These observations and inferences outline the require-ments for a basal model. Thermal state of the bed and inte-grated water flux must be calculated. Sliding can be estimatedusing a relation including cavitation, with cavitation increas-ing with water supply and water pressure but limited by theincrease in water flux that occurs with increased cavitation. Inregions of soft subglacial materials, water pressure will affecttill properties and the mobilization of more till. Till continuityis essential, because thicker till will bury more sticky bedrock.

The references cited here include candidate relations forthe low-stress-exponent till flow law, the sliding relation, thetill-generation relation, and the description of the water systemneeded for a complete basal model. Much work remains tochoose among these candidate relations and then refine them.

This work was supported in part by National ScienceFoundation grants OPP 89-15995, its continuation, and EAR90-58193.

ReferencesAlley, R.B. 1989a. Water-pressure coupling of sliding and bed defor-

mation: I. Water system.Journal of Glaciology, 35(119),108-118.Alley, R.B. 1989b. Water-pressure coupling of sliding and bed defor-

mation: II. Velocity-depth profiles. Journal of Glaciology, 35(119),119-129.

Alley, R.B. 1991. Deforming-bed origin for southern Laurentide tillsheets? Journal of Glaciology, 37(125), 67-76.

Alley, R.B. 1993. In search of ice-stream sticky spots. Journal ofGlaciology, 39(133), 437-447.

Alley, R.B. In press. Toward a hydrologic model for computerized ice-sheet simulations. Hydrological Processes.

Alley, R.B., S. Anandakrishnan, C.R. Bentley, and N. Lord. In press. Awater-piracy hypothesis for the stagnation of ice stream C. Annalsof Glaciology.

Alley, R.B., D.D. Blankenship, C.R. Bentley, and S.T. Rooney. 1987. Tillbeneath ice stream B. 3. Till deformation: evidence and implica-tions. Journal of Geophysical Research, 92(B9), 8921-8930.

Anandakrishnan, S., and R.B. Alley. In press. Ice stream C sticky spotsdetected by microearthquake monitoring. Annals of Glaciology.

Bentley, C.R. 1987. Antarctic ice streams: A review. Journal of Geo-physical Research, 92(B9), 8843-8858.

Blake, E.W. 1992. The deforming bed beneath a surge-type glacier:Measurement of mechanical and electrical properties. (Ph.D. the-sis, University of British Columbia, Vancouver).

Blankenship, D.D., C.R. Bentley, S.T. Rooney, and R.B. Alley. 1987. Tillbeneath ice stream B. 1. Properties derived from seismic traveltimes. Journal of Geophysical Research, 92(B9), 8903-8911.

Boulton, G.S., and R.C.A. Hindmarsh. 1987. Sediment deformationbeneath glaciers: Rheology and geological consequences. Journalof Geophysical Research, 92(B9), 9059-9082.

Clark, P.U. 1992. Surface form of the southern Laurentide ice sheetand its implications to ice-sheet dynamics. Geological Society ofAmerica Bulletin, 104(5), 595-605.

Fowler, A.C. 1987. Sliding with cavity formation. Journal of Glaciology,33(115), 255-267.

Humphrey, N.F. 1987. Coupling between water pressure and basalsliding in a linked-cavity hydraulic system. International Associa-tion of Hyd rological Sciences Publication, 170, 105-119.

Humphrey, N.F., B. Kamb, M. Fahnestock, and H. Engelhardt. 1993.Characteristics of the bed of the lower Columbia Glacier, Alaska.Journal of Geophysical Research, 98(B1), 837-846.

Johnson, W.H., A.K. Hansel, and B.J. Stiff. 1991. Glacial transportrates, late Wisconsinan Lake Michigan Lobe in Central Illinois:Implication for transport mechanisms and ice dynamics. Geologi-cal Society ofAmerica Abstracts with Programs, 23(5),A61.

Kamb, B. 1987. Glacier surge mechanism based on linked cavity con-figuration of the basal water conduit system. Journal of Geophysi-cal Research, 92(B9), 9083-9100.

Kamb, B. 1991. Rheological nonlinearity and flow instability in thedeforming-bed mechanism of ice-stream motion. Journal of Geo-physical Research, 96(B10), 16585-16595.

MacAyeal, D.R. 1992. The basal stress distribution of ice stream E,Antarctica inferred by control methods. Journal of GeophysicalResearch, 97(B1), 595-603.

Shabtale, S., and C.R. Bentley. 1987. West antarctic ice streams drain-ing into the Ross Ice Shelf: Configuration and mass balance. Jour-nal of Geophysical Research, 92(B2), 1311-1336.

Walder, J.S. 1982. Stability of sheet flow of water beneath temperateglaciers and implications for glacier surging. Journal of Glaciology,28(99), 273-293.

Walder, J.S., and A.C. Fowler. 1994. Channelized subglacial drainageover a deformable bed. Journal of Glaciology, 40(134), 3-15.

Weertman, J. 1972. General theory of water flow at the base of a glaci-er or ice sheet. Reviews of Geophysics and Space Physics, 10(1),287-333.

Weertman, J., and G.E. Birchfield. 1982. Subglacial water flow underice streams and west antarctic ice-sheet instability. Annals ofGlaciology, 3, 316-320.

Whilans, I.M., and C.J. Van der Veen. 1993. New and improved deter-minations of velocity of ice streams B and C, West Antarctica,Journal of Glaciology, 39(133), 483-490.

Ice velocities near a relict flow feature on Siple DomeR.W. JACOBEL, C.W. DORSEY, andA.M. HARNER, Department of Physics, St. Olaf College, Northfield, Minnesota 55057

A dvanced very-high-resolution radiometer (AVHRR)magery of the western Ross Embayment shows a linear

feature crossing the northwest flank of Siple Dome, the ridgeseparating ice streams C and D. Bindschadler and Vornberger(1990) have tentatively identi9fied this as a possible relict icestream margin, traversing an area which today is interstreamice. If this is the case, it would indicate a different configura-tion of these ice streams in the past and have importantimplications for the behavior of the west antarctic ice sheet.

In an effort to gain more information about this relict marginfeature, we have made enhancements of AVHRR imagery togain greater spectral resolution. We have also used repeatLandsat thematic mapper (TM) scenes and the methodsdeveloped by Bindschadler and Scambos (1991) to determinesurface-ice velocities at one location near the relict margin.

From the National Oceanic and Atmospheric Administra-tion's archive of AVHRR imagery, we have selected five nearlycloud-free images of the western Ross Embayment. Three of

ANTARCTIC JOURNAL - REVIEW 199462

the images were acquired at approximately the same Sunazimuth and two others with the Sun approximately 900 and1800 relative to this. Surface features illuminated at low Sunangle may appear with high contrast in images recorded at aparticular solar azimuth, but other Sun positions can give use-ful information about the feature as well. Working with band 2data, we first coregistered all the images and then performed aprincipal component analysis on the three with different solarazimuths to extract brightness information common to them.

After several trials, we found that the best enhancementof the surface features of interest came from using the firstprincipal component of the three images acquired at approxi-mately the same solar azimuth. A subscene of approximately380 by 260 kilometers (km) from this composite is shown infigure 1. The image is centered on the Siple Dome and shows

the hypothesized relict margin feature striking diagonallyacross the northeast flank of the dome, just right of center inthe image. Ice stream D enters the scene at the upper rightand flows into the Ross Ice Shelf near the upper center, whileice stream C flows from right to left across the lower part ofthe image. The relict margin feature appears to result fromflow out of ice stream C, across the Siple Dome divide andinto the Ross Ice Shelf adjacent to ice stream D. The enhance-ment in this image over any of the individual scenes does notresult from improvement in spatial resolution, which is still1.2 km per pixel, but because the original radiometric bright-ness values are now mapped into a much larger range due tothe principal component analysis.

To study this feature in more detail, we have examinedLandsat TM imagery of the area. Figure 2 is a portion of one

Figure 1. Composite AVHRR image of Siple Dome and surrounding ice streams C and D made by extracting the first principal component of threeindividual scenes at approximately the same Sun angle. Image size is 260 by 380 km; true north is toward the top. The prominent linear featurecrossing the northeast flank of Siple Dome is hypothesized to be a relict ice stream margin.

ANTARCTIC JOURNAL - REVIEW 199463

TM scene showing the northeast corner of Siple Dome andpart of the ice stream D grounding line area. The image isapproximately 53 by 42 km on a side; true north is to the left.The south margin of ice stream D trends from the upper righttoward the center of the image where it joins the Ross IceShelf. A train of crevasses and buckled ice generated at thisshear margin continues diagonally across the image on thefloating ice. The Siple Dome relict margin feature enters theimage at the lower right trending parallel to the modern shearmargin and terminates where ice becomes ungrounded, rightof center in the lower part of the image. Long wavelength con-

trast changes in the image (for example, between the relictmargin and the modern ice stream D shear margin) are theresult of albedo variations due to new snow accumulation.The TM scenes have been processed by combining bands 2, 3,4, and 5, and then using a high-pass filter to remove most ofthe edge-to-edge solar brightness differences (Bindschadlerand Scambos 1991).

Figure 3 is a 25-by-16-km subscene of the TM image infigure 2 shown at full resolution. The heavily crevassed andbuckled ice zone running diagonally across the left side of theimage is the extension of the shear margin of ice stream D

1-igure 2. Landsat TM image of the northeast corner of Siple Dome and part of the ice stream D grounding line area. The south margin of icestream D trends from the upper right toward the center of the image where it joins the Ross Ice Shelf. The Siple Dome relict margin feature cutsdiagonally across the image at the lower right. Scene dimensions are approximately 53 by 42 km; true north is toward the left.

ANTARCTIC JOURNAL - REVIEW 199464

flowing onto the Ross Ice Shelf. The relict margin, thoughmuted in this enhancement, runs diagonally across this lowerright portion of the image. Between this relict margin and themodern shear margin of ice stream D (just right of center inthe image) is a small group of crevasses.

These crevasses have been located in two other images ofthis area taken 2 years before and 2 years after the scene infigure 3. Using the methods of Bindschadler and Scambos(1991), we have coregistered all three images and then used apattern recognition algorithm to measure the displacementsof these crevasses. The resulting velocities are given in thetable. Because most of the uncertainty in the crevasse posi-tions comes from the coregistration, the velocities for individ-

ual crevasses from each image pair show good agreement,whereas the means for the 2- and 4-year intervals clusterabout slightly different values. This does not indicate achange in velocity but simply that the ice here is moving at arate of 20 and 30 meters per year.

Using these same image pairs, we have measured veloci-ties within ice stream D of greater than 500 meters per year inagreement with results obtained independently by Scamboset al. (in press). In contrast, ice in the interstream ridgesmoves typically at a rate of less than a few meters per year.Thus, at least one area of the ice within the hypothesized for-mer relict ice stream channel is moving at speeds intermedi-ate between those of interstream ridge ice and a modern ice

Figure 3. Landsat TM image of a subportion of figure 2 at full resolution. The scene is approximately 25 by 16 km; true north is to the left. A trainof crevasses can be seen inside the paleo ice stream channel, just west (above) of the relict margin feature (right of scene center).

ANTARCTIC JOURNAL - REVIEW 1994

65

Velocities of crevasse cluster inside relict margin

20 3621 3122 4421 2721 2918 3121 32

aln meters per year ±10. Mean 20.5 meters per year. Standarddeviation of the mean = 1.2 meters per year.b ln meters per year ±14. Mean = 32.7 meters per year. Standarddeviation of the mean = 5.7 meters per year.

stream. This is an intriguing result, which if confirmed, sug-gests that the configuration of the ice streams has beenaltered dramatically in response to environmental change.

We are presently engaged in fieldwork to study the relict mar-gin area using surface-based ice-penetrating radar to examinethe continuity of internal layers, and geodetic surveying tomeasure ice velocities and strain rates.

We wish to acknowledge R. Bindschadler, M. Fahnestock,1'. Scambos, and P. Vornberger, who gave us invaluable assis-tance with the images and the software for obtaining veloci -ties. This work was supported by National Science Founda-tion grant OPP 93-00165 to St. Olaf College.

References

Bindschadler, R.A., and P.L. Vornberger. 1990. AVHRR imageryreveals antarctic ice dynamics. EOS, Transactions of the AmericanGeophysical Union, 71(23), 741-742.

Bindschadler, R.A., and T.A. Scambos. 1991. Satellite-image-derivedvelocity field of an antarctic ice stream. Science, 252(5003),242-252.

Scambos, T.A., K.A. Echelmeyer, M.A. Fahnestock, and R.A. Bind-schadler. In press. Development of enhanced ice flow at the south-ern margin of ice stream D, Antarctica. Annals of Glaciology, 20.

Internal layer folding patterns from radar studies ofice streams B and C

ROBERT W. JACOBEL and BEN J. GROMMES, Department of Physics, St. Olaf College, Northfield, Minnesota 55057

We have continued studies of the folding patterns of inter-nal layers seen in ground-based radar studies of ice

streams B and C. The echoes from these layers arise fromchanges in the dielectric properties of ice due to deposition ofdebris on the ice surface. Therefore, the internal layers repre-sent isochrones that can be analyzed to gain clues about icedynamics (Whillans and Johnsen 1983). In earlier studies usingdata we collected in collaboration with the U.S. Geological Sur-vey during the 1987-1988 and 1988-1989 field seasons (Wrightet al. 1990), we found that the folding patterns of internal layersare not related to bed topography or to areas of high basalshear stress—"sticky spots" (Jacobel et al. 1993). Instead, weconcluded that the folds are initiated as the ice transits fromthe inland ice sheet to fast-streaming flow by some processthat is not yet well understood. Additional evidence placing theorigin of folding well upstream of the point of detection in theice streams was our observation of the tilting of axial foldplanes in the flow direction. Assuming that folds form with theaxial fold plane vertical and using the flow law, we calculatedthat it would take on the order of several hundred years to pro-duce the observed tilting, therefore placing the origin of thefolds many kilometers upstream (Jacobel et al. 1993).

Our recent work has focused on the three-dimensionalnature of the fold structures because we have found that thereis also considerable deformation in the direction transverse toflow, and this deformation needs to be incorporated into an

understanding of ice-stream dynamics. Figure 1 shows twointersecting radar profiles acquired at the downstream B loca-tion, one approximately along the flow direction and theother transverse to it. The same prominent fold in the internallayers can be seen in both profiles. This can be the case only ifthe actual fold structure trends oblique to the flow directionand the profiles depict a projection of the fold on each axis.

In figure 1, the horizontal scales in each of the projec-tions are different, and this information has been used todetermine the strike of the fold, which in this case is atapproximately 58 0 to the iceflow direction. Bed topographyon the ice plain at the downstream B location is essentiallyflat, and the ice is only marginally grounded on a bed ofdeformable till (Blankenship et al. 1988; Rooney 1988). Basalshear stresses have been determined from an analysis ofstrain rates by Bindschadler et al. (1987) and are found to bevery small. Variations in them are thus unlikely to be respon-sible for producing the large folds we observe.

At the downstream B location, three-dimensional infor-mation is limited to the region of the two crossing profilesshown; however, at the upstream C location, we have radardata from a grid of approximately 5 by 30 kilometers (km),with 1-km spacing between the profile lines. Echoes from aprominent internal layer have been identified in nearly all theprofiles at a depth of approximately 725 meters, and figure 2shows a mesh surface depiction of this isochrone. This map is

ANTARCTIC JOURNAL - REVIEW 1994

66