Sedimentation beneath ice shelves — the view from ice stream B

Marine Geology, 85 (1989) 101-120

101

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

SEDIMENTATION BENEATH ICE SHELVES VIEW FROM ICE STREAM B

THE

R.B. ALLEY, D.D. BLANKENSHIP, S.T. ROONEY and C.R. BENTLEY Geophysical and Polar Research Center, University of Wisconsin, 1215 West Dayton St., Madison, WI 53706 (U.S.A.) (Received August 3, 1987; revised and accepted May 23, 1988)

Abstract Alley, R.B., Blankenship, D.D., Rooney, S.T. and Bentley, C.R., 1989. Sedimentation beneath ice Shelves -- the view from ice stream B. In: R.D. Powell and A. Elverhei (Editors), Modern Glacimarine Environments: Glacial and Marine Controls of Modern Lithofacies and Biofacies. Mar. Geol. 85:101 120. Ice-shelf development is favored by rapid flow of cold ice from outlet glaciers or ice streams into protected embayments with localized high spots. Basal melting of ice shelves is rapid near the ice front and m a y occur near the grounding line. Ice from outlet glaciers m a y contain significant englacial debris that is deposited as a dropstone diamicton in regions of basal melting. Englacial debris is sparse or absent in ice streams. Evidence from ice stream B, draining into the Ross Ice Shelf of West Antarctica, suggests that the rapid ice velocity arises from deformation of a several-meter-thick, water-saturated basal tilllayer that is eroding an unconformity on sediments beneath and that has deposited a "till delta" tens of meters thick and tens of kilometers long at the grounding line. Sea-level fall would cause "conveyor belt" recycling of this till delta and grounding-line advance across the Ross Sea to the edge of the continental shelf,forming an ice sheet with a low, ice-stream profileresting on a several meter-thick deforming tilllayer eroding an unconformity. The modern Ross Sea is characterized by a regional unconformity overlain by a diamicton of probable latest Pliocene-Pleistocene age measuring several meters to tens of meters thick. W e hypothesize that this diamicton is a deformed glacial tilland that the Ross Sea sediments record one or more expansions of the till-lubricatedWest Antarctic ice sheet to the edge of the continental shelf.

Introduction Ice shelves are important elements of glacial systems and may play a significant role in glaciomarine sedimentation. A considerable amount of theoretical effort has been addressed to identifying the sedimentologic signature of ice shelves (e.g., Carey and Ahmad, 1961; Drewry and Cooper, 1981; Orheim and Elverhoi, 1981; Powell, 1984; Drewry, 1986, pp.201-216) and this effort probably has been largely successful. Unfortunately, very few hard data are available on the subject. The only comprehensive observations taken beneath a large ice shelf 0025-3227/89/$03.50

are those from site J9 on the Ross Ice Shelf (Figs.1 and 2; Clough and Hansen, 1979). There, only sparse biological activity was observed and no measurable thickness of modern sediments was present (Webb et al., 1979). As discussed below, this may be typical of large areas of modern and ancient ice shelves, but it provides little help in recognizing sub-ice-shelf sediments when they are deposited. Until,.an ambitious project to drill through ice shelves is mounted or a scientific submersible is taken beneath an ice shelf, we can expect few new observations of sub-ice-shelf sedimentation. Any contribution we make to the literature here, thus, must follow earlier woi-kers in

© 1989 Elsevier Science Publishers B.V.

102 20 ° W

~O°W

0°

IO°E

20°E

o~ t

i

LAZAREV

j¢

"%,.

,%

i

I

! . \ ,-i,oo.

-,,

. GEORGE

4

-90°W

,':-..

•

,,

lO°S

o.~.

ICE SHELF /' 0 I

km i

t000 I

)

20"S

%

Fig.1. L o c a t i o n map of A n t a r c t i c a . Ice shelves are s h o w n stippled. On grid scale, 1° = 111 kin.

relying heavily on hypotheses, speculation, and probabilities rather t h a n direct observations. Bearing this in mind, we will first summarize the glaciological constraints on ice-shelf formation and on types and rates of sediment deposition beneath ice shelves. We will then discuss the new observations regarding till

deformation beneath ice stream B at the head of the Ross Ice Shelf (Fig.2) and their implications for sub-ice-shelf sedimentation, leading to a new hypothesis for the sedimentary history of the Ross Sea area. We will present this hypothesis, compare its predictions to our observations, and highlight those areas where more data are required.

103 ( t 8 0 °)

6°W

4=

2=

0°

4°

:::i:° : :i ::ii!i SIPLE

~ I ( ~ ERISE

ed9 i:

~

!i! ::::::i~::

0 I

I

km

I

300 I

Fig.2. Map showing Ross ice streams flowing into Ross Ice Shelf; ice streams are s h o w n stippled. Modified from Shabtaie and Bentley (1987). Grounding line of Rose (1979) is shown on the ice streams by light dashed line and grounding line of Shabtaie and Bentley (1987) is shown by ticked solid line; ice plains occur between the two. Major camps and features are indicated. See text for further explanation.

Conditions for i c e - s h e l f f o r m a t i o n

An ice shelf is "a sheet of very thick ice, with a level or gently undulating surface, which is attached to the land along one side but most of which is afloat and bounded on the seaward side by a steep cliff"; the term was used by Sir Douglas Mawson in 1912 (Bates and Jackson, 1980, p.309). With very few exceptions (e.g., Gow, 1967; Jeffries et al., 1988) the upglacier end of an ice shelf consists of snow and ice that accumulated on a grounded ice sheet and then flowed downglacier until it crossed the grounding line (Hollin, 1962) and began to float. Along flow on an ice shelf, ice formed from snow accumulation becomes increasingly important (Thomas and MacAyeal, 1982) and ice formed from basal freeze-on of sea water may become important (Morgan, 1972; Robin, 1979; Lange and MacAyeal, 1986; Engelhardt and Determann, 1987).

An unconfined ice shelf spreads rapidly under its own weight at a rate that is easily calculated (Weertman, 1957; Robin, 1958; Thomas, 1973a,b). If an ice shelf were composed of temperate ice, spreading would be very rapid and water-filled surface crevasses would be able to join with basal crevasses (Weertman, 1973) and cause rapid calving (Rasmussen and Meier, 1982) and ice-shelf disintegration. Iceshelf spreading and calving are reduced by lower temperatures, confinement in an embayment, and interaction with local "pinning points" where the ice shelf becomes grounded on high places in the sea floor. Rapid ice flux across the grounding line also tends to counter the rapid spreading of an ice shelf. Existence of an ice shelf thus is favored by rapid input from grounded ice (ice streams or outlet glaciers; Robin, 1979; Bentley, 1984; Powell, 1984) into an embayment with local high spots in the bed and with cold temperatures (average midsummer temperature < 0°C (Mercer, 1978) or mean annual temperature < - 1 0 ° C (Robin, 1979)). This does not mean that temperate ice shelves cannot exist, but only that the requirements of lateral constraint and protection from bottom melting to sustain a temperate ice shelf are sufficiently severe that none exists today; however, the George IV Ice Shelf (Fig.l; Paren and Cooper, in press) comes close, possibly in response to recent climatic warming. Most modern ice shelves exist in the Antarctic (Fig.l), where they occupy (including ice rises, which are local regions of ice outflow formed over pinning points) about 47% of the coastline of permanent ice and 12% of the surface area of permanent ice, but contain < 3% of the total volume of permanent ice (Drewry et al., 1982). The small volume of ice shelves and ice rises occurs because ice shelves average 475 m thick and ice rises 670 m thick, whereas grounded ice excluding ice rises averages 2450 m thick (Drewry et al., 1982). The huge Ross and Filchner-Ronne ice shelves (Fig.l), again including ice rises, together account for 66% of the area and 74% of the volume of all Antarctic ice shelves, but only 9% of the ice-shelf coastline (Drewry et al.,

104

1982; Doake, 1985). The Ross and Filchner Ronne ice shelves also drain about 30% of the mass output from the Antarctic Ice Sheet (Giovinetto and Bentley, 1985). Constraints on sub-ice-shelf sedimentation

Sedimentation beneath ice shelves has been discussed thoroughly by a number of authors (e.g., Drewry and Cooper, 1981; Orheim and Elverh~i, 1981; Powell, 1984). The basic conclusion of these papers is that debris melting out of the base of an ice shelf will settle through the water column to form a dropstone diamicton, which may be subject to slumping or winnowing by currents following initial deposition. Deposition is limited to regions of basal melting and is limited by the amount of debris carried in the ice.

Basal melting Identification of areas of basal melting beneath an ice shelf, and thus of potential sedimentation, is easy in theory but difficult in practice. Possible methods are borehole and core studies, including identification of frozenon saline ice and analysis of temperaturedepth profiles (Morgan, 1972; MacAyeal and Thomas, 1979; Zotikov et al., 1979; Budd et al., 1982; Engelhardt and Determann, 1987), radar sounding (Neal, 1979; Thyssen, in press), resistivity measurements (Shabtaie and Bentley, 1979), mass-balance calculations (e.g., Thomas, 1976), water-mass studies in oceans adjacent to ice shelves (Doake, 1985; Jacobs et al., 1985), and thermodynamically coupled ice-shelfocean models (MacAyeal, 1985). All such methods suffer serious shortcomings, however. Borehole studies and resistivity are point measurements and would require extensive ground programs to characterize an ice shelf. Mass-balance calculations require a long time series of data to remove nonsteady effects. Radar sounding and thermodynamic models are not sufficiently well characterized to be relied upon routinely, and water-mass studies

cannot determine where beneath an ice shelf melting and freezing are occurring. Despite these difficulties, some useful data have been collected. Most ice shelves experience net basal melting; two measured values are about 0.3 m a ' averaged over the Ross Ice Shelf (Pillsbury and Jacobs, 1985) and 2 m a 1 for the George VI Ice Shelf (Doake, 1985). Basal melting is most rapid toward the seaward edges of ice shelves. Large and highlatitude ice shelves have significant regions of basal freeze-on in interior zones (Morgan, 1972; Zotikov et al., 1979; Engelhardt and Determann, 1987). Many authors have made the (untested) hypothesis that there is a zone of basal melting near the grounding line even if basal freezing occurs downglacier, owing to the pressure dependence of the melting point of ice and the sloping base of the ice shelf (the "ice pump"; Robin, 1979; Drewry and Cooper, 1981; Orheim and Elverhoi, 1981; Powell, 1984). A parcel of water in thermal equilibrium with ice at some depth (and thus pressure) will melt ice if it is forced to a lower level (higher pressure) adiabatically, and will allow ice formation if forced to a higher level (lower pressure); an icepump circulation thus can be established along the sloping base of an ice shelf in which ice is melted off at great depth and frozen back on at lesser depth (Foldvik and Kvinge, 1974; Doake, 1976; Robin, 1979; Lewis, 1985).

Sediment supply Small amounts of sediment may be generated biogenically beneath an ice shelf (Clough and Hansen, 1979), transported beneath an ice shelf by currents, or deposited on the surface of an ice shelf through eolian processes and eventually melted off the bottom. However, such processes are likely to be insignificant on a large ice shelf, where sedimentation will depend primarily on debris transport from regions of grounded ice. Such transport can be subglacial, basal, high englacial, or supraglacial. (Supraglacial debris rests on top of a glacier, subglacial debris is beneath a glacier,

105

and englacial debris, debris within a glacier, comprises basal debris near the bed of the glacier and high-englacial debris farther from the glacier bed.) Debris transported supraglacially or englacially eventually must be deposited as a dropstone diamicton beneath the ice shelf or from icebergs calved from the ice-shelf front, whereas subglacially transported debris must be deposited at or near the grounding line. As noted above, most ice flowing into ice shelves does so from outlet glaciers (fastmoving ice flowing between rock walls) or from ice streams (fast-moving ice flowing between regions of slow-moving ice; Bentley, 1987). Outlet glaciers receive supraglacial debris from rockfalls, and may contain high-englacial debris from formation of medial moraines where tributaries join. Basal melting may concentrate some of this debris in a basal or subglacial transport zone, and erosion of the bed may produce further debris for basal or subglacial transport. However, polar ice sheets capable of generating ice shelves typically lack large areas of rock outcrops and thus lack abundant supraglacial debris and medial moraines. For the Ross Ice Shelf, about two-thirds of the ice crossing the grounding line is fed from ice streams without rock walls (Shabtaie and Bentley, 1987). This ice may contain some highenglacial debris picked up during flow over subglacial peaks upstream (Bentley, 1971); however, the lack of seismic reflections from such debris beneath the Upstream B camp (UpB, Fig.2) on ice stream B suggests that such debris is sparse if present. It is also unlikely that the ice streams contain abundant basal debris. Basal debris may be entrained through freezing (regelation or net freeze-on) or structural processes (shearing or folding-in). The thickness of the debrisrich basal layer depends on the balance of further structural deformation, net vertical strain (compressive or extensional), basal freezing or melting, and clast dispersion. Ice streams typically are regions of extensional flow and basal melting. This means that

downward ice velocities exceed upward debris velocities from local deformation or dispersion (RSthlisberger, 1968; Weertman, 1968), restricting debris to a thin regelation layer (Weertman, 1964). Debris-rich basal ice formed in a catchment area would need to be quite thick to survive passage down an ice stream. Taking ice stream B as an example, about 25 m of ice are melted off the base during passage down the ice stream, and layers of ice undergo strain thinning by about a factor of two (calculated using data from Bindschadler et al. (1987a,b), Shabtaie and Bentley (1987), Whillans et al. (1987) and Alley and Bentley (in press)). A debris-rich basal layer would need to be more than 50 m thick to survive passage down the ice stream, much thicker than is likely to occur (e.g., Gow et al., 1979). Debris-rich ice flowing from an interstream ridge has a shorter path along the ice stream and is more likely to survive to the ice shelf, but the ice volumes involved are small (Shabtaie and Bentley, 1987). Two possible special circumstances may increase the englacial debris of ice stream B. The downstream end of the ice stream is an ice plain (Bentley, 1987; see below) with low heat of sliding. Depending on the exact geothermal flux, this region may experience slow basal melting or slow freeze-on of perhaps 0.1 m of debris-rich basal ice during passage to the grounding line. In addition, backpressure from Crary Ice Rise causes flow to become longitudinally compressive near the downstream end of ice stream B, although with lateral extension of typically greater magnitude (Bindschadler et al., 1987a,b). Debris may be sheared or folded into the ice there, or squeezed up into longitudinal basal crevasses. We consider it unlikely that such structurally entrained debris will be large compared to other fluxes discussed below, and we make this assumption here, but further data are needed. The englacial debris flux from ice stream B or a similar ice stream thus is likely to be limited to an active regelation layer only a few

] 06

millimeters thick (Weertman, 1964). Taking the average thickness of this layer to be about 10mm (Kamb and LaChapelle, 1964; Weertman, 1964) and the debris concentration in it to be 5% by volume (Sugden and John, 1976, p.161), and taking the porosity of sub-ice-shelf sediments to be 25%, the debris flux t hr ough the regelation layer would equal about 0.3 m 3 a ~ of sediment per meter-width across the 100km-wide mouth of ice stream B (Shabtaie and Bentley, 1987). Melting of this layer along the approximately 103-km-long flow band of the Ross Ice Shelf fed by ice stream B would cause an average deposition rate of about 0.3 m Ma 1. If we allow for possible structurally entrained material, basal freezing on the ice plains, and possible higher debris concentration for the regelation layer, the estimated englacial debris flux from ice stream B might be increased by one or two orders of magnitude. However, the discussion above, plus our seismic observations at UpB and at the Downstream B camp (DnB, Fig.2) make it highly unlikely th at englacial debris flux could be increased by an additional two to three orders of magnitude beyond this t hr ough occurrence of a basal debris zone about 100 m thick and with 10-50% debris by volume (although with slower ice velocity), as proposed by Drewry and Cooper (1981). These calculations suggest t hat englacial debris transport from ice streams is very slow, but place no constraints on subglacial debris transport. Subglacial transport can occur by deformation of subglacial sediment (Boulton, 1979; Alley et al., 1986) or by stream or waterfilm transport. Here, however, it is necessary to remember t h a t water fluxes across ice-shelf grounding lines are relatively small because ice shelves generally lack a supply of surface meltwater to the bed. The water flux from ice stream B, which averages about 50 km wide, 400 km long, and drains an area of about 217x103 km 2 (Shabtaie and Bentley, 1987) is probably about 20m 3 s -1 (Weertman and Birchfield, 1982; Bindschadler, 1983; Lingle and Brown, 1987;

this study). By comparison, the summer water flux from Columbia Glacier, an Alaskan tidewater glacier only about 5 km wide in its lower reaches, 67 km long, and draining an area of 1.1 × 103 km 2 (Meier et al., 1985) is about 400 m * s i (Walters et al., 1986). The water flux from ice stream B would fill a single channel about 1.4 m in radius, or a water film about 0.01 m thick (calculated following Weertman. 1972 and Weertman and Birchfield, 1982). For a given water flux, the sediment flux is approximately inversely proportional to the square of the number of channels through which it flows (Leopold and Maddock, 1953, p.21; Bloom, 1978, p.216). To estimate icestream sediment discharge if flow were fully channelized, we can use an analogy to Variegated Glacier in its pre-surge state. At t hat time, Variegated Glacier had a sliding velocity similar to ice stream B, had channelized water discharge from its lower stream about equal to the estimated flux from ice stream B, and had a suspended sediment load of 2.5 kg m - 3 (Humphrey, 1986; Humphrey et al., 1986). Assuming that suspended sediment is 50% of the total load (Bloom, 1978, p.213; Hooke et al., 1985), this would suggest a sediment flux of about 100 kg s 1 from a single channel draining ice stream B, equivalent to a sub-ice-shelf sedimentation rate of about 10 m 3 a 1 m - 1 width of ice stream or about 10 m Ma i averaged over the ice-shelf length. This is an upper limit for sediment transport in streams beneath ice stream B. In the lower limit of l0 T channels 10 mm in diameter, (approximately equal to a 100 km-wide, 10 mm-deep water film) the simple inverse-square law indicates a drastic reduction in sediment flux (by about 14 orders of magnitude). A distributed water system would also be limited to transporting clay and silt, whereas a single channel would transport clasts as large as cobbles (Hjulstr6m, 1935, 1939; Singer and Anderson, 1984). Several lines of evidence suggest t hat the distributed water system limit is more nearly correct than the single-channel limit for most polar ice streams and outlet glaciers. Although channels have lower water pressures t han

107

films (Weertman, 1972), theoretical calculations show t h a t the stress distribution around a channel incised upward into the ice (a RSthlisberger or R channel) prevents it from collecting basal water, and that a system of channels incised downward into rock (Nye or N channels) can exist and collect basal water only under special circumstances if channelized surficial meltwater is not supplied to the bed (Weertman and Birchfield, 1983). Below we argue that ice stream B rests on a deforming till. If so, this strengthens the argument against channelized water flow. Deforming till will creep rapidly into a lowpressure region such as an incipient channel, unless previously channelized flow reduces the water pressure enough to increase the till strength and suppress local deformation (Boulton and Hindmarsh, 1987; Alley, in press). Thus, channelized surficial meltwater supplied to the bed can remain channelized and remove large quantities of debris by sweeping across the bed; however, in the absence of a channelized source, basal water will tend to remain in a distributed system and will transport relatively small quantities of relatively fine-grained sediments. The high basal water pressures observed directly at Byrd Station (Fig.l; Alley et al., 1987c) and inferred seismically at UpB (Blankenship et al., 1987) are significantly above what would occur if water flow were well channelized (Bindschadler, 1983), confirming out theoretical discussion for those locations. We thus will assume t h a t there is little sediment transport by channelized water beneath ice stream B and similar cold glaciers. The reader should note t h a t the above maximum estimate of stream transport is one to two orders of magnitude less than the debris flux by bed deformation estimated below, so errors in our treatment of stream transport should not be too significant. Also, Columbia Glacier has regions of deforming subglacial till 0.1-0.5 m thick (Meier, in press) despite summertime water drainage per unit width of grounding line about 400 times larger than for ice stream B, so it seems unreasonable to

expect water drainage to dominate the sediment budget of ice stream B. Very close to the ice-stream grounding line (within 0 (1 km)? where the symbol 0(x) means "of the order of magnitude of x") the ice approaches flotation as the basal water pressure approaches the overburden pressure, and this reduces both the till strength and the tendency for till to creep into R channels. The possibility then exists of channelized flow of water or of fluidized debris in this region (Boulton and Hindmarsh, 1987). However, this region probably is wide compared to its length, so it is unlikely t h a t the flow can become concentrated in one or a few channels. Estimated sediment fluxes from cold ice streams and outlet glaciers are listed in Table 1, although order(s) of magnitude uncertainties are associated with most of these values. High-englacial and basal fluxes from outlet glaciers are based in part on Drewry and Cooper (1981). Transport in subglacial water streams beneath an ice stream was calculated above, and we assume similar behavior for outlet glaciers. Debris flux by till deformation beneath ice streams is taken from the estimated flux beneath ice stream B, discussed below; no data are available on whether till deformation occurs beneath outlet glaciers. The table illustrates that till deformation, if active, is likely to dominate sediment flux and t h a t englacial and stream transport are likely to be small on outlet glaciers and smaller on ice streams. TABLE 1

Possible debris fluxes to an ice shelf from ice streams and outlet glaciers, in cubic meters of glaciomarine sediment per meter width of grounding line per year Debris flux (m 3 m-i a-i)

High englacial Basal Streams Till deformation

Ice stream

Outlet glacier

0-I 0.i-I < 0.i 100-1000

0.1-10 0.1-100 0-10 ?

Thus far we have considered debris transport from the main body of grounded ice (although discussion of debris transport by subglacial deformation is deferred to below), but we have ignored the possible role of localized grounding of the ice shelf at pinning points. Two types of grounding can occur. If the ice shelf contacts the bed but the region of contact continues to participate in the main ice-shelf flow, then ice rumples develop. If the region of grounding ceases to participate in the main ice-shelf flow and becomes a local center of ice outflow, it is an ice rise. Both can contribute sediment to the ice shelf, but are unlikely to be significant sources. From Drewry et al. (1982) and Doake (1985), ice rises account for only about 0.5% of the area of grounded ice draining into the Ross Ice Shelf. Flow off ice rises is very slow, because they have small accumulation areas. Ice rumples are inferred to be wet-based (because they exhibit the high velocities but low driving stresses of ice-shelf flow) and could experience erosion and supply debris to the ice shelf. However, ice rumples are relatively localized regions where the ice is not grounded too strongly, so rapid erosion in the absence of falling sea level would quickly remove the bedrock high causing the ice rumples and thus remove the source of debris. E v i d e n c e f r o m ice s t r e a m B

Till deformation For the last several years, investigators from the University of Wisconsin, Ohio State University, NASA/Goddard Space Flight Center, the University of Chicago, and other institutions have been conducting a cooperative glaciological and geophysical survey of the Siple Coast, which is that region where ice from West Antarctica drains into the Ross Ice Shelf (Fig.2). Seismic surveys at the UpB camp on ice stream B have shown that the ice stream there is underlain by a layer, several meters thick, of water-saturated, unconsolidated sediment with about 40% porosity in which the water pressure is within about 50 kPa (0.5 bar)

of the overburden pressure (Btankenship et al., 1986, 1987). The layer is continuous, or nearly so, along a 9 k m line transverse to flow (Rooney et al., 1987) and along a 12.5 km line in the direction of flow (Rooney et al., in press b). The layer has a smooth upper interface with the ice, but the lower surface is an angular unconformity that is eroded into flutes about 10m deep and 300-1000m across, oriented parallel to ice flow. The rocks beneath the angular unconformity probably are poorly consolidated Neogene marine or glaciomarine sediments (Rooney et al., in press a). Preliminary seismic results from the DnB camp (Fig.2) suggest that the layer is present there and has a similar thickness (Blankenship et al., in press). The high porosity of this several-meter-thick subglacial layer indicates that it is dilated and thus is deforming; force-balance and waterbalance calculations also suggest that it is deforming. We thus have hypothesized (Alley et al., 1986, 1987a,b) that this layer is a deforming, water-saturated till similar to that described by Boulton (1979) beneath BreidamerkurjSkull in Iceland, that it extends beneath the entire ice stream, that the ice velocity arises largely from deformation within this till, and that erosion/remobilization of soft subjacent sediments by clasts in this till has eroded the observed angular unconformity. (Drewry (1986, pp.61-62) has discussed the analogous case of erosion by clasts in ice frozen to its bed.) In the discussion that follows we will assume that our interpretation of the seismic data from UpB as indicating a deforming till is correct; however, the reader should remember that direct measurement of a velocity profile within the several meter thick layer beneath a kilometer of ice has not been conducted, although such measurements are planned. We have estimated the velocity profile and thus the till flux in the layer, and find that it corresponds to a steady-state erosion rate of the order of tenths of a millimeter of rock per year averaged over the catchment area and the upstream part of the ice stream (Alley et al.,

109 1987a). This sediment flux also requires relatively rapid deposition at the grounding line. We have estimated (Alley et al., 1987a) a rock flux of hundreds of cubic meters per year per meter-width of grounding line, which would have formed a deposit tens of kilometers long into water tens of meters deep if the grounding line has been near its present position for the last 5-10 ka (Thomas and Bentley, 1978; Greischar and Bentley, 1980). Geophysical data suggest that such a deposit does exist at the grounding line of ice stream B (see below). Till deltas

Terminology for such an extensive grounding-line deposit poses a small problem. Powell (1981, 1984) argues that grounding-line deposits of ice shelves should be called "morainal banks", but describes such banks as "elongate ridges or isolated mounds" comprising "grounding-line melt-out till, dropped, compound, and residual para-tills...fluvial sediment and sediment gravity flow deposits" (Powell, 1984, p.19). The features we propose here are not elongate ridges or isolated mounds, and, as described below, we believe that they comprise basal-till topsets with minor sorted sediments, and gravity-flow foresets and bottomsets; the topsets will parallel the base of the ice and may dip upstream. Other possible terms, such as kame delta, delta kame, kame moraine, and delta moraine carry implications of dominance by meltwater or subaerial, ice-marginal deposition (Sugden and John, 1976; Bates and Jackson, 1980). We have informally termed these deposits "till deltas" to emphasize their deltalike nature and the likely dominance of till in the topset beds (Alley et al., 1987a), and we will continue this informal usage here until such time as a more formal terminology can be established. A thick, extensive accumulation of sediment near the grounding line, where the water pressure is almost as large as the overburden pressure, would be quite soft and would support only a small basal shear stress. This in

turn requires a small ice-air surface slope which implies a small pressure gradient driving water flow, a thickened water film, and enhanced sliding between ice and till in addition to ongoing till deformation. The base of an ice shelf typically rises downstream, and if sediment filling the sub-ice-shelf cavity retained this slope, water drainage would be slowed further. The downglacier end of a till delta is the grounding line, where flotation begins, and we have called the upglacier end the "coupling line" (Fig.3; Alley et al., 1987a) where the ice-stream surface slope decreases onto the delta. It seems unlikely that water drainage or possible basal freeze-on over the till delta could remove most of the till supplied from upglacier by deformation, as discussed above, although limited sediment sorting might occur. Deformation then must continue across the delta, creating a several-meter-thick topset bed that may have a shallow (_< 1 °) upglacier dip. The highly unconsolidated, water-saturated sediment transported through this topset must lose contact with the ice at the grounding line, leading to slumping and development of foreset and bottomset beds of turbidites and debrisflow deposits (Prior et al., 1981; Powell, 1984). The depositional dip of foreset beds might be similar to dips in other low-energy, turbiditedominated, progradational clastic settings, or about 1° or less in the downstream direction (Mitchum et al., 1977; Sangree and Widmier, 1977).

COUPLING GROUNDING ICE LINE LINE FRONT CE STREAM I DELTA I CE SHELF ISEA WATER

I

MAX.

I

ORESETS

I

%~%~

BEDROCK

Fig.3. C a r t o o n o f the l i k e l y c o n f i g u r a t i o n o f the ice stream, t i ] ] delta, and ice shelf.

110

When we first predicted the existence of a large, lobate region of grounded ice at the mouth of ice stream B, the grounding line was mapped as shown by the lightly dashed line in Fig.2, with a surface slope immediately upstream of the grounding line about equal to values farther upstream (Rose, 1979). Detailed mapping using satellite altimetry and airborne-radar data now has shown that the grounding line is as indicated by the solid line in Fig.2, tens of kilometers downstream of the old grounding line (Shabtaie and Bentley, 1987). The surface slope on the newly discovered regions of grounded ice, or "ice plains" (Bentley, 1987), is significantly less than on the main part of the ice stream and is almost as small as ice-shelf values (Shabtaie and Bentley, 1987). Reinterpretation of seismic data collected during RIGGS on the ice plains suggests that, at least in most places, they are underlain by water-saturated, unconsolidated sediments probably tens of meters thick (Shabtaie and Bentley, 1987). Preliminary analysis of new seismic data from near the DnB camp (Fig.2) shows a several-meters-thick layer overlying a unit tens of meters thick containing beds dipping downstream at about one half of a degree; these may be the predicted topset and foresets (Blankenship et al., in press). The grounding line mapped by Rose (1979) probably represents the coupling line. Of course, the simple existence of unconsolidated sediments beneath an ice shelf does not demonstrate that the sediments were transported by subglacial deformation. Rapid melting of debris-rich basal ice at the grounding line and discharge of subglacial melt streams can form morainal banks at grounding lines (Powell, 1984). Melt-stream discharge would build a deltaic deposit, but this would have sorted sediments and thus would tend to have steeper foreset beds, perhaps dipping 1°-10 ° (Mitchum et al., 1977; Sangree and Widmier, 1977). Basal melt-out probably would produce a dropstone diamicton layer lacking deltaic form .

Based on radar data, Drewry et al. (1980,

p.48) suggested that part of the grounding line of the Filchner Ice Shelf is "just touching a sea floor, composed of soft, water saturated sediment". Drewry and Cooper (1981) noted that this zone was tens of kilometers wide and suggested that it was formed by sedimentation owing to debris melt-out filling the sub-iceshelf cavity. A similar zone of soft basal sediments in contact with ice near the grounding line in a region previously assumed to be afloat was discovered by drilling on the Lazarev Ice Shelf (Korotkevich et al., 1978). In the case of ice stream B, the sediment volume beneath the ice plains seems too large to have been deposited by melt-out or by subglacial streams without a deforming till in the time during which the grounding line has been near its present position (assuming that this sediment is post-Wisconsinan), and the seismically observed sedimentary structures resemble those expected from a deforming till. A deforming till cannot be ruled out for the Filchner Ice Shelf and elsewhere, but data to assess this are not available.

Implications of till deltas The existence of till deltas introduces certain other interesting possibilities. Because of the small surface slope over a till delta, a relatively small rise in sea level could cause a relatively large grounding-line retreat. For example, on ice stream B a sea-level rise of about 25 m would cause a grounding-line retreat of about 75 km (Shabtaie and Bentley, 1987). However, this would introduce a water layer varying from 25 m thick at the old grounding line to zero thickness at the new grounding line, and the estimated sediment flux from the ice stream of hundreds of cubic meters per year per meterwidth would fill this water layer with sediment in only 0(103 a). Sea-level rise slower than 0(10- 2 m a- 1) would be compensated entirely by sedimentation on the till delta of ice stream B and would cause no grounding-line retreat (Shabtaie and Bentley, 1987). The stabilizing effect of sedimentation on grounding-line position would he lost if the

111 sediment supply were terminated. It now is clear t h a t ice stream C was an active ice stream t h a t stopped about two centuries ago (Shabtaie and Bentley, 1987). This stoppage reduced the ice flux across the grounding line and ongoing ice-shelf spreading is causing thinning of the ice there; the effect of such thinning on grounding-line position is analogous to the effect of sea-level rise. The stoppage also ended sediment supply to the grounding line (presuming t h a t ice stream C moved by a simiPar mechanism to ice stream B). The grounding line of ice stream C now is retreating in response to the thinning ice (Thomas et al., in press), and the shape of the grounding line (Fig.2) suggests t h a t it has retreated tens of kilometers along most of the ice-shelf front since ice stream C became inactive. (The region of ice stream C closest to ice stream B may have remained grounded on its till delta thus far because flow lines from ice stream B have bent towards ice stream C (Shabtaie and Bentley, 1987), partially replacing the ice loss from stoppage of ice stream C and preventing rapid thinning of ice there.) Another interesting possibility comes from our calculation, above, that partial ice-till decoupling across a thickened water film on the till delta reduces the flux of deforming till below its maximum possible value. If any perturbation to the system were to thin the water film, till flux across the delta would increase by erosion of the head of the delta. For example, a falling sea level would increase the interaction of the ice shelf with pinning points downstream, increasing the backstress and probably causing the ice over the till delta to thicken and steepen to maintain force balance. This would increase the pressure gradient driving water flow, thin the water film, increase ice-till coupling, and cause till flux across the delta to exceed till input from upstream of the delta (Alley et al., 1987a). This would give the classic '~conveyor belt" grounding-line advance (Powell, 1984), in which the upstream end of the delta (the coupling line) and the grounding line advance through sediment recycling. Behind the advancing till delta

would be an extended ice stream lubricated by a thin till layer. This idea figures in our hypothesis for the Quaternary history of the Ross Embayment, discussed below.

Sedimentary history of the Ross Embayment: A unified hypothesis The sedimentary history of the Ross Sea now is receiving considerable attention and has spawned much controversy; both depositional environments and ages of sediments sampled are subject to dispute. This is due largely to a shortage of samples, the highly reworked character of most materials sampled, and ongoing refinement of biostratigraphic zonations (Karl et al., 1987). Our new evidence concerning the physics of glacier flow on ice stream B gives us considerable insight into how the system may have responded to sea-level fluctuations t h a t are known to have occurred in the past. We find (Alley et al., 1987a) that the likely response of the ice sheet to sea-level fluctuations would lead to a sedimentary sequence similar to that described by some observers. Here we will give a brief summary of observations of sediments from the Ross Embayment, including areas of controversy, and then will present our hypothesis for the sedimentary history and show how it relates to the observations. Observations To summarize from Houtz and Davey (1973), Hayes and Frakes (1975), Anderson et al. (1984), Dunbar et al. (1985), and Karl et al. (1987), the modern Ross Sea (Fig.4) is underlain by a sedimentary column typically hundreds of meters thick. The upper part of this sedimentary column contains a prominent unconformity, often called the Ross Sea unconformity, at a depth of < 2 m to about 40 m below the sea floor. Erosion on this unconformity may have amounted to as much as several hundred meters. This erosion probably occurred beneath grounded ice, although marine bottom currents have been suggested as a possible

8°

i0 o I%

0 8 0 o) 4o

6o

2°

0

-

2+

4+

/'

:1

,/

/

/, d~

\ BYRD 5 o ~ - - i

]oc

j !

~I~O~/,A"

750"

J

\\',,,

e^

3

7°

9oii

I1°! k

EDWARD"

I.

t i +3o~L-

/]

i 15o

]

. . . . . .

/

/

/

" " ~

;'5oS-_

/

¢

0

!

17°

0°

i

+

o

Fig.4. B a t h y m e t r y (depth below sea level in meters of the base of the water column or of the grounded ice) of Ross Sea and most of the area of W e s t A n t a r c t i c a draining into it. T h e contour interval c h a n g e s from 50 m in the Ross Sea and under the Ross Ice Shelf to 250 m b e n e a t h the West Antarctic inland ice. On a grid scale, ] ° = 111 km. Byrd Station, Upstream B camp (UpB), and J9 are shown. From B e n t l e y and J e z e k {1981).

113 cause (Mercer and Sutter, 1982). This unconformity is overlain by an unstratified or poorly stratified diamicton, which in turn is overlain by a thin veneer of Holocene sediments comprising ice-rafted clasts, terrigenous silt and clay, and biogenic silica; in this Holocene layer ice-rafted debris is sparse near the front of the Ross Ice Shelf, indicating that little englacial debris reaches the ice-shelf front but is more abundant where outlet glaciers drain into the Ross Sea. The transition from the unstratified diamicton just above the unconformity to the modern sediments is marked by a water-sorted unit 0.1-0.5 m thick at some sites (Kellogg et al., 1979) but not in most locations (Anderson et al., 1984). • The ages of the Ross Sea unconformity and of the overlying diamicton are uncertain (Hayes and Frakes, 1975; Kellogg et al., 1979; Savage and Ciesielski, 1983), but the unconformity probably is Pliocene or Pleistocene in age. Based on appearance, grain-size distribution and other characteristics, Kellogg et al. (1979), Anderson et al. (1980, 1984), and others have argued that the diamicton above the Ross Sea unconformity is a basal till. Anderson et al. (1984) show that populations of transported clasts in this material are not mixed, but are traceable to discrete sources in the Transantarctic Mountains and in West Antarctica. Truswell and Drewry (1984) demonstrated that pollen grains in the diamicton also are traceable to discrete sources. This shows that deposition occurred from a grounded ice sheet, or from an ice shelf with marine currents too slow to mix pollen, rather than from floating icebergs. Estimated sedimentation rates of 6-8 m Ma 1 (Hayes and Frakes, 1975) seem too high for sedimentation beneath an ice shelf fed by West Antarctic ice streams, as discussed above, so we consider that this diamicton probably is a basal till. Notice, however, that several authors including Hayes and Frakes (1975) and Fillon (1979) have interpreted the diamicton as glaciomarine rather than as basal till. Beneath the Ross Ice Shelf, data are available only from the J9 site, downstream of ice

stream B, where the bed is about 590 m below sea level (Figs.2 and 4; see Clough and Hansen, 1979). Sediments were collected there to a maximum depth of about 1 m. These sediments originally were interpreted as a dropstone diamicton of Miocene age containing clasts transported from the Siple Coast; observed differences between the upper 0.1-0.2 m and deeper material were interpreted as the result of diagenesis in situ (Webb et al., 1979). Vigorous debate has focused on whether Pliocene or Pleistocene fossils are present in the material, which certainly is dominated by Miocene forms (e.g., Kellogg and Kellogg, 1981, 1983; Brady, 1983; Kellogg and Kellogg, 1986). More recent studies (Harwood, 1986; Harwood et al., in press a, b) question whether any undoubtedly post-Miocene diatoms occur at J9 and support a Middle-Late Miocene age for the youngest documented event of marine productivity in the interior Ross Embayment. Anderson et al. (1980) have reconsidered the physical properties of the J9 sediments and conclude that the sediment probably is a basal till rather than a dropstone diamicton. However, Harwood et al. (in press b) present evidence suggestive of a dropstone origin. A basal till might contain only recycled fossils from its source area, whereas a water-lain sediment probably would contain some fossils indicative of the time of deposition. In this regard it is worth noting that Raiswell and Tan (1985) have interpreted the chemistry of the J9 cores as indicating Pleistocene deposition of at least the upper 0.1-0.2m, and possibly the entire length. North of the Ross S e a , the continental rise and abyssal plain are blanketed by a Tertiary sedimentary wedge hundreds of meters to kilometers thick (Hayes and Frakes, 1975). Miocene to Recent sediments in this wedge contain clasts transported by ice (Hayes and Frakes, 1975).

Glaciological considerations Some glaciological models, including those of Thomas and Bentley (1978) and Stuiver et al.

114

(1981) have reconstructed the Wisconsinanmaximum West Antarctic ice sheet as having advanced to the edge of the continental shelf and having developed an equilibrium, East Antarctic-type surface profile characterized by a steep surface slope near the coast and thick ice with a gradual surface slope inland. Other authors have suggested that the Wisconsinanmaximum West Antarctic ice sheet was grounded in what is now the Ross Sea but exhibited a low, ice-stream surface profile (Thomas, 1979; Denton et al., 1986; Alley et al., 1987a). Drewry (1979) summarizes evidence from West Antarctica against an equilibrium high-profile ice sheet in the Ross Sea, noting that such an ice sheet would require greater increases in ice thickness in the vicinity of Byrd Station than are allowed by the data of Whillans (1976) and Robin (1977). Drewry (1979) then presents the hypothesis that the Ross Ice Shelf expanded during the Wisconsinan but that the grounding line advanced only about as far as J9. The Ross Embayment is relatively deep near the modern grounding line, shallows outward to the edge of the continental shelf, and has relatively constant width from the modern grounding line to the shelf edge (Fig.4). In the absence of large increases in marginal ablation, a sufficiently large sea-level drop for a sufficiently long time necessarily would allow the Ross Ice Shelf to become fully grounded and allow grounded ice to expand to the edge of the continental shelf, regardless of basal conditions. The minimum sea-level drop required for this to occur has been termed a critical value and estimated as about 120-130m for 0(103--104 a) (Weertman, 1974; Thomas and Bentley, 1978; Drewry, 1979; see also discussion by R.H. Thomas appended to Drewry, 1979). The ability of a morainal bank or till delta to cause grounding in water t h a t otherwise is too deep (Powell, 1984) suggests that conveyor belt recycling of a till delta would allow groundingline advance to the edge of the continental shelf for a sea-level drop less than the critical value. In this case, the rate of grounding-line

advance would be limited by the rate of' tilldelta recycling. Data summarized by Drewry (1979) show that the actual Wisconsinanmaximum drop in sea level was within a few meters or tens of meters of this critical value, but whether the critical value was achieved for a sufficiently long time, if at all, is uncertain.

Hypothesis Wisconsinan-maximum sea-level fall caused the grounding line of the West Antarctic Ice Sheet to advance across the Ross Sea to the edge of the continental shelf. Sea-level fall caused increased interaction of the ice shelf with pinning points, increasing backstress on grounded ice and causing ice over the heads of till deltas to steepen and thicken to maintain force balance. This caused the water film at the ice-till interface to thin and increased the icetill coupling and the till flux across the delta. The resulting conveyor-belt recycling of the delta accompanied the grounding-line advance, and may have been required to allow the grounding-line advance if the actual sealevel fall was less t h a n the critical value for a grounded ice sheet in the Ross Sea without till deltas. Grounding-line advance led to a low-profile ice sheet. Much of the newly grounded ice sheet was occupied by ice streams, but slow-moving ridges betweem ice streams may have existed, perhaps at Crary Ice Rise and elsewhere, and may have been frozen to their beds locally. The ice streams were lubricated by water-saturated till layers some meters thick, with erosion (including remobilization of older till) occurring beneath the till. Lubricating till for the ice streams was supplied by advection from upstream, by local remobilization and erosion, and by recycling of till deltas. This till was transported to the grounding line at the edge of the continental shelf, where it built till deltas and/or slumped downward to the abyssal plain. During post-Wisconsinan sea-level rise the ice was floated off the bed to leave a continuous layer of basal till several meters thick

115 across the Ross Sea. Sorted sand layers of the type reported by Kellogg et al. (1979) were left locally during grounding-line retreat where water flows were concentrated, and the till layer may have thickened locally as till deltas began to develop during any pauses in grounding-line retreat. The grounding line eventually stabilized near its present position about 5,000-10,000 years ago owing to cessation of sea-level rise and to backstresses from interactionq of the Ross Ice Shelf with its sides and with pinning points (Thomas and Bentley, 1978); deposition of modern till deltas then began. Floating of frozen-on regions of slowmoving grounded ice during grounding-line retreat may have allowed localized dropstone sedimentation during subsequent basal melting, as proposed for the Filchner-Ronne Ice Shelf by Orheim and Elverhoi (1981). These events may have occurred several times, during the latest Pliocene-Pleistocene, and possibly before. In each case, erosion beneath the grounded ice may have occurred wholly within the basal tills from earlier advances or may have cut through earlier tills to the older glaciomarine sediments beneath the Ross Sea unconformity. The Ross Sea unconformity thus may represent one or several latest Pliocene-Pleistocene erosional events, and the overlying till may have been deposited by the latest advance or may include material from several advances. The number of advances that contributed to erosion of the Ross Sea unconformity and to deposition of the overlying till may vary geographically. The unconformity observed at the base of the deforming till at UpB is the inland extension of the Ross Sea unconformity, and is still being eroded.

Test of hypothesis The hypothesis presented above makes a number of testable predictions. The hypothesis requires that the material resting on the Ross Sea unconformity is a basal till, as argued by Anderson et al. (1980). It also seems to require that the material at site J9 is a basal till or some part of a till delta that was deposited

during the Pleistocene. However, the hypothesis does not require the J9 material to contain Pleistocene fossils. This is because the large fluxes of till envisioned here should allow ~'flushing out" of younger forms deposited near J9, so that only sediments eroded upglacier beneath the grounded ice of the West Antarctic ice sheet or from deeper, older sediments in the vicinity of J9 would be observed in the till there. We thus would expect the youngest abundant fossils at J9 to be no younger than the most recent period of marine productivity in the region now occupied by the grounded West Antarctic ice sheet. Testing of our hypothesis clearly requires resolution of the existing conflicts about the age and depositional mode of sediments in the Ross Sea and beneath the Ross Ice Shelf. In addition, further geophysical and drilling studies are needed of deforming till and the till delta beneath ice stream B. We also need to develop the modeling capability to quantify this hypothesis and to use it to make further testable predictions regarding sediments in the Ross Sea.

Summary Development of ice shelves is favored by rapid flow of cold ice from outlet glaciers or ice streams into protected embayments with localized high spots. Modern ice shelves largely are restricted to the Antarctic, and the Ross and Filchner-Ronne Ice Shelf together contain most of the world's shelf ice. Sediment can be transported to an ice shelf englacially (either basally, high englacially, or rarely, supraglacially) or subglacially (by meltwater streams or a deforming bed). Ice from outlet glaciers may contain significant englacial debris, and subsequent basal melt-out will cause deposition of a dropstone diamicton. In contrast, ice supplied from ice streams is unlikely to contain significant englacial debris, and meltwater transport is likely to be small because of the small water production of polar glaciers. In the absence of a deforming basal till, an ice-stream-fed ice shelf will have slow sedimentation, probably

116

concentrated near the grounding line from melt-out of debris in a regelation layer several millimeters thick and from water transport, with essentially zero sedimentation away from the grounding line. Evidence from the UpB camp on ice stream B draining into the Ross Ice Shelf suggests that much of the ice velocity arises from deformation of a several-meter-thick, water-saturated subglacial till that is eroding an angular unconformity on subjacent sediments. Continuity arguments then led us to suggest that the till flux has caused post-Wisconsinan deposition of lobate "till deltas" tens of kilometers long and tens of meters thick at the mouth of the ice stream th at are characterized by partial decoupling between ice and till across a water film and by a small surface slope. The subsequent discovery by Shabtaie and Bentley (1987) o f " i c e plains" at the mouth of ice stream B and adjacent ice streams, lobate in form, tens of kilometres long and having a small surface slope, and the limited seismic evidence that these ice plains are underlain by unconsolidated, water-saturated sediments tens of meters thick, lends credence to our ideas. The estimated till flux from ice stream B is about 1 0 2 1 0 3 m 3 a 1 m i width of grounding line, with deposition concentrated almost exclusively in grounding-line till deltas. The till deltas probably consist of a topset region of deformed till 0(10m) thick with debris-flow foresets and bottomsets; they may be capped by a thin layer (0(0.1 m) or less) of water-sorted sediment in some areas. Observations of age and depositional environment of younger sediments in the Ross Sea are disputed, but it seems likely t hat the Ross Sea contains a glacially eroded, regional unconformity (the Ross Sea unconformity) at a depth of < 2 m to about 40 m below the sea bed t h a t is overlain by a continuous layer of latest Pliocene-Pleistocene basal till; the till is capped by a horizontally discontinuous, thin (0(0.1 m)) layer of sorted sediment and by a thin (0(0.1-1m)) layer of Holocene biogenic sediment with a varying amount of ice-rafted debris.

Our new observations of processes beneath ice stream B suggest the hypothesis that sealevel fall caused by growth of n o r t h e r n hemisphere ice sheets would have allowed the grounding line of the Ross Ice Shelf to advance to the edge of the continental shelf, aided by "conveyor-belt" recycling of till deltas at the grounding line and perhaps requiring the thickness of till deltas to allow grounding. The newly grounded ice sheet would have had a low, ice-stream profile and would have been lubricated by a several-meter-thick, watersaturated till layer that eroded subjacent sediments. Subsequent sea-level rise would have refloated the ice, leaving a several-meterthick till layer exposed at the sea floor or locally capped by a t hi nner layer of sorted sediment formed by water flow at the ice-till interface. If this hypothesis is correct the Ross Sea unconformity was formed by erosion beneath one or more latest Pliocene-Pleistocene grounded ice sheets in the Ross Sea, and the till on top of the unconformity represents the lubricating material from one or more advances; these advances probably include a Wisconsinan-maximum event. Also, the unconformity being eroded beneath the deforming tilt at UpB is the inland extension of the Ross Sea unconformity. In the absence of special pleading, this hypothesis requires t hat the material sampled at J9 is a basal till or till delta deposited during the last (Wisconsinanmaximum?) ice advance. However, owing to the high till fluxes in our model, we expect that the material at J9 is almost entirely transported from upstream beneath currently grounded ice or eroded from deeper, older sediments near J9 and may contain few or no fossils indicative of the age of the most recent reworking event. Geophysical and glaciological studies planned on the Siple Coast over the next few years by us and by cooperating investigators should shed much light on the processes of till deformation and till-delta deposition; further work on the sediments of the Ross Sea and beneat h the Ross Ice Shelf also is required to test our hypothesis.

117

Acknowledgements We thank J.B. Anderson and D.M. Harwood for helpful comments, R. Powell and A. Elverhoi for organizing a fruitful symposium, A.N. Mares for manuscript preparation, and P.B. Dombrowski and S.H. Smith for figure preparation. This work was supported in part by the U.S. National Science Foundation under grant DPP84-12404. This is contribution number 479 of the Geophysical and Polar Research Center, University of Wisconsin-Madison.

References Alley, R.B., in press. Water-pressure coupling of sliding and bed deformation: I. Water system. J. Glaciol. Alley, R.B. and Bentley, C.R., in press. Ice-core analysis on the Siple Coast of West Antarctica. Ann. Glaciol., 11. Alley, R.B., Blankenship, D.D., Bentley, C.R. and Rooney, S.T., 1986. Deformation of till beneath ice stream B, West Antarctica. Nature, 322(6074): 57-59. Alley, R.B., Blankenship, D.D., Bentley, C.R. and Rooney, S.T., 1987a. Till beneath ice stream B. 3. Till deformation: Evidence and implications. J. Geophys. Res., 92 (Bg): 8921-8929. Alley, R.B., Blankenship, D.D., Rooney, S.T. and Bentley, C.R., 1987b. Till beneath ice stream B. 4. A coupled icetill flow model. J. Geophys. Res., 92 (B9): 8931-8940. Alley, R.B., Blankenship, D.D., Rooney, S.T. and Bentley, C.R., 1987c. Continuous till deformation beneath ice sheets. Int. Assoc. Hydrol. Sci. Publ. No. 170, pp.81-91. Anderson, J.B., Kurtz, D.D., Domack, E.W. and Balshaw, K.M., 1980. Glacial and glacial marine sediments of the Antarctic continental shelf. J. Geol., 88: 399-414. Anderson, J.B., Brake, C.F. and Myers, N.C., 1984. Sedimentation on the Ross Sea continental shelf, Antarctica. Mar. Geol., 57: 295-333. Bates, R.L. and Jackson, J.A. (Editors), 1980. Glossary of Geology. Am. Geol. Inst., Falls Church, Va., 2nd ed., 749 pp. Bentley, C.R., 1971. Seismic evidence for moraine within the basal Antarctic ice sheet. In: A.P. Crary (Editor), Antarctic Snow and Ice Studies, II. (Antarctic Research Series, 16.) Am. Geophys. Union, Washington, D.C., pp.89-129. Bentley, C.R., 1984. Some aspects of the cryosphere and its role in climatic change. In: Climate Processes and Climate Sensitivity. (Geophysical Monograph 29, Maurice Ewing Vol.5). Am. Geophys. Union, Washington, D.C., pp.207-220. Bentley, C.R., 1987. Antarctic ice streams: A review. J. Geophys. Res., 92 (Bg): 8843-8858. Bentley, C.R. and Jezek, K.C., 1981. RISS, RISP, and RIGGS: Post-IGY glaciological investigations of the Ross Ice Shelf in the U.S. program. J. R. Soc. N.Z., 11(4): 355-372.

Bindschadler, R., 1983. The importance of pressurized subglacial water in separation and sliding at the glacier bed. J. Glaciol., 29(101): 3-19. Bindschadler, R.A., MacAyeal, D.R. and Stephenson, S.N., 1987a. Ice stream-ice shelf interaction in West Antarctica. In: C.J. van der Veen and J. Oerlemans (Editors), The Dynamics of the West Antarctic Ice Sheet. Reidel, Hingham, Mass., pp.161-180. Bindschadler, R.A., Stephenson, S.N., MacAyeal, D.R. and Shabtaie, S., 1987b. Ice dynamics at the mouth of ice stream B, Antarctica. J. Geophys. Res., 92 (B9): 8885-8894. Blankenship, D.D., Bentley, C.R., Rooney, S.T. and Alley, R.B., 1986. Seismic measurements reveal a saturated porous layer beneath an active Antarctic ice stream. Nature, 322(6074): 54-57. Blankenship, D.D., Bentley, C.R., Rooney, S.T. and Alley, R.B., 1987. Till beneath ice stream B. 1. Properties derived from seismic travel times. J. Geophys. Res., 92 (Bg): 8903-8911. Blankenship, D.D., Rooney, S.T., Alley, R.B. and Bentley, C.R., in press. Seismic evidence for a thin basal layer at a second location on ice stream B (abstract). Ann. Glaciol., 12: Bloom, A.L., 1978. Geomorphology: A Systematic Analysis of Late Cenozoic Landforms. Prentice-Hall, Englewood Cliffs, N.J., 510 pp. Boulton, G.S., 1979. Processes of glacier erosion on different substrata. J. Glaciol., 23(89): 15-38. Boulton, G.S. and Hindmarsh, R.C.A., 1987. Sediment deformation beneath glaciers: Rheology and geological consequences. J. Geophys. Res., 92 (B9): 9059-9082. Brady, H.T., 1983. Interpretation of sediment cores from the Ross Ice Shelf site J-9, Antarctica. Nature, 303: 510-511. Budd, W.F., Corry, M.J. and Jacka, T.H., 1982. Results from the Amery Ice Shelf Project. Ann. Glaciol., 3: 36-41. Carey, S.W. and Ahmad, N., 1961. Glacial marine sedimentation. In: Proc. Int. Symp. Arctic Geology, 1st. 2, pp.865 894. Clough, J.W. and Hansen, B.L., 1979. The Ross Ice Shelf Project. Science, 203(4379): 433-434, Denton, G.H., Hughes, T.J. and KarlAn, W., 1986. Global ice-sheet system interlocked by sea level. Quat. Res., 26(1): 3-26. Doake, C.S.M., 1976. Thermodynamics of the interaction between ice shelves and the sea. Polar Rec., 18: 37-41. Doake, C.S.M., 1985. Glaciological evidence: Antarctic Peninsula, Weddell Sea. In: Glaciers, Ice Sheets and Sea Level: Effect of a CO2-Induced Climatic Change. U.S. Dep. Energy, Comm. Glaciol., Polar Res. Board, Washington, D.C., Attachment 11, pp.197-209. Drewry, D.J., 1979. Late Wisconsin reconstruction for the Ross Sea region, Antarctica. J. Glaciol., 24(90): 231-243. Drewry, D.J., 1986. Glacial Geologic Processes. Edward Arnold, London, 276 pp. Drewry, D.J. and Cooper, A.P.R., 1981. Processes and models of Antarctic glaciomarine sedimentation. Ann. Glaciol., 2: 117-122. Drewry, D.J., Meldrum, D.T. and Jankowskil E., 1980.

11~

Radio echo and magnetic sounding of the Antarctic ice sheet, 1978 1979. Polar Rec., 20(124): 43 51. Drewry, D.J., Jordan, S.R. and Jankowski, E., 1982. Measured properties of the Antarctic ice sheet: Surface configuration, ice thickness, volume and bedrock characteristics. Ann. Glaciol., 3:83 91. Dunbar, R.B., Anderson, J.B., Domack, E.W. and Jacobs, S.S., 1985. Oceanographic influences on sedimentation along the Antarctic continental shelf. In: S.S. Jacobs (Editor), Oceanology of the Antarctic Continental Shelf. (Antarctic Research Series, 43.) Am. Geophys. Union, Washington, D.C., pp.291 312. Engelhardt, H. and Determann, J., 1987. Borehole evidence for a thick layer of basal ice in the central Ronne Ice Shelf. Nature, 327(6120): 318 319. Fillon, R.H., 1979. Comment on "Late Quaternary extent of the West Antarctic ice sheet: New evidence from Ross Sea cores". Geology, 7:518 519. Foldvik, A. and Kvinge, T., 1974. Conditional instability of sea water at the freezing point. Deep-Sea Res., 21: 160 174. Giovinetto, M.B. and Bentley, C.R., 1985. Surface balance in ice drainage systems of Antarctica. Antarct. J. U.S., 20(4): 6 13. Gow, A.J., 1967. Antarctic glaciological studies. Antarct. J. U.S., 2(4): 121 122. Gow, A.J., Epstein, S. and Sheehy, W., 1979. On the origin of stratified debris in ice cores from the bottom of the Antarctic ice sheet. J. Glaciol., 23(89): 185 192. Greischar, L.L. and Bentley, C.R., 1980. Isostatic equilibrium grounding line between the West Antarctic inland ice sheet and the Ross Ice Shelf. Nature, 283(5748): 651 654. Harwood, D.M., 1986. Diatom biostratigraphy and paleoecology with a Cenozoic history of Antarctic ice sheets. Ph.D. Thesis, Ohio State Univ., Columbus, 592 pp. Harwood, D.M., Greene, D.G. and Scherer, R.P., in press a. New age data for sediments beneath the Ross Ice Shelf at RISP site J9 from Foraminifera and siliceous microfossils. Antarct. J. U.S., 22. Harwood, D.M., Scherer, R.P., Webb, P.-N. and Greene, D.G., in press b. Middle late Miocene age of sediments beneath the Ross Ice Shelf (site J-9) and interpretation of Miocene Antarctic paleoenvironments. Mar. Micropaleontol. Hayes, D.E. and Frakes, L.A., 1975. General synthesis, Deep Sea Drilling Project. In: Initial Reports of the Deep Sea Drilling Project, Vol.28. U.S. Gov. Print. Off., Washington, D.C., pp.919-942. HjulstrSm, F., 1935. Studies of the morphological activity of rivers as illustrated by the River Fyris. Geol. Inst. Univ. Uppsala, Bull., 25: 221-527. Hjulstr6m, F., 1939. Transportation of detritus by moving water. In P.O. Trask (Editor), Recent Marine Sediments. Am. Assoc. Pet. Geol., Tulsa, Okla., pp.5 31. Hollin, J.T., 1962. On the glacial history of Antarctica. J. Glaciol., 4:173 195. Hooke, R. LeB., Wold, B. and Hagen, J.O., 1985. Subglacial hydrology and sediment transport at Bondhusbreen, southwest Norway. Geol. Soc. Am. Bull., 96(3): 388-397.

Houtz, R. and Davey, F.,I., 1973. Seismic profiles and sonobuoy measurements in Ross Sea, Antarctica. J. Geophys. Res., 78:3448 3468. Humphrey, N., 1986. Suspended-sediment discharge from Variegated Glacier during pre-surge and surge phases of motion. In: Chapman Conf. Fast Glacier Flow (Whistler Village, B.C., May 1986) (Abstr.). Humphrey, N., Raymond, C. and Harrison, W., 1986. Discharges of turbid water during mini-surges of Variegated Glacier, Alaska, USA. J. Glaciol., 32(111): 195- 207. Jacobs, S.S., Fairbanks, R.G. and Horibe, Y., 1985. Origin and evolution of water masses near the Antarctic continental margin: Evidence from t-I2180/H2160 ratios in seawater. In: S.S. Jacobs (Editor), Oceanology of the Antarctic Continental Shelf (Antarctic Research Series, 43.) Am. Geophys. Union, Washington, D.C., pp.59 85. Jeffries, M.O., Sackinger, W.M., Krouse, H.R. and Serson, H.V., 1988. Water circulation and ice accretion beneath Ward Hunt Ice Shelf (Northern Ellsmere Island, Canada), deduced from salinity and isotope analysis of ice cores. Ann. Glaciol., 10:68 72. Kamb, B. and LaChappelle, E.R., 1964. Direct observation of the mechanism of glacier sliding over bedrock. J. Glaciol., 5: 159-172. Karl, H.A., Reimnitz, E. and Edwards, B.D., 1987. Extent and nature of Ross Sea Unconformity in western Ross Sea, Antarctica. In: A.K. Cooper and F.J. Davey (Editors), The Antarctic Continental Margin: Geology and Geophysics of the Western Ross Sea. (Earth Science Series, 5B.) Circum-Pac. Counc. Energy Miner. Resour., Houston, pp.77 92. Kellogg, D.E. and Kellogg, T.B., 1986. Diatom biostratigraphy of sediment cores from beneath the Ross Ice Shelf. Micropaleontology, 32(1): 74-94. Kellogg, T.B. and Kellogg, D.E., 1981. Pleistocene sediments beneath the Ross Ice Shelf. Nature, 293:130 133. Kellogg, T.B. and Kellogg, D.E., 1983. Reply to "Interpretation of sediment cores from the Ross Ice Shelf site J-9, Antarctica", by H.T. Brady, Nature, 303: 510 511. Nature, 303:511 -513. Kellogg, T.B., Truesdale, R.S. and Osterman, L.F., 1979. Late Quaternary extent of the West Antarctic ice sheet: New evidence from Ross Sea cores. Geology, 7: 249-253. Korotkevich, Y.S., Savatyugin, L.M. and Morev, V.A., 1978. Drilling through the ice shelf in the vicinity of Novolazarevskaya Station. Soy. Antarc. Exped., Inf. Bull., 98: 49-52. Lange, M.A. and MacAyeal, D.R., 1986. Numerical models of the Filchner-Ronne ice shelf: An assessment of reinterpreted ice thickness distributions. J. Geophys. Res., 91 (B10): 10,457 10,462. Leopold, L.B. and Maddock, T., 1953. Hydraulic geometry of stream channels and some physiographic implications. U.S. Geol. Surv., Prof. Pap., 252:57 pp. Lewis, E.L., 1985. The "ice pump", a mechanism for iceshelf melting. In: Glaciers, Ice Sheets, and Sea Level: Effect of a CO2-Induced Climatic Change. U.S. Dep. Energy, Comm. Glaciol., Polar Res. Board, Washington, D.C., A t t a c h m e n t 20, pp.275-278. Lingle, C.S. and Brown, T.J., 1987. A subglacial aquifer bed

119 model and water pressure-dependent basal sliding relationship for a West Antarctic ice stream. In: C.J. van der Veen and J. Oerlemans (Editors), The Dynamics of the West Antarctic Ice Sheet. Reidel, Hingham, Mass., pp.249-285. MacAyeal, D.R., 1985. Tidal rectification below the Ross Ice Shelf, Antarctica. In: S.S. Jacobs (Editor), Oceanology of the Antarctic Continental Shelf. (Antarctic Research Series, 43.) Am. Geophys. Union, Washington, D.C., pp.109-132. MacAyeal, D.R. and Thomas, R.H., 1979. Ross Ice Shelf temperatures support a history of ice-shelf thickening. Nature, 282(5740): 703-705. Meier, M.F., in press. Relation between water input, basal water pressure, and sliding of Columbia Glacier. Ann. Glaciol., 12: Meier, M.F., Rasmussen, L.A., Krimmel, R.M., Olsen, R.W. and Frank, D., 1985. Photogrammetric determination of surface altitude, terminus position, and ice velocity of Columbia Glacier. U.S. Geol. Surv., Prof. Pap., 1258-F: 41 pp. Mercer, J.H., 1978. West Antarctic ice sheet and C O 2 greenhouse effect: A threat of disaster. Nature, 271(5643): 321-325. Mercer, J.H. and Sutter, J.F., 1982. Late Miocene-Earliest Pliocene glaciation in southern Argentina: Implications for global ice-sheet history. Palaeogeogr., Palaeoclimatol., Palaeoecol., 38: 185-206. Mitchum Jr., R.M., Vail, P.R. and Sangree, J.B., 1977. Seismic stratigraphy and global changes of sea level, part 6: Stratigraphic interpretation of seismic reflection patterns in depositional sequences. In: C.E. Payton (Editor), Seismic Stratigraphy -- Applications to Hydrocarbon Exploration. Am. Assoc. Pet. Geol. Mere., 26: 117-133. Morgan, V.I., 1972. Oxygen isotope evidence for bottom freezing on the Amery Ice Shelf. Nature, 238(5364): 393-394. Neal, C.S., 1979. The dynamics of the Ross Ice Shelf revealed by radio echo-sounding. J. Glaciol., 24(90): 295-307. Orheim, O. and Elverhoi, A., 1981. Model for submarine glacial deposition. Ann. Glaciol., 2: 123-128. Paren, J.G. and Cooper, S., in press. Thermal regime of George VI Ice Shelf. Ann. Glaciol., 11: Pillsbury, R.D. and Jacobs, S.S., 1985. Preliminary observations from long-term current meter moorings near the Ross Ice Shelf, Antarctica. In: S.S. Jacobs (Editor), Oceanology of the Antarctic Continental Shelf (Antarctic Research Series, 43.) Am. Geophys. Union, Washington, D.C., pp.87-I07. Powell, R.D., 1981. A model for sedimentation by tidewater glaciers. Ann. Glaciol., 2: 129-134. Powell, R.D., 1984. Glacimarine processes and inductive lithofacies modelling of ice shelf and tidewater glacier sediments based on Quaternary examples. Mar. Geol., 57: 1-52. Prior, D.B., Wiseman Jr., J.W. and Bryant, W.R., 1981. Submarine chutes on the slopes of fjord deltas. Nature, 290: 326-328.

Raiswell, R. and Tan, M.M., 1985. Diagenesis of sediments beneath the Ross Ice Shelf and their sedimentary history. Nature, 315: 483-485. Rasmussen, L.A. and Meier, M.F., 1982. Continuity equation model of the predicted drastic retreat of Columbia Glacier, Alaska. U.S. Geol. Surv., Prof. Pap., 1258-A: 23 pp. Robin, G. de Q., 1958. Glaciology Ill.Seismic shooting and related investigations. In: Norwegian-British-Swedish Antarctic Expedition, 1949-1952. Sci. Results, 5. Norsk Polarinst., Oslo. Robin, G. de Q., 1977. Ice cores and climatic changes. Philos. Trans. R. Soc. London, Ser. B, 280: 143-168. Robin, G. de Q., 1979. Formation, flow, and disintegration of ice shelves. J. Glaciol., 24: 259-271. Rooney, S.T., Blankenship, D.D., Alley, R.B. and Bentley, C.R., 1987. Till beneath ice stream B. 2. Structure and continuity. J. Geophys. Res., 92 (B9): 8913-8920. Rooney, S.T., Blankenship, D.D., Alley, R.B. and Bentley, C.R., in press a. Seismic reflection profiling of a sediment-filled graben beneath ice stream B, West Antarctica. In: Proc. Int. Syrup. Antarc. Earth Sci., 5th. Rooney, S.T., Blankenship, D.D., Alley, R.B. and Bentley, C.R., in press b. Seismic reflection profiling of a widespread tillbeneath ice stream B, West Antarctica. Ann. Glaciol., 11: (Abstr.}. Rose, K.E., 1979. Characteristics of ice flow in Marie Byrd Land, Antarctica. J. Glaciol., 24(90): 63-74. R5thlisberger, H., 1968. Erosive processes which are likely to accentuate or reduce the bottom relief of valley glaciers. Int. Assoc. Sci. Hydrol. Publ. No. 79, pp.87-97. Sangree, J.B. and Widmier, J.M., 1977. Seismic stratigraphy and global changes of sea level, part 9: Seismic interpretation of clastic depositional facies. In: C.E. Payton (Editor), Seismic Stratigraphy -- Applications to Hydrocarbon Exploration. Am. Assoc. Pet. Geol. Mere., 26: 165-184. Savage, M.L. and Ciesielski,P.F., 1983. A revised history of glacial sedimentation in the Ross Sea region. In: R.L. Oliver, P.R. James and J.B. Jago (Editors), Antarctic Earth Science. Aust. Acad. Sci., Adelaide, pp.555-559. Shabtaie, S. and Bentley, C.R., 1979. Investigation of bottom mass-balance rates by electrical resistivity soundings of the Ross Ice Shelf, Antarctica. J. Glaciol., 24(90): 331-343. Shabtaie, S. and Bentley, C.R., 1987. West Antarctic ice streams draining into the Ross Ice Shelf: Configuration and mass balance. J. Geophys. Res., 92 (Bg): 1311-1336. Singer, J.K. and Anderson, J.B., 1984. Use of total grainsize distributions to define bed erosion and transport for poorly sorted sediment undergoing simulated bioturbation. Mar. Geol., 57: 335-359. Stuiver, M., Denton, G.H., Hughes, T.J. and Fastook, J.L., 1981. History of the marine ice sheet in West Antarctica during the last glaciation: A working hypothesis. In: G.H. Denton and T.J. Hughes (Editors), The Last Great Ice Sheets. Wiley, New York, pp.319-436. Sugden, D.E. and John, B.S., 1976. Glaciers and Landscape. Edward Arnold, London, 376 pp.

120

Thomas, R.H., 1973a. The creep of ice shelves: Theory. ,J. Glaciol., 12(64): 45-53. Thomas, R.H., 1973b. The creep of ice shelves: Interpretation of observed behaviour. J. Glaciol., 12(64): 55 70. Thomas, R.H., 1976. Thickening of the Ross Ice Shelf and equilibrium state of the West Antarctic ice sheet. Nature, 259: 180-183. Thomas, R.H., 1979. West Antarctic ice sheet: Present-day thinning and Holocene retreat of the margins. Science, 205(4412): 1257-1258. Thomas, R.H. and Bentley, C.R., 1978. A model for Holocene retreat of the West Antarctic ice sheet. Quat. Res., 10: 150--170. Thomas, R.H. and MacAyeal, D.R., 1982. Derived characteristics of the Ross Ice Shelf, Antarctica. J. Glaciol., 28(100): 397412. Thomas, R.H., Stephenson, S.N., Bindschadler, R.A., Shabtale, S. and Bentley, C.R., in press. Thinning and grounding line retreat on the Ross Ice S h e l l Ann. Glaciol., 11: Thyssen, F., in press. Special aspects of the central part of the Filchner/Ronne Ice Shelf. Ann. Glaciol., 11: Truswell, E.M. and Drewry, D.J., 1984. Distribution and provenance of recycled palynomorphs in surficial sediments of the Ross Sea, Antarctica. Mar. Geol., 59: 187 214. Walters, R.A., Josberger, E.G. and Driedger, C.L., 1986. Subglacial discharge from a tidewater glacier. In: Chapman Conf. Fast Glacier Flow (Whistler Village, B.C., May 1986). (Abstr.). Webb, P.N., Ronan Jr., T.E., Lipps, J.H. and DeLaca, T.E.,

1979. Miocene glaciomarine sediments from beneath the southern Ross Ice Shelf, Antarctica. Science, 203(4379): 435 437. Weertman, J., 1957. Deformation of floating ice shelves. J. Glaciol., 3:38 42. Weertman, J., 1964. Glacier sliding. J. Glaciot., 5(39): 287- 303. Weertman, J., 1968. Diffusion law for the dispersion of hard particles in an ice matrix that undergoes simple shear deformation. J. Glaciol., 7(50): 161 165. Weertman, J., 1972. General theory of water flow at the base of a glacier or ice sheet. Rev. Geophys. Space Phys., 10(1): 287 333. Weertman, J., 1973. Can a water-filled crevasse reach the bottom surface of a glacier? Int. Assoc. Sci. Hydrol. Publ., No. 95, pp.139 145. Weertman, J., 1974. Stability of the junction of an ice sheet and an ice s h e l l J. Glaciol., 13(67): 3-11. Weertman, J. and Birchfield, G.E., 1982. Subglacial water flow under ice streams and West Antarctic ice-sheet stability. Ann. Glaciol., 3: 316-320. Weertman, J. and Birchfield, G.E., 1983. Stability of sheet water flow under a glacier. J. Glaciol., 29(103): 374- 382. Whillans, I.M., 1976. Radio-echo layers and the recent stability of the West Antarctic ice sheet. Nature, 264(5582): 152-155. Whillans, I.M., Bolzan, J. and Shabtaie, S., 1987. Velocity of ice streams B and C, Antarctica. J. Geophys. Res., 92 (B9): 8895 8902. Zotikov, I.A., Zagorodnov, V.S. and Raikovsky, J.V., 1979. Sea ice on bottom of the Ross Ice S h e l l Antarct. J. U.S., 14:65 66.