Greenland glacier discharge variability from automated ground-based imaging

Project Personnel

Jason E. Box

Principal Investigator

Assistant Professor, Department of Geography

Byrd Polar Research Center

The Ohio State University

Henry H. Brecher

Byrd Polar Research Center

The Ohio State University

Dean Merchant

President Topo Photo Corp.

Professor Emeritus, The Ohio State University,

Fellow of American Society for Photogrammetry & Remote Sensing

Introduction and Motivation

Very recently, scientists have been surprised by how quickly such a large ice mass as the Greenland ice sheet can respond via its outlet glaciers to inter-annual climate variability. Regions of enhanced flow on the ice sheet, i.e. ice streams, have demonstrated large fluctuations in speed, thickness, and end position (Joughin et al. 2004; Podlech et al. 2004; G. Hamilton), in apparent response to positive and negative temperature trends. Even the ice sheet flow rates outside regions of streaming exhibit a significant sensitivity to the duration and intensity of melting (Zwally et al. 2002). While glacier discharge estimates have been made for Greenland’s largest outlet glaciers (Rignot et al. 2000; 2001; 2004), these have not captured what we now understand to be a system characterized by significant seasonal and interannual flow variability (Joughin et al. 2004; Luckman and Murray, 2005). The need for higher temporal resolution discharge data is evident.

This proposal describes autonomous ground-based mono- and stereo-imaging systems capable of monitoring surface velocity across a glacier surface at high temporal resolution using standard terrestrial photogrammetry techniques. Support for the maintenance of existing systems already ‘watching’ three Greenland outlet glaciers is requested. The addition of a second camera station at each sites is proposed to achieve unambiguous velocity determination from using photogrammetry techniques. Two additional glacier monitoring sites are proposed to represent regional ice sheet behavior. Graduate student support is also sought under this project.

Major Objectives

1. measure seasonal and interannual variability in Greenland outlet glacier surface velocities

2. use surface velocity cross section and existing ice thickness data to estimate variations in glacier ice-volume discharge near the grounding line

3. determine how much of the ice flow speed variance can be explained by climate variability

4. evaluate inter-regional correlation in glacier flow variability

This work is proposed to coincide with and commemorate the International Polar Year 2007/08.

Existing Measurements

In 2004, investigator Box was awarded pilot funding (Spatial variability in glacial ablation rates from digital photography, NASA grant NAG05-GA66G, Oct 01 2004 - Sep 30 2006, $34 k) from the NASA ICESat program to develop and deploy automatic camera systems in western Greenland. ‘IceCam’ systems were installed May 2005 featuring different targets. One IceCam captured the formation and drainage timing of a supraglacial melt lake (Figure 1) near the JAR 1 continuous GPS and Greenland Climate Network automatic weather station (Steffen et al. 1996).

Midnight, 05 Jun 2005

Midnight, 14 June 2005

Midnight, 27 June 2005

10 AM, 09 August 2005

Figure 1. IceCam image sequence from Lake JAR showing different stages of development. Midnight images are shown, as they provide high brightness contrast. Daytime images provide useful color contrast.

Monitoring Spatial Variations in Ablation

The Smart Stake 3 IceCam monitored the surface surrounding an automatic weather station 16 May 2005 – 14 August 2005. The position of six fixed targets in the field of view of the camera have been surveyed. Two of six targets are shown in the Figure 2 image subsets, i.e. the Smart Stake and the pole to its right. All six targets will be used to assess spatial variability of ablation rates at this site. Analysis of this data is planned as an honors undergraduate project beginning Fall 2005.

Figure 2. IceCam image sequence from Smart Stake 3 showing different stages of the melt season. One of four metal poles, in addition to the Smart Stake mast are available to determine spatial variations in surface ablation rate.

Sermeq Avannarleq (Dead Glacier)

An IceCam was activated 12 August 2005 to monitor variations in the Sermeq Avannarleq glacier, near the line of flight between Ilulissat and Swiss Camp (Figure 3). The first four days of data were collected from this site on 16 August with helicopter flight time provided by The Greenpeace Project Thin Ice expedition, July-August 2005 (Figure 3). A visit to this site is planned in 2005 as part of field work at Swiss Camp (K. Steffen letter attached). This is one of three outlet glacier sites that host single IceCam, i.e. mono systems. Mono imaging can provide information on flow rates with precise camera orientation measurements and flow direction assumption. However, uncertainties for mono systems are significant. At modest additional cost, outlet glaciers can be monitored with two cameras, i.e. stereo systems, which provide unambiguous position detection and higher likelihood of continuous surveying, in the event that one of the cameras fails. Figure 4 features existing and proposed IceCam sites.

Figure 3. Image sequence showing IceCam coverage of Sermeq Avannarleq and ice front changes over a 24-hour period (2000 UTC 15 August (left) and 2000 UTC 16 August (right) 2005) show calving of an ice bulge near the center of the ice front.

Figure 4. Greenland location map showing existing and proposed IceCam sites.

A note on Isbraes, Ice streams, and Glaciers

Isbrae-type outlet glaciers: have very high driving stresses; flow through a deep bedrock channel significantly deeper than the surrounding ice; have relatively steep surface slopes; and have relatively high ice flux, as compared to ‘ice streams’ and especially as compared to ‘glaciers’ (Truffer and Echelmeyer, 2002). Here the term ‘outlet glacier’ is used to refer to all such ice flow systems. Imaging both isbrae and ‘glacier’-type’ outlet glaciers is proposed to test the hypothesis: only isbrae-type outlet glaciers are sensitive to short term climate variations through meltwater interactions.

Jakobshavns Isbrae

After three months of measurements at the JAR1 melt lake (Figure 1), the Ice Cam equipment was relocated to a site along side the Jakobshavns Isbrae (Figure 5), at the easternmost land position with a view of nearly the entire 7 km long ice front. The Jakobshavns Isbrae is one of the most productive glaciers in the world and is perhaps the best documented of all Greenland outlet glaciers (e.g. Weidick et al. 1990; 2003; Weidick 1992; 1994; Abdalati et al. 2001; Thomas et al. 2003; Podlech and Weidick 2005). Jakobshavns Isbrae has been observed by repeat satellite measurements to be in a recent state of doubled flow speed (Joughin et al. 2004), thinning (Abdalati et al. 2001; Krabill et al. 2004), break-up and retreat (e.g. Weidick et al. 2003), in particular since 2002. Understanding its speed fluctuations (Luckman and Murray, 2005) and continued retreat is of importance to ice sheet mass balance and global sea level assessments. Installation of this camera in this ‘World Heritage’ site has been formally approved by the Greenland Home Rule government.

Figure 5. IceCam placed near the Jakobshavns Isbrae ice front 11 August 2005.

Figure 6 illustrates the proposed camera setup at Jakobshavns Isbrae and Sermeq Avannarleq. NASA ice sounding radar (ISR) and laser altimeter flights from 22 May 2005 are shown (J. Sonntag, E. Frederick personal communication). An IceCam horizontal field of view (66 degrees) is shown for the Jakobshavns Isbrae situation. Details of stereo calculations are provided in appendix A.

Figure 6. Existing and proposed IceCam sites at the Jakobshavn and Sermeq Avannarleq outlet glaciers. 07 August 2005 MODIS image resolution is 250 m, processed by J. Box.

Sermilik Brae

A new IceCam was installed 04 August 2005 near the Sermilik Isbrae glacier in extreme south Greenland (Figure 7). Sermilik Isbrae has thinned 120 m since 1985 (Podlech et al. 2004). Work at this site is in collaboration with Carl Bøggild of the Geologic Survey of Denmark and Greenland (GEUS), and is complemented by an existing automatic weather station network, letter of support attached. Collaboration is planned to study this glacier response using the IceCam, weather station data and an ice dynamics model. This is one of three existing sites where a mono system is proposed to be upgraded to a stereoscopic system.

‘trim line’

Figure 7. IceCam at the ice front of Sermilik Isbrae, installed 4 August 2005 (left). The ice front is visible on the left half of the image. The glacier ‘trim line’ to the right of the glacier is visible as the interface of darker vegetated land above bare land, suggesting 245 m (800 ft) thinning since c. 1850, Half of the thinning appears to have occurred since 1985 (Podlech et al. 2004). MODIS image featuring site positions (right).

Proposed Measurement Sites

Rinks Isbrae

Rinks Isbrae has high fjord walls advantageous for IceCam viewing of surface roughness features (Figure 8). MODIS imagery show a relatively small front retreat for this glacier, as compared to Jakobshavns Isbrae. A motivating question in the site selection is: will glaciers further north begin to accelerate as southern glaciers already have? This glacier is in the neighborhood of Uummannaq, serviced by commercial air flights and Air Greenland Bell 212 helicopter charter. The Bell 212 range is 225 Nautical miles (416 km) or 2.3 h, with sufficient fuel to divert to alternate airport.

Figure 8. Proposed IceCam sites at the Rinks Isbrae outlet glacier, overlain on MODIS image.

Helheim Glacier

Representing east Greenland glaciers is important to provide a larger-scale perspective. Recent observations from G. Hamilton and L. Stearns suggest that the Helheim Glacier has begun to flow more rapidly since 2001. Figure 9 illustrates the proposed camera setup.

Figure 9. Proposed IceCam sites at the Helheim glacier, south east Greenland, overlain on MODIS image.

Equipment

IceCams were developed in consultation with equipment manufacturers and photographers. Advice from T. Pfeffer (Department of Civil Engineering, University of Colorado, Boulder) was most valuable. Commercial-grade cameras costing $400-$800 are triggered at 4-6 hour intervals using an intervalometer ($250). Data are stored on 2-4 Gb memory cards ($200-400). Stations do not transmit data, although this capability is considered, but has not been budgeted as it effectively doubles station cost. Camera enclosures ($114) protect equipment from rain and snow. Chemical desiccant inside enclosures prevent enclosure window fogging. Sealed gel cell batteries ($150) supply power that is recharged by a 20 W solar panel and charging regulator ($450). The integrated system was tested in a Byrd Polar Research Center cold-room March 2005. The cameras self-heat, even when quiescent, keeping the imaging sensor working even at sub-freezing temperatures surrounding the camera. A detailed list of equipment and costs is included with the budget materials. Incorporating micro-computers is proposed to provide system restart capability and to log health data, i.e. battery voltage, temperature, humidity. The cost per single camera system is $5069 with all capabilities proposed, including automatic system restart.

Surface Velocity Measurements from Stereo Imaging

Terrestrial stereo photogrammetry is a well-known technique (e.g. Moffit, 1967; Slama, 1980) and has been applied extensively to studies of glacier motion for many years (e.g. Brecher and Thompson, 1993). The determination of positions of points in three dimensions in a local coordinate system is a ‘space intersection’ problem. That is, by knowing the positions of two camera stations with respect to each other (at the ends of a ‘baseline’) and the three rotations of the axes of each camera and measuring the image coordinates of the terrain points on each of the two images, the positions of these points result from the intersection of the rays to the points. Appendix A gives details.

The positions and rotations of the cameras can be derived by ‘orienting’ the images to (at least three) points whose positions are known in the local coordinate system (‘control points’), by measuring them directly or by a combination of the two. The latter technique is often employed for various reasons, such as ease or difficulty of establishing control points, ability to measure the baseline distance accurately, etc. (e.g. Brecher and Thompson, 1993).

In this work, it is easy and straightforward to measure the positions of the two camera stations with sufficient accuracy by differential GPS (giving the added benefit of a global, rather than local, coordinate system). It is also proposed to level the cameras to the horizontal plane, which will allow easy definition and re-definition (after maintenance visits) of two of the three required camera axis rotation angles. Given that the camera orientations are to remain fixed throughout a given measurement period, the camera azimuthal rotation angles will be valid for all the subsequent photography. In the case of Sermeq Avannarleq, and possibly some of the other sites, it is proposed to establish at least one control point on (fixed) rock outcrops in order to allow the determination of the third rotation angle at subsequent time periods, to evaluate errors associated with camera motion, i.e. from wind vibrations or other displacements, e.g. by snow creep.

In the case of Jakobshavns Isbrae, it appears that fixed terrain can be included in the images, but at some detriment to the stereo-coverage. Alternatively, it may be also be feasible to measure the third rotation angle directly, i.e. measuring the viewing azimuth angles of the cameras using a precision compass. However, establishing at least one targeted control point on the (moving) ice appears to be the option that provides the most precise control. Photos from the existing IceCam site and some from the air suggest that a distinct feature on the ice front may be selected as a control point and thus can be measured with a helicopter hovering over and making a GPS measurement. However, the use of one or more artificial targets (such as a black plastic banner) to unambiguously define this control point is also considered, although the Greenland Home Rule Government may not approve landing permission for such a site within the World Heritage site.

Several features across the glacier, perpendicular to its flow, will be identified in image pairs for instants in time. The positions of these features can be calculated in ‘terrain coordinates’ from measurement in a ‘stereo model’. See Appendix A for details. The same features will be identified for image pairs at some later time, with time interval on the order of days, for the same time of day, to have surface illumination roughly the same, so shadows contribute a small amount to feature re-identification ambiguity. It may be possible and would be desirable to place at least a few artificial targets on the ice in order to ensure unambiguous identification at each measurement. The series of feature-displacements, in the 3D ‘terrain’ coordinate system, will then provide a velocity measurement. Depending on glacier velocity, daily to weekly velocity measurements are expected to exceed one standard deviation of the expected uncertainty.

Ice Volume Flux from Surface Velocities

Ice volume flux can be estimated using surface velocity cross sections and ice thickness for floating ice just below the grounding line (Rignot et al. 1997) because the ice vertical velocity profile and basal melt rates seem to be negligible. The position of the grounding line can be identified by synthetic aperture radar interferometry (Rignot et al. 1997). The positions of the proposed IceCams will feature ice near the grounding line. Ice sounding radar (ISR) (Gogineni et al. 2001) bedrock depth measurements in the vicinity of proposed sites may be obtained from NASA Wallops Facility investigators (W. Krabill, J. Sonntag, E. Frederick, P. Kanagaratnam). Airborne Topographic Mapper (ATM) altimeter data provide changes in surface elevation (Krabill et al. 2000; 2004) and thus help constrain the ice thickness. Ice Cloud and Land Elevation Satellite (ICESat) and CryoSat altimeter data will be of use in the determination of ice thickness at times not available from ATM surveys. Given ice thickness information from ISR and altimetry, it is feasible to derive seasonal variations in glacier volume flux, given IceCam-derived surface velocity measurements, ice thickness data, and targeting the part of the glacier that is at the grounding line position. For situations where the surveyed part of the glacier is above the grounding line, an estimate of the vertical velocity profile would be needed. To address this potential challenge, the incorporation of a glacier flow model would be planned. Podlech (2004) details such a model and its use is planned in collaboration with Carl Bøggild of GEUS, letter attached.

Correlation with Climate Data

Box et al. (2004; 2005) have applied a mesoscale atmospheric data assimilation model to compute spatial and temporal patterns of surface mass balance, including rates of meltwater production at sub-daily time-scales. The use of Automatic Weather Station (AWS) data from the Greenland Climate Network (GC-Net) (Steffen et al. 1996) has proven vital in this regard, for model error assessment and calibration. Melt water production information is currently available at 24 km horizontal resolution, with a 12 km product planned for 2006. These data will be used to compute the time-variation of the available meltwater volume for drainage basins contributing to individual outlet glacier flow rates. The flow rates derived from IceCam data will be compared with meltwater flux information to test the hypothesis: outlet glacier flow rates correlate with meltwater flux. Other parameters, such as daily to seasonal temperature, albedo, and precipitation anomalies, will also be compared with the outlet glacier flow rates to determine how much of the ice dynamics variance can be explained by short-term climate variability.

Lens Calibration

Although the Nikon cameras in use are not sold as ‘metric’ cameras, the careful determination of interior orientation parameters (focal length, lens distortion, principal point offset) for each camera, through calibration procedures, illustrated in Figure 10, allows retrieved metric imagery to be of more than sufficient accuracy for the purposes of this study.

Figure 10. Dean Merchant (left) and Henry Brecher (right) maneuver inexpensive Nikon digital camera into a variety of positions to obtain lens calibration data at a target field at Topophoto Inc., London, OH, USA.

Error Considerations

We calculate conservative uncertainties of ±10 to ±40 m in position determinations for features moving between 60±40 degrees of the line of sight at 2 km to 8 km distances, respectively. Given daily glacier motions of 10-70 m, 4-day velocity determinations significantly larger than the uncertainty. Uncertainty calculations that include sensor dimensions, uncorrectable lens distortions, and a small amount of vibration, suggest that the Nikon 5400 camera yields measurements at 2 km to 4 km positions good to ±10 and ±20 m, respectively. However, to resolve a feature 4 pixel in diameter, uncertainties are expected to be ±40 m. In cases when uncertainties are largest, the time interval for displacement calculations must be increased. Even with 2-week time resolution, seasonal variability in ice flow can be determined.

Maintaining camera stability is important, to minimize the need to co-register images using fixed image features on land (Harrison et al. 1992). Camera stability, as such, is a problem of the current design that will be minimized by using a more robust camera enclosure mounting flange. In almost all cases, unmoving features in image foreground and background help identify error from camera orientation changes.

References

Abdalati, W., W. Krabill, E. Frederick, S. Manizade, C. Martin, J. Sonntag, R. Swift, R. Thomas, W. Wright, J. Yungel, 2001: Outlet glacier and margin elevation changes: Near-coastal thinning of the Greenland ice sheet, J. Geophys. Res., 106(D24), 33,729 (2001JD900192)

Bamber, J. L., R. L. Layberry, S. P. Gogineni, 2001: A new ice thickness and bedrock data set for the Greenland ice sheet, 1, Measurement, data reduction, and errors, J. Geophys. Res., 106(D24), 33,773-33,780.

Box, J.E., D.H. Bromwich, L-S Bai, 2004: Greenland ice sheet surface mass balance for 1991-2000: application of Polar MM5 mesoscale model and in-situ data, J. Geophys. Res, Vol. 109, No. D16, D16105, 10.1029/2003JD004451.

Box, J.E., D.H. Bromwich, B.A. Veenhuis, L-S Bai, J.C. Stroeve, J.C. Rogers, K. Steffen, T. Haran, S-H Wang, Variability of Greenland surface mass balance (1988-2004) using calibrated Polar MM5 output, J. Climate, provisionally accepted.

Bennike O., N. Mikkelsen, H. K. Pedersen and A. Weidick, 2004: Ilulissat Icefjord - A world heritage site, GEUS publication, 116 pages, hardcover

Brecher H. and L. Thompson, 1993: Measurement of the retreat of Qori Kalis glacier in the tropical Andes of Peru by terrestrial photogrammetry, Photogram. Eng. Rem. Sen., 59(6), 1017-1022.

Gogineni, S. D. Tammana, D. Braaten, C. Leuschen, T. Akins, J. Legarsky, P. Kanagaratnam, J. Stiles, C. Allen, and K. Jezek, 2001: Coherent radar ice thickness measurements over the Greenland ice sheet, J. Geophys. Res., 106(D24), 33761-33772.

Harrison W.D., K.A. Echelmeyer, D.M. Cosgrove, and C.F. Raymond, 1992: The determination of glacier speed by time-lapse photography under unfavorable conditions, J. Glaciol. 38(129), 257-265

Joughin, I., W. Abdalati, and M. Fahnestock, 2004: Large fluctuations in speed on Greenland's Jakobshavn Isbrae glacier. Nature, 432, 608-610.

Krabill, W., W. Abdalati, E. Frederick, S. Manizade, C. Martin, J. Sonntag, R. Swift, R. Thomas, W. Wright, and J. Yungel, Greenland ice sheet: high-elevation balance and peripheral thinning. Science, 289, 428-430, 2000.

Krabill, W., E. Hanna, P. Huybrechts, W. Abdalati, J. Cappelen, B. Csatho, E. Frederick, Manizade, C. Martin, J. Sonntag, R. Swift, R. Thomas, J. Yungel, 2004: Greenland Ice Sheet: Increased coastal thinning, Geophys. Res. Lett., 31(24), L24402

Luckman A. and T. Murray, 2005: Seasonal variation in velocity before retreat of Jakobshavn Isbrae, Greenland, Geophys. Res. Lett., 32, L08501, doi:10.1029/2005GL022519, 2005

Moffit, F. H. 1967: Photogrammetry, International Textbooks in Civil Engineering, International Textbook Company, Scranton, PE, USA, p. 448-528.

Podlech S., The Qagssimiut Ice Lobe, South Greenland: An assessment of significant ice margin thinning, GEUS Rapport, 2004/67 (Ph.D. thesis)

Podlech, S. and A. Weidick, 2004: Observations on the catastrophic break-up of the front of Jakobshavn Isbrae, West Greenland, 2002 – 2003. J. Glaciol., 50(168), pp.153-154.

Podlech S., Mayer C., Bøggild C.E., 2004: Glacier retreat, mass-balance and thinning: the Qagssimiut ice margin, South Greenland. Geografiska Annaler, 86A(4), pp.305 – 317.

Rignot, E., G. Buscarlet, B. Csathó, S. Gogineni, W. Krabill, M. Schmeltz, Mass balance of the northeast sector of the Greenland ice sheet: a remote-sensing perspective, J. Glaciol,, 46(153), 265-273

Rignot, E., W. Krabill, S. Gogineni, and I. Joughin, 2001: Contribution to the glaciology of northern Greenland from satellite radar interferommetry, J. Geophys. Res., 106(D24), 34,007–34,020.

Rignot E., D. Braaten, S. P. Gogineni, W. B. Krabill, and J. R. McConnell, 2004: Rapid ice discharge from southeast Greenland glaciers, Geophys. Res. Lett., 31, L10401, doi: 10.1029/2004 GL019474

Slama, C.C. 1980: Manual of photogrammetry, American Society of Photogrammetry, 4^th Ed., Falls Church, VA, USA, p. 828-831.

Steffen, K., J.E. Box and W. Abdalati, 1996: Greenland climate network: GC-Net. CRREL, 98-103.

Thomas, R. H., W. Abdalati, E. Frederick, W. B. Krabill, S. Manizade, and K. Steffen, 2003: Investigation of surface melting and dynamic thinning on Jakobshavn Isbrae, Greenland, J. Glaciol., 49(165), 231–239.

Truffer M. and K. Echelmeyer, 2002, Of Isbrae and Ice Streams, Ann. Glaciol., Symp. on Fast Glacier Flow. Ann. Glaciol., 36, 67-72.

Weidick, A. 1992: Jakobshavn Isbrae area during the climatic optimum. Rapport Grønlands Geologiske Undersøgelse 155, 6672

Weidick, A. 1994: Fluctuations of West Greenland calving glaciers. In: Reeh, N. (ed.): Workshop on the calving rate of West Greenland glaciers in response to climate change, 143168. Copenhagen, Denmark: Danish Polar Center

Weidick, A., Oerter, H., Reeh, N., Thomsen, H.H. & Thorning, L. 1990: The recession of the Inland Ice margin during the Holocene climatic optimum in the Jakobshavn Isfjord area of West Greenland. Palaeogeography, Palaeoclimatology, Palaeoecology 82, 38939

Weidick A., N. Mikkelsen, C. Mayer and S. Podlech. 2003: Jakobshavn Isbrae, West Greenland: the 2002-2003 collapse and nomination for the UNESCO World Heritage List, Geological Survey of Denmark and Greenland Bulletin Nr. 4, Review of Survey activities 2003, pp. 85-88

Zwally, H. J., W. Abdalati, T. Herring, K. Larson, J. Saba, and K. Steffen, 2002: Surface melt-induced acceleration of Greenland ice-sheet flow, Science, 297, 218-222.

Additional Materials

Appendix: Digital Stereo Photogrammetry Basics

Figure A1 shows proposed IceCam situation in a two-dimensional horizontal plane. Camera orientation angles a₁ and a₂, focal lengths (f₁ and f₂) are key parameters to set up a stereo model. What remains are angle measurements derived from image pixels coordinates relative to the camera lens axis. Pixel units are converted to distance units given the camera focal length and ‘format size’ of the camera detector. With the focal lengths of cameras known from the calibration procedure (featured in Figure 10) and measurements of the image coordinates of a feature at positions c_i from the center of each image, the angles d_i are calculated from which the angles f_i to the feature of rays d_i are determined. 6 to 10 features across the glacier surface are expected to be tracked in image pairs with time intervals of 1-7 days. Interaction of the rays yields the coordinates x_i and y_i.

Figure A1. Schematic representation of proposed terrestrial photogrammetry.