@icesat2.gsfc.nasa.gov

During 2003–08, the mass gain of the Antarctic ice sheet from snow accumulation exceeded the loss from ice discharge by 73 ± 23 Gt a–1 (3.7% of input), as derived from ICESat laser altimetry. The 131 Gt a–1 gain in East Antarctica (EA) and the 70 G a–1 gain in four drainage systems (DS) of West Antarctic (WA2) exceeded combined losses of 98 Gt a–1 from three coastal DS of West Antarctic (WA1) and 28 Gt a–1 from the Antarctic Peninsula (AP). Re-analysis of ERS radar-altimeter data, including a new post-glacial-rebound correction, indicates an even larger overall gain of 120 ± 51 Gt a–1 during 1992–2001. In WA2 and EA, persistent dynamic thickening (deficiency of ice flow relative to long-term accumulation) contributed more than 200 Gt a–1 to the net positive balance in both periods. Consistent with observed outlet-glacier accelerations, loss increases of 38 Gt a–1 in WA1 and 21 Gt a–1 in AP from increased dynamic thinning dominated a gain increase of 9 Gt a–1 from positive accumulation anomalies in WA1 and AP. These decade-scale changes are small relative to the long-term dynamic thickening in EA and WA2, which may buffer additional dynamic thinning for several decades.


Antarctic radio frequency albedo and implications for cosmic ray reconstruction


Corresponding author: Jessica STOCKHAM

Corresponding author e-mail: jegab8@ku.edu

The ANtarctic Impulsive Transient Antenna (ANITA) experiment, a balloon-borne suite of horn antennas, is designed to detect radio frequency (RF) signals from the interaction of neutrinos with nucleons in the Antarctic ice sheet. Subsequent to the the ANITA II 2008/09 flight, no neutrino events were discovered, but analysis did show the detection of cosmic ray air shower events. The majority of these events were detected as reflections from the ice surface. Reconstructing these reflected events requires analysis and modeling of the reflection properties of the air–ice interface. Using data obtained during the ANITA II 2008/09 flight, the direct and reflected solar signals are employed to estimate the the power reflection coefficients as a function of incident elevation angle. The results show a general agreement between averaged measurements and expected values given by the Fresnel equations.


Estimating radio frequency attenuation lengths in Antarctic and Greenlandic ice from radar depth-sounding data


Corresponding author: Mark STOCKHAM

Corresponding author e-mail: hammer@ku.edu

The balloon-borne ANITA experiment in Antarctica is designed to detect in-ice neutrino collisions, which produce radio waves that propagate upward to the suite of 32 horn antennas that constitute ANITA. The primary virtuate of ANITA is the ability to simultaneously observe 20 000 km3 of ice given from its 38 km altitude vantage point. The radio frequency signal strength observed at the balloon, however, depends on the radio frequency attenuation length of the ice through which the neutrino-generated signal must travel. Ice attenuation length varies as a function of surface temperature, depth, bedrock depth and ice chemistry (imperfections). The CReSIS project has data from many locations in Antarctica and Greenland produced by radar depth sounding. Using methods developed by analyzing the continuum signal in radar depth-sounding data from Greenland, an approximation to the attenuation length, as a function of location, in Antarctic ice is presented.