Scientific Basis for IceBridge Science Requirements

Scientific Basis for IceBridge Science Requirements

1. Ice Sheets, Ice Caps, Glaciers Baseline Justification

Here, is the density, g is acceleration due to gravity, H is the ice thickness and h is the surface topography. Driving stress on West Antarctic Ice streams is approximately 10 kPa. Using that number as a required measurement accuracy, and selecting 1000 m of ice as a reasonable average value, the required accuracy on driving stress is written as

Assuming the total driving stress uncertainty is distributed equally between the two terms, the required surface slope accuracy is 0.06 degrees. Taking the surface slope (S) as

and assuming negligible error in the longitudinal coordinate, the uncertainty in S is

From here we can estimate either the requirement on surface elevation accuracy or the required sample spacing. For the 0.06 degree slope accuracy requirements and choosing the instrumental surface elevation error to be 0.5 m, the required surface elevation sample spacing is about 700 m (or about an ice thickness for this case) in the along flow direction. Alternatively, the sample spacing could be chosen and the elevation accuracy estimated. So if we set the sample interval at 1 km, the requirement on elevation accuracy is 0.75 m. We retain the 0.5 m requirement based on the IGOS recommendation of 0.5 m at 1 km sampling intervals (IGOS, 2007, p. 86).

The requirements on spatial repeatability accuracy are as follows. For regions where 2-dimensional data are acquired (1x1 km), the exact distribution of independent samples within the area is not important. For places where we simply have at least 500 m long tracks of narrow swaths (ATM for example), we need to have the swaths overlap to say 50% in order to reduce slope induced errors. ATM has achieved this overlap goal.

The flux across an output gate is given as

Here V is the speed orthogonal across the outward boundary S of the basin. Vertical variation in the speed is accounted for by a weighting parameter .

Assuming the errors on each measurement parameter are normally distributed, the variance of the flux is

Accounting for errors in the interface balance terms, the error on the equivalent, basin-averaged thickening rate is

The equation illustrates that there are 7 distinct error terms contributing to the error in ice sheet thickening rate. Taking the error on elevation changes as 15 cm/yr and given that the length of the outflux gate S, the depth-averaged-velocity tuning parameter (), and the velocity V, and the basin area A can be accurately measured or reasonably modeled, the allowable error on the remaining terms probably can be relaxed to (15/(3)^1/2) cm/yr or about 9 cm/yr ice equivalent. This now allows an estimate on the required thickness accuracy from

Taking V to be 1000 m/yr, S to be 1200 km,  is unity, and A is 10^6 km (roughly the dimensions of the Amery drainage and around the grounding line) the average error in ice thickness needs to be about 75 m. Given that this is one of if not the largest glacier drainage in the world, it seems reasonable to reduce the ice thickness error to 50 m on average for a basin half this size.

Ice sounding radars are capable of achieving these levels of accuracy as illustrated by following Skolnick (1962, p 464) who adopts the following relations for estimating travel time accuracy on radar measurements in the presence of noise

Here T is the echo arrival time uncertainty, t_r is the echo rise time, S is the signal strength and N is the noise (clutter) level. Conservatively taking the rise time as proportional to the inverse bandwidth (bw)

And the ice thickness error is

UHF and VHF radars are presently operating at bandwidths of about 20 MHz. For 0 dB SNR, the echo time error results in about 4 m error in ice thickness. Allowing for small errors in wave speed (well documented in both the laboratory and in the field), an ice thickness uncertainty of 50 m is well realizable with existing radar systems. The two confounding factors which reduce SNR are surface clutter and temperature-dependent absorption. Both processes are important in fast flowing outlet glaciers. This is illustrated in the coverage map showing where ice thickness data have been successfully acquired by the University of Kansas (Figure 1 ). Consequently we impose a less restrictive requirement of 10% of the ice thickness for outlet glaciers. Note also that warmer, more absorptive ice in southern Greenland also has restricted ice thickness coverage (figure 2)

Figure 1. Left: Ice thickness coverage (red) and all flight lines (red and blue) from 1993-2009 from the University of Kansas. Right: 1993-2009 ice thickness data collected by the University of Kansas over Jacobshavn Glacier. The red points represent places where automatically picked ice thickness data are available. Thickness data coverage can be expanded (especially along the length of the glacier channel) by using advanced signal processing to enhance bed echoes and manual picking techniques, as has been demonstrated on other glaciers . Consequently, these maps are a conservative estimate of coverage. (Figure compiled by A. Hock and K. Jezek)

Figure 2. Basal topography and ice thickness data coverage (IS4). Figure prepared by E. Rignot.

IS4) Gravity Measurement Accuracy. Gravity measurements are key to constraining ice sheet/glacier bed topography, grounding line configurations, bathymetry beneath ice shelves and ice tongues, fjord geometry and bed geology which can inform about basal conditions. Each of these parameters acutely impacts ice dynamics. Gravity data are essential to constrain bathymetry beneath ice shelves, floating ice tongues as well as the geometry of ice covered fjords inaccessible to classic surface ship marine geophysical measurements. Gravity can be used to identify sills in front of major outlet glaciers that will influence the transfer of warm marine water towards grounding lines. Oceanographic studies require topographic estimates +/- 50m to constrain the nature of the ocean-ice interaction. Finally gravity can be used to constrain the nature of the bed beneath ice sheets, outlet glaciers and ice streams. Regions of soft and hard bed can be linked to the underlying bedrock geology. Identifying the presence of any subglacial sediments or till beneath regions of fast flowing ice has the potential to improve ice sheet models.

In optimal conditions, airborne gravity measurements are capable of providing topographic estimates that are accurate to better than 50m. In the ideal world, airborne gravity measurements accurate to .5 mGal would produce a topographic model accurate to ~7.5m. Topographic and bathymetric models derived from gravity are limited by geologic “noise” or density variations within the regional topography and the fundamental resolution limitation associated with making measurements from a moving platform. As airborne gravity data requires a time-based filter, faster aircraft speeds result in decreased spatial resolution. Geologic “noise” can be reduced by linking bathymetric models to known geologic structures or by use of complimentary data sets such as magnetics. In Greenland and Antarctica, magnetic data can be used to help constrain subice geology along with the gravity data. For example, where the gravity data detects a ridge in the basement, the magnetic data can be used to differentiate between glacial till and crystalline basement for the ridge composition. An airborne magnetic system can image sub-ice geology such as volcanic rocks, crystalline basement and sedimentary basins.

Documenting the movement of water beneath the large ice shelves is key to understanding ocean-ice interactions but requires an accurate knowledge of the ice shelf cavity. Cavity geometry has been very difficult to obtain. Both the ice shelf thickness and the bathymetry of the continental margin are necessary to define the cavity geometry. Radar can measure the upper surface of the cavity but does not penetrate the water underneath. Exploration with autonomous underwater vehicles is possible but provides relatively little spatial coverage. Airborne gravity can support new bathymetric models beneath ice shelves. Examples of targets for recovering ice shelf bathymetry include the Petermann Glacier in Greenland and the Larsen C, Getz, Abbott and King George V ice shelves in Antarctica. An orthogonal grid of flight lines acquired at 5-10 km line spacing is appropriate to provide a regional bathymetry model. Over Larsen C a 20-50 km spaced airborne grid provides preliminary insights into the broad regional bathymetric trends, specifically overdeepenings adjacent to the grounding line and cross shelf troughs. (Cochran and Bell 2011).

Sill geometry is important as these ridges of elevated topography can provide pinning points for ice shelves or ice tongues or alternatively serve as a basic barrier to the flux of water between the global ocean to the ice sheet grounding line. 40 km in front of Thwaites Glacier a prominent ridge or sill serves as a pinning point for the remaining ice tongue. The sills in front of Thwaites and Pine Island Glaciers in Antarctica as well as 70-550m sills in front of many Greenland fjords are a basic valve on the circulation of ocean water to the grounding line. Estimating sill depths from gravity is best based on profiles perpendicular to the sill, generally profiles along the fjord, orthogonal to the grounding line. For the optimal bathymetric solutions 5 km line spacing orthogonal to the targeted sill and with cross lines is the best survey design.

Basal conditions are key to understanding the evolution of ice sheet velocity. Identification of sedimentary basins in West Antarctica have been aligned to the onset of fast flow of the ice streams (Bell et al, 1998, Anandakrishnan et al, 1998). Gravity can be used to identify the presence of sediments and hence the presence of a soft bed. The trough beneath Jakobshaven is filled by over 1 km of sediments based on the gravity data acquired from a Twin Otter (Block and Bell, 2010).

Gravity data collected as part of OIB have been demonstrated to achieve repeatabilities and accuracies better than several tenths of a milligal, which is an extraordinary accomplishment (see http://bprc.osu.edu/rsl/IST/documents/IceBridge%20AIRGrav%20Accuracy.pdf). High accuracies where achieved over long, straight, low elevation flights over sea ice and the Sanders result was independently confirmed by the IST. Over ice sheets, we expect that the instrument accuracy is similar but complications from geophysical structures can degrade repeatability. In the Sanders report quoted above, cross over difference standard deviations are about 1.5 milligal for filter lengths of about 83 sec. Accuracies degrade for both longer and shorter filters. Independent analysis by the IST generally confirms the ice sheet result. As noted above, we can use a simple Bouguer slab analysis to estimate that a 2 milligal uncertainty roughly equates to a 50 m uncertainty in the thickness of a water slab (density contrast 1 gm/cc).

Figure 3. Flight lines designed to coincide with Cryosat orbit tracks. Yellow lines indicate boundaries between Cryosat operating modes (figure provided by K. Jezek)

Figure 4. Temporal evolution of ice sheet elevation change in Greenland from ICESat and ATM data. Major changes are over the large outlet glaciers (Jak, Kanger and Helheim,) and over the marine terminating glaciers of NW Greenland. Increasing thinning is indicated in N Greenland, spreading over higher elevations on the NE ice stream. Maps provided by B. Csatho.

Figure 5. GRACE-derived, Antarctic Ice Sheet annual balance showing substantial mass loss along the western and southeastern flanks of the ice sheet (map provided by S. Luthcke)

IS7) OIB surface elevation measurements improve the ICESat data set by refining off-track slopes and ICESat-derived digital elevation models used for ice dynamics modeling. Because of the natural dimensions of Greenland coastal glaciers, ICESAT-1 tracks under-sample many narrow outlets near the coast where discharge and dynamics thinning rates are high (Figure 6). At more southerly latitudes where orbit tracks dirverge, the orbit geometry also limits knowledge of cross track slope corrections to ICESat data. Finally, cloud cover obscured observations over Greenland (figure 7). IceBridge will acquire a reference elevation set for coastal portions of Greenland and selected areas of Antarctica. The data will improve upon ICESAT-1 sampling for long term (>5 yr) elevation change estimates once ICESAT-2 is launched. This bridge data set will allow ICESAT-2 to begin making long-term (~5 year) measurements of thinning rates in the first year at a much improved spatial resolution.

Figure 6. ICESat-1 track coverage illustrates the undersampling of Greenland outlet glaciers (figure provided by I Joughin).

a.

b. c.

Figure 7. Coverage gaps in the ICESat record (a) (Figure provided by B. Smith). 100 km boundar around Greenland (b) and Antarctica (c). (Maps provided by M. Studinger).

IS9)  The net discharge from Greenland has been monitored successfully with InSAR at flux gates across the major outlets but many other imporant glaciers have never been measured (for example, in the northwest, in central east Greenland and in southeast Greenland). Moreover, the thicknesses across many of these gates have not been directly measured and in particular almost none has been successfully sounded close to the ice front or grounding lines. Rather, most data were collected upstream of the grounding line and approximations are necessary to deduce the grounding line ice fluxes. With the near-terminus regions of many glaciers thinning by 10s of meters during speedups, an up-to-date time series of near-terminus thickness is important for tracking ice discharge variability, particularly as missions such as DESDynI provide more frequent velocity mapping.

To obtain a complete and comprehensive estimation of ice fluxes into the ocean, not only for mass balance purposes but also for freshwater fluxes into the ocean, OIB will strive to collect ice thickness measurements as close as possible to the glacier grounding lines. Cross flow profiles and along flow profiles will be necessary to evaluate the precision of the mapping and reduce residual uncertainties. A one-time OIB radar sounding mapping to determine the bedrock depths will allow thickness to be tracked through time with subsequent altimeter-only surface mappings (i.e., OIB, ICESAT-2, and DESynI). With this objective in mind, OIB will measure thickness at gates located 3 and 8-km from the present termini of the ~200 outlet glacier in Greenland with widths of 2-km or more. The 3-km flux gates will provide the thicknesses necessary for a comprehensive determination of the ice flux discharge from Greenland. The 8-km gate will provide a second gate for instances where the terminus retreats by more than 3-km. In addition to providing flux gates, these transects along with the along-flow transects (IS8) will provide at least some bed elevation/thickness information for ice sheet modeling of all major glaciers.

In addition, OIB will document ice fluxes farther upstream as this will enable the determination of the pattern of net mass balance of the glacier farther upstream, and will enable the determination of future ice fluxes when glaciers retreat farther inland.

IS10) To, increase the number of ICESat orbits sampled by OIB and to facilitate flux measurement for comparison with altimeter derived estimates of ice thickness change, OIB shall collect elevation and ice thickness data along continuous lines parallel to 3 closed elevation contours (1000, 200 and 2500 m) (figure 9). Measurements at 2000 m will also extend the record begun in 1995 when a series of surface and airborne campaigns measured ice elevation and ice thickness.

B

Figure 9. 1000, 2000 and 2500 m contours about the Greenland Ice Sheet (left). Contours in the vicinity of Jacobshavn Glacier and ICESat reference orbits (right).

IS11) . The inability to place an upper bound on sea level change in the last IPCC reports stems largely from the uncertainty in our knowledge of glacier dynamics, the physics of which are not included in current whole ice-sheet models. OIB will acquire a comprehensive set of measurements for process modeling studies aimed at understanding outlet glacier dynamics. The data set will allow the development of process-level model experiments to determine the physics that govern fast flow and to develop the parameterizations needed for larger-scale predictive models. A representative set of glaciers should be chosen, including Jakobshavn, Kanger, Helheim, a northern glacier, and 2 or 3 glaciers each from the currently changing regions in the northwest and southeast. Measurements will extend to interior regions where surface velocities decrease to 50 m/yr (Figure 10). Coverage requirements are justified in the following sections.



Figure 10. Surface velocity thresholds on the Greenland Ice Sheet. (Figure provided by E. Rignot).

Detailed bed mapping in selected region.

The level to which subglacial properties affect ice dynamics and the level of details required to capture these effects remains an open question. As illustrated in Figure 11, the details in the bedrock topography in the Jakobshaven region influence the friction and driving stress resulting from inversion methods, and therefore the basal condition to be applied in any numerical model. The driving stress depends on the ice thickness and to compensate for the smaller thicknesses obtained with the 5km resolution models, the model need to drop the basal drag in order to fit the surface velocities. This example illustrates that any inversion made on the wrong bed will result in the wrong basal boundary condition and any prognostic simulations resulting from this dataset should be view with caution.

OIB has the potential to test the level of knowledge needed from the basal topography, and an answer to this question requires a bedrock known at a very fine resolution. Further refinement to the 2km grid flown in 2010 of the land terminating Russell glacier (Greenland) is a high priority and should ideally be complemented by a similar grid in a fjord region and an Antarctic grounding zone region.

Figure 11: Basal topography (top), driving stress (middle), and basal velocity (bottom) in the Jakobshaven region from 3 distinct bedrocks applied to the ISSM model. Bedrock used are the 5km Bamber dataset (left), the 5km (middle) and the 1km (right) SeaRISE datasets that incorporates the fine scale Cresis data. Figure courtesy of Mathieu Morlinghem, Helene Seroussi, S. Nowicki, E. Larour.

To improve prognostic predictions of future sea-level contribution from the ice Greenland and Antarctic ice sheets obtained with whole ice sheet models, IOB is required to continue measurements that fill the gap in our current knowledge in bedrock elevations over coastal regions. OIB is required to make bedrock measurements on a 5km grid over areas of slow flow, and on a grid of the order of at least the ice thickness for regions of fast flows, or regions of complex stress regime (eg: grounding line or shear margins). Priority should be given to regions where it is suspected that canyons might be present, or regions with bedrock below sea-level that could connect to the interior of the ice sheet (ie: high risk regions for ice loss).

The motivation for the strong focus in coastal regions compared to interior regions is due to Stone et al. (2010), where a good agreement in ice thickness is obtained in the interior of the Greenland ice sheet between the Bamber and Letreguilly dataset even thought the bedrock are significantly different in the two simulations (Figure 12).

Figure 12: The ratio of the difference in ice thickness (a) and bedrock topography (b) between the Bamber and Letreguilly datasets expressed as a percentage (z_bamber – z_letreguilly/z_letreguilly). Figure adapted from Stone et al. (2010).

The motivation for more refined bedrock measurement in regions of fast or complex flow is based on experience with the SeaRISE simulations, where a 5km bedrock in the Jakoshaven region produced from the fine scale Cresis measurements (see Figure 11 “Glen 5km”) is necessary in order to reproduce a surface velocity similar to the observed velocity. These regions are also regions where the coarse resolution of whole ice sheet models often prevents them from reliable diagnostic or prognostic simulations, such that regional models have an important role.

Flow band models.

Continue the time series annually in a few selected regions with existing long term measurements (ex: Helheim Glacier or Jakobshavn Isbrea). These regions allow a study of the dynamics of glacier retreat and inversion of surface properties and elevation could provide insight on whether the basal condition are changing in time, or to test our current understanding of the underlying physics at play.

IS12) For Antarctica, important mass balance and dynamical changes are associated with processes occurring at the grounding line – the boundary where the inland ice sheet begins to float on the ocean - or ice sheet land terminus (ISMASS, 2003). (Figure 13). Therefore OIB shall make measurements of surface elevation, ice thickness and gravity along the grounding line and along a parallel track approximately 10 km upstream of the grounding line. The data will be crucial for ice flux measurements and to investigate dynamical processes at the grounding line. The two tracks are required partly because grounding likely occurs over a zone and partly to increase information available for ice dynamics models (e.g. surface and basal topography gradients). Success will constitute the first, complete circum-ice-sheet measurements of ice thickness and basal topography. Combined with surface velocity data acquired using spaceborne synthetic aperture radar data, the data will enable the most accurate estimate of the total volume flux of ice being discharged from the ice sheet in the vicinity of the ice margin or grounding line. Differencing the ice flux across our flight path with the integrated accumulation rate over the interior surface will provide an independent estimate of total ice sheet volume change.

Figure 13. Radarsat Antarctic Mapping Project (RAMP) coast line (green), ASAID grounding line (yellow courtesy of R. Bindschadler) overlain on RAMP coherence mosaic. The Riiser-Larsen Ice Shelf grounding line is distinguishable in the coherence data by the dark band separating the smooth ice shelf from the textured interior ice sheet. The ASAID grounding line and the coherence band generally agree to about 3 km in this area. Map provided by K. Jezek.

IS13) The vast majority of the largest Greenland glaciers terminate in the ocean and melt in contact with the ocean waters. Recent studies suggest that this melt process is orders of magnitude larger than at the surface and probably plays a central role in glacier stability, grounding line retreat and ice calving. A better understanding of these ice-ocean interactions and their impact on glacier is essential to interpret recent changes in glacier mass balance and in turn ice sheet mass balance, and also to improve our predictive capability of ice sheet evolution. Taken another way, we will not be able to predict glacier and ice sheet evolution at all until we have made major progress in our understanding and characterization of these interactions.

To achieve this goal, it is critical to document the sea floor bathymetry in front of the glaciers as this determines the pathways for ocean heat to reach land ice. This includes the determination of the depth of the glacial fjords, the presence of sills from past grounding line positions, the presence of troughs generated by paleo ice streams, the exact depth of the fjords at the glacier front and how troughs and pathways are connected to the sea floor bathymetry on the surrounding continental ice shelf.

While data exist on the continental ice shelf, most glacial fjords in Greenland have never been surveyed (figure 14). Similarly, we have no sounding of the sea floor underneath floating ice shelves in the northern part of Greenland where ice-ocean interactions also play a fundamental role in glacier mass balance and evolution.

Therefore OIB shall collect measurements of the sea floor bathymetry in front of glaciers terminating in the ocean, in the glacial fjords extending to the mouth of the fjords where sills are usually present, and underneath floating ice shelves in the north. This will require at least a survey flight along the main axis of the fjords and a series of cross-fjord profiles as permitted by the local geography and time constraints on the mission. The goal is to determine the dominant geometric characteristics of these fjords to help better constrain ocean numerical models aiming at modeling ocean-fjord circulation and heat exchanges, with obvious implications for the modeling of ice-ocean interactions and glacier flow.

Figure 14. Bathymetry tracks used to do the best compilation todate of Arctic sea floor bathymetry, in particular around Greenland. See attached plot from IBCAO. Apart from southwest Greenland and fjords in east Greenland, the coastline is particularly devoid of data within 100-200 km of the coast (e.g. northwest Greenland, Kennedy channel in north Greenland, northeast greenland, and the area between Kangerdlugssuaq and Helheim Glaciers). bathymetry tracks have been used to do the best compilation todate of Arctic sea floor bathymetry, in particular around Greenland. (Plot is from IBCAO and provided by E. Rignot).

IS14) Measurements of the distribution of suglacial water have been made since the 1970’s using ice sounding radar (Kapitsa and others, 1996) and also radar and laser altimeters. Essentially, ice sounding radar data collected over large subglacial lakes reveal smooth, highly reflective basal topography. Flow over subglacial lakes also results in characteristically smooth surface topography that is detectable with radar and laser altimetry. Radar interferometry and laser altimetry have also been used to measure changes in lake volume by detected elastic deflection of the ice sheet surface as subglacial lakes drain (Gray and others, 2005, Fricker and others, 2007) . For thinner water layers or for sparsely distributed water, reflectivity data from ice sounding radars have also been used to map water distributions but this analysis is complicated by the several parameters that contribute to the final radar intensity (basal roughness, temperature dependent absorption through the ice). Consequently radar tends to remain an uncertain proxy indication for basal water

IS15) Submeter, stereo mapping photography will contribute to better crosss track slope corrections of altimeter data in coastal regions of Greenland and Antarctica where slope-magnitudes are highest and where the surface is complicated by crevassing. While stereo mapping will qualitatively improve the IceBridge data set, the quantitative improvement is still unknown because of overall accuracy uncertainties and the amount of data that can be reasonably processed from what could be a very voluminous data set.

2. 
   
   Directory: rsl
   rsl -> Abstract a new formal theory dt of truth extending pa
   rsl -> Bay Area Safeguarding/slr workshop Invite for bcdc contact list
   
   
   
   Download 150.46 Kb.
   
   Share with your friends: