ICE-SHELF STABILITY - Judith Currycurry.eas.gatech.edu/.../Chapter10/Ency_Oceans/Ice-shelf_Stability.pdf · future. What are Ice Shelves? Physical and geographical setting An ice

ICE-SHELF STABILITY

C. S. M. Doake, British Antarctic Survey,Cambridge, UK

Copyright ^ 2001 Academic Press

doi:10.1006/rwos.2001.0005

Introduction

Ice shelves are Soating ice sheets and are foundmainly around the Antarctic continent (Figure 1).They can range in size up to 500 000 km2 and inthickness up to 2000m. Most are fed by ice streamsand outlet glaciers, but some are formed by icebergswelded together by sea ice and surface accumula-tion. They exist in embayments where the shape ofthe bay and the presence of ice rises, or pinningpoints, plays an important role in their stability. Iceshelves lose mass by basal melting, which modiResthe properties of the underlying water mass andeventually inSuences the circulation of the globalocean, and by calving. Icebergs, sometimes morethan 100 km in length, break off intermittently fromthe continent and drift to lower latitudes, usuallybreaking up and melting by the time they reach theAntarctic Convergence. The lowest latitude that iceshelves can exist is determined by the mean annualair temperature. In the Antarctic Peninsula a criticalisotherm of about !53C seems to represent thelimit of viability. In the last 40 years or so, severalice shelves have disintegrated in response to ameasured atmospheric warming trend in the westernand northern part of the Antarctic Peninsula. How-ever, this warming trend is not expected to affectthe stability of the larger ice shelves furthersouth such as Filchner}Ronne or Ross, in the nearfuture.

What are Ice Shelves?

Physical and geographical setting

An ice shelf is a Soating ice sheet, attached to landwhere ice is grounded along the coastline. Nourish-ed mainly by glaciers and ice streams Sowing off theland, ice shelves are distinct from sea ice, which isformed by freezing of sea water. Most ice shelvesare found in Antarctica where the largest can coverareas of 500 000 km2 (e.g. Ross Ice Shelf andFilchner}Ronne Ice Shelf). Thicknesses vary fromnearly 2000 m, around for example parts of thegrounding line of Ronne Ice Shelf, to about 100 mat the seaward edge known as the ice front.

Typically, ice shelves exist in embayments, con-strained by side walls until they diverge too muchfor the ice to remain in contact. Thus the geometryof the coastline is important for determining bothwhere an ice shelf will exist and the position of theice front. There are often localized grounding pointson sea bed shoals, forming ice rises and ice rumples,both in the interior of the ice shelf and along the icefront. These pinning points provide restraint andcause the ice shelf to be thicker than if it were notpinned. Ice Sows from the land to the ice front.Input velocities range from near zero at a shearmargin (e.g., with a land boundary), to several hun-dred meters per year at the grounding line where icestreams and outlet glaciers enter. At the ice frontvelocities can reach up to several kilometers peryear. A characteristic of ice shelves is that the (hori-zontal) velocity is almost the same at all depths,whereas in glaciers and grounded ice sheets thevelocity decreases with depth (Figure 2).

In the Antarctic, most ice shelves have net surfaceaccumulation although there may be intensive sum-mer melt which Soods the surface. The basal regimeis controlled by the subice circulation. Basal meltingis often high near both the grounding line and theice front. Marine ice can accumulate in the inter-mediate areas, where water at its in situ freezingpoint upwells and produces frazil ice crystals. Sur-face temperatures will be near the mean annual airtemperature whereas the basal temperature will beat the freezing point of the water. Therefore thecoldest ice is normally in the upper layers.

Ice shelves normally form where ice Sowssmoothly off the land as ice streams or outlet gla-ciers. In some areas, however, ice breaks off at thecoastline and reforms as icebergs welded together byfrozen sea ice and surface accumulation. The pro-cesses of formation and decay are likely to operateunder very different conditions. An ice shelf is un-likely to reform under the same climatic conditionswhich caused it to decay. Complete collapse isa catastrophic process and rebuilding requires amajor change in the controlling parameters. How-ever, there is evidence of cyclicity on periods of afew hundred years in some areas.

Collapse of the northernmost section of LarsenIce Shelf within a few days in January 1995 indi-cates that, after retreat beyond a critical limit, iceshelves can disintegrate rapidly. The breakup historyof two northern sections of Larsen Ice Shelf (LarsenA and Larsen B) between 1986 and 1997 has been

1282 ICE-SHELF STABILITY

VVCrGayathrirScanrPadmarRWOS 0005

Figure 1 Distribution of ice shelves around the coast of Antarctica.

used to determine a stability criterion for ice shelves.Analysis of various ice-shelf conRgurations revealscharacteristic patterns in the strain-rates near the icefront which have been used to describe the stabilityof the ice shelf.

Why Are Ice Shelves Important?

Ice shelves are one of the most active parts of the icesheet system. They interact with the inland ice sheet,glaciers and ice streams which Sow into them, andwith the sea into which they eventually melt, eitherdirectly or as icebergs. A typical Antarctic ice shelfwill steadily advance until its ice front undergoesperiodic calving, generating icebergs. Calving canoccur over a wide range of time and spatial scales.Most ice fronts will experience a quasi-continuous‘nibbling’ away of the order of tens to hundreds of

meters per year, where ice cliffs collapse to formbergy bits and brash ice. The largest icebergs may bemore than 100 km in size, but will calve only atintervals of 50 years or more. Thus when chartingthe size and behavior of an ice shelf it can bedifRcult to separate and identify stable changes fromunstable ones.

Mass balance calculations show that calving oficebergs is the largest factor in the attrition of theAntarctic Ice Sheet. Estimates based on both shipand satellite data suggest that iceberg calving is onlyslightly less than the total annual accumulation. Iceshelf melting at the base is the other principal ele-ment in the attrition of the ice sheet, with approx-imately 80% of all ice shelf melting occurring atdistances greater than 100 km from the ice front.Although most of the mass lost in Antarctica isthrough ice shelves, there is still uncertainty about

ICE-SHELF STABILITY 1283


Figure 2 Cartoon of ice shelf in bay with ice streams.

the role ice shelves play in regulating Sow off theland, which is the important component for sea-level changes. Because ice shelves are already Soat-ing, they do not affect sea level when they breakup.

Ice shelves are sensitive indicators of climatechange. Those around the Antarctic Peninsula haveshown a pattern of gradual retreat since about1950, associated with a regional atmospheric warm-ing and increased summer surface melt. The effectsof climate change on the mass balance of the Ant-arctic Ice Sheet and hence global sea level are unclear.

There have been no noticeable changes in theinland ice sheet from the collapse of the AntarcticPeninsula ice shelves. This is because most of themargins with the inland ice sheet form a sharptransition zone, making the ice sheet dynamics inde-pendent of the state of the ice shelf. This is not thecase with a grounding line on an ice stream wherethere is a smooth transition zone, but generally theimportance of the local stresses there diminishesrapidly upstream.

An important role of ice shelves in the climatesystem is that subice-shelf freezing and melting pro-cesses inSuence the formation of Antarctic BottomWater, which in part helps to drive the oceanicthermohaline circulation (see Sub Ice-shelf Circula-tion and Processes).

How Have Ice Shelves Changedin the Past?

As the polar ice sheets have waxed and wanedduring ice age cycles, ice shelves have also grownand retreated. During the last 20}25 million years,the Antarctic Ice Sheet has probably been grounded

out to the edge of the continental shelf many times.It is unlikely that any substantial ice shelves couldhave existed then, because of unsuitable coastlinegeometry and lack of pinning points, although theextent of the sea ice may have been double thepresent day area. At the last glacial maximum,about 20 000 years ago, grounded ice extended outto the edge of the continental shelf in places, forexample in Prydz Bay, but not in others such as theRoss Sea. When the ice sheets began to retreat, someof the large ice shelves seen today probably formedby the thinning and eventual Sotation of the former-ly grounded ice sheets.

Large ice shelves also existed in the Arctic duringthe Pleistocene. Abundant geologic evidence showsthat marine Northern Hemisphere ice sheetsdisappeared catastrophically during the climatictransition to the current interglacial (warm period).Only a few small ice shelves and tidewater glaciersexist there today, in places like Greenland, Svalbard,Ellesmere Island and Alaska.

There have been periodic changes in Antarcticice-shelf grounding lines and ice fronts during theHolocene. The main deglaciation on the westernside of the Antarctic Peninsula which started morethan 11 000 years ago was initially very rapid acrossa wide continental shelf. By 6000 years ago, the icesheet had cleared the inner shelf, whereas in theRoss Sea retreat of ice shelves ceased about the sametime. Retreat after the last glacial maximum wasfollowed by a readvance of the grounding line dur-ing the climate warming between 7000 and 4000years ago. Open marine deposition on the continen-tal shelf is restricted to the last 4000 years or so.Short-term cycles (every few hundred years) and



Figure 3 Cartoon showing retreat of Wordie ice front between 1966 and 1989.

longer-term events (approximately 2500 year cycles)have been detected in marine sediment cores thatare likely related to global climate Suctuations.

More recently, ice-shelf disintegration aroundthe Antarctic Peninsula has been associated witha regional atmospheric warming which has beenoccurring since about 1950. Ice front retreat ofmarginal ice shelves elsewhere in Antarctica (e.g.,Cook Ice Shelf, West Ice Shelf) is probably alsorelated to atmospheric warming. There is littlesign of any signiRcant impact on the groundedice.

Where Are Ice Shelves DisintegratingNow? Two Case Studies

The detailed history of the breakup of two iceshelves, Wordie and Larsen, illustrate some of thecritical features of ice-shelf decay.

Wordie Ice Shelf

Retreat of Wordie Ice Shelf on the west coast of theAntarctic Peninsula has been documented using highresolution visible satellite images taken since 1974and the position of the ice front in 1966 mappedfrom aerial photography. First seen in 1936, the ice

front has Suctuated in position but with a sustainedretreat starting around 1966 (Figure 3). The iceshelf area has decreased from about 2000km2 in1966 to about 700 km2 in 1989. However, deRningthe position of the ice front can be very uncertain,with large blocks calving off to form icebergs beingdifRcult to classify as being either attached to, orseparated from, the ice shelf.

Ice front retreat until about 1979 occurred mainlyby transverse rifting along the ice front, creatingicebergs up to 10 km by 1 km. The western area wasthe Rrst to be lost, between 1966 and 1974. Resultsfrom airborne radio-echo sounding between 1966and 1970 suggested that the ice shelf could bedivided into a crevassed eastern part and a riftedwestern part where brine could well-up and inRl-trate the ice at sea level. By 1974 there were manytransverse rifts south of Napier Ice Rise and by1979 longitudinal rifting was predominant. Manylongitudinal rifts had formed upstream of several icerises by 1979, some following preexisting Sowlinefeatures. A critical factor in the break-up was thedecoupling of Buffer Ice Rise; by 1986 ice wasstreaming past it apparently unhindered, in contrastto the compressive upstream folding seen in earlierimages. Between 1988 and 1989 the central part of



the ice shelf, consisting of broken ice in the lee ofMount Balfour, was lost, exposing the coastline andeffectively dividing the ice shelf in two. By 1989,longitudinal rifting has penetrated to the groundingline north of Mount Balfour. Further north, a grow-ing shear zone marked the boundary between fast-Sowing ice from the north Forster Ice Piedmont andslower moving ice from Hariot Glacier. There hasbeen no discernible change in the grounding lineposition.

The major ice rises have played several roles incontrolling ice-shelf behavior. When embedded inthe ice shelf, they created broken wakes down-stream, and zones of compression upstream whichhelped to stabilize the ice shelf. During ice frontretreat, they temporarily pinned the local ice frontposition and also acted as nucleating points forrifting which quickly stretched upstream, suggestingthat a critical fracture criterion had been exceeded.At this stage, an ice rise, instead of protecting theice shelf against decay, aided its destruction byacting as an indenting wedge.

Breakup was probably triggered by a climaticwarming which increased ablation and the amountof melt water. Laboratory experiments show thatthe fracture toughness of ice is reduced at highertemperatures and possibly by the presence of water.Instead of refreezing in the upper layers of Rrn, freewater could percolate down into crevasses and, byincreasing the pressure at the bottom, allow them togrow into rifts or possibly to join up with basalcrevasses. Processes like these would increase theproduction rate of blocks above that required fora ‘steady state’ ice front position. The blocks willdrift away as icebergs if conditions, such as baygeometry and lack of sea ice, are favorable. Thus,ice front retreat would be, inter alia, a sensitivefunction of mean annual air temperature.

Some ice shelves, such as Brunt, are formed fromblocks that break off at the coast line and, unable toSoat away, are ‘glued’ together by sea ice and snow-fall. These heterogeneous ice shelves contrast withthose, such as Ronne, where glaciers or ice streamsSow unbroken across the grounding line to forma more homogeneous type of ice shelf. Before 1989,Wordie Ice Shelf consisted of a mixture of bothtypes, the main tongues being derived from glacierinputs, while the central portion was formed fromblocks breaking off at the grounding line and at thesides of the main tongues. The western rifted areathat broke away between 1966 and 1974 was de-scribed in early 1967 as ‘snowed-under icebergs’and it was the heterogeneous central part that brokeback to the coastline around Mount Balfour in1988/89. Increased ablation would not only en-

hance rifting along lines of weakness but would alsoloosen the ‘glue’ that held the blocks together.

Larsen Ice Shelf

The most northerly ice shelf in the Antarctic, LarsenIce Shelf, extends in a ribbon down the east coast ofthe Antarctic Peninsula from James Ross Island tothe Ronne Ice Shelf. It consists of several distinct iceshelves, separated by headlands. The ice shelf inPrince Gustav Channel, between James Ross Islandand the mainland, separated from the main part ofLarsen Ice Shelf in the late 1950s and Rnally disap-peared by 1995.

The section between Sobral Peninsula and Robert-son Island, known as Larsen A, underwent a cata-strophic collapse at the end of January 1995 when itdisintegrated into a tongue of small icebergs, bergybits and brash ice. Previously the ice front had beenretreating for a number of years, but in a morecontrolled fashion by iceberg calving. Before theRnal breakup, the surface that had once been Satand smooth had become undulating, suggesting thatrifting completely through the ice had occurred.Collapse during a period of intense north-westerlywinds and high temperatures was probably aided bya lack of sea ice, allowing ocean swell to penetrateand add its power to increasing the disintegrationprocesses. The speed of the collapse and the smallsize of the fragments of ice (the largest icebergs wereless than 1 km in size) implicate fracture as thedominant process in the disintegration (Figure 4).

The ice front of the ice shelf known as Larsen B,between Robertson Island and Jason Peninsula,steadily advanced for about 6 km from 1975 until1992. Small icebergs broke away from the heavilyrifted zone south of Robertson Island after July1992 and a major rift about 25 km in length hadopened up by then. This rift formed the calvingfront when an iceberg covering 1720 km2 and small-er pieces corresponding to a former ice shelf area of550 km2 broke away between 25 and 30 January1995, coincident with the disintegration of LarsenA. The ice front has retreated continuously by a fewkilometers per year since then (to 1999) and the iceshelf is considered to be under threat of disappear-ing completely.

Why Do Ice Shelves Break-up?

Calving and Fracture

Calving is one of the most obvious processes in-volved in ice shelf breakup. Although widespread,occurring on grounded and Soating glaciers,including ice shelves, as well as glaciers ending in



Figure 4 Photo of the Larsen breakup. Picture taken approximately 10 km north of Robertson Island, looking west north-west. Inthe far background the mountains of the Antarctic Peninsula can be seen.

freshwater lakes, it is not well understood. Thebasic physics, tensile propagation of fractures, maybe the same in all cases but the predictability ofcalving may be quite different, depending on thestress Reld, the basal boundary conditions, theamount of surface water, etc. Fracture of ice hasbeen studied mainly in laboratories and applyingthese results to the conditions experienced in nat-urally occurring ice masses requires extrapolationswhich may not be valid. Questions such as how iscrevasse propagation inSuenced by the ice shelf ge-ometry and by the physical properties of the icesuch as inhomogeneity, crystal anisotropy, temper-ature and presence of water, need to be answeredbefore realistic models of the calving process can bedeveloped. The engineering concepts of fracture mech-anics and fracture toughness promise a way forward.

Fracture of ice is a critical process in ice shelfdynamics. Mathematical models of ice shelf behav-ior based on continuum mechanics, which treat iceas a nonlinear viscous Suid will have to incorporatefracture mechanics to describe iceberg calving andhow ice rises can initiate fracture both upstream anddownstream. High stresses generated at shear mar-gins around ice rises, or at the ‘corner points’ of theice shelf where the two ends of the ice front meetland, can exceed a critical stress and initiate crevass-ing. These crevasses will propagate under a favor-

able stress regime, and if there is sufRcient surfacewater may rift through the complete thickness.Transverse crevasses parallel to the ice front act assites for the initiation of rifting and form lineswhere calving may eventually occur.

Calving is a very efRcient method of getting rid ofice from an ice shelf } once an iceberg has formed, itcan drift away within a few days. Sometimes, how-ever, an iceberg will ground on a sea bed shoal,perhaps for several years. Once an Antarctic iceberghas drifted north of the Antarctic Convergence, itusually breaks-up very quickly in the warmer waters.

Climate Warming

Circumstantial evidence links the retreat of iceshelves around the Antarctic Peninsula to a warmingtrend in atmospheric temperatures. Observations atmeteorological stations in the region show a rise ofabout 2.53C in 50 years. Ice shelves appear to existup to a climatic limit, taken to be the mean annual!53C isotherm, which represents the thermal limitof ice-shelf viability. The steady southward migra-tion of this isotherm has coincided with the patternof ice-shelf disintegration.

There are too few sea temperature measurementsto show if the observed atmospheric warming hasresulted in warmer waters, and thus increased basal



melting. Although basal melting may have in-creased, the indications are that it is surface pro-cesses that have played the dominant role in causingbreakup. The mechanism for connecting climatewarming with ice-shelf retreat is not fully under-stood, but increased summer melting producing sub-stantial amounts of water obviously plays a part inenhancing fracture processes, leading to calving.

The retreat of ice shelves around the AntarcticPeninsula that has occurred in the last half of thetwentieth century has raised questions aboutwhether or not this reSects global warming orwhether it is only a regional phenomenon. Thewarming trend seems to be localized to the Antarc-tic Peninsula region and there is no signiRcantcorrelation with temperature changes in the rest ofAntarctica.

Stress Patterns

Numerical models of ice shelves have been used toexamine their behavior and stability criteria. Thestrain-rate Reld can be speciRed by the principalvalues (�� 1 and �� 2 where �� 1'�� 2) and the direction ofthe principal axes. A simple representation is givenby the trajectories, which are a set of orthogonalcurves whose directions at any point are the direc-tions of the principal axes. Analyses of the strain-rate trajectories for Filchner Ronne Ice Shelf and fordifferent ice shelf conRgurations of Larsen Ice Shelfshow characteristic patterns of a ‘compressive arch’and of isotropic points (Figure 5). The ‘compressivearch’ is seen in the pattern formed by the smallestprincipal component (�� 2) of the strain-rate trajecto-ries. Seaward of the arch both principal strain-ratecomponents are extensive, whereas inland of thearch the �� 2 component is compressive. The archextends from the two ends of the ice front across thewhole width of the ice shelf. It is a generic feature ofice shelves studied so far and structurally stable tosmall perturbations. It is probably related to thegeometry of the ice shelf bay. A critical arch, con-sisting entirely of compressive trajectories, appearsto correspond to a criterion for stability. If the icefront breaks back through the arch then an irrevers-ible retreat occurs, possibly catastrophically, to an-other stable conRguration. The exact location of thecritical arch cannot yet be determined a priori, but it isprobably close to the compressive arch delineated bythe transition from extension to compression for �� 2.

Another pattern seen in the strain-rate trajectoriesis that of isotropic points. They are indicators ofgeneric features in the surface Sow Reld which arestable to small perturbations in the Sow. This meansthat their existence should not be sensitive either toreasonable errors in data used in the model or to

simpliRcations in the model itself. They act as (per-manent) markers in a complicated (tensor) strain-rate Reld and are often located close to points wherethe two principal strain-rates are equal or where thevelocity Reld is stationary (usually either a max-imum or a saddle point). Isotropic points are classi-Red by a number of properties, but only twocategories can be reliably distinguished observation-ally, by the way the trajectory of either of theprincipal strain-rates varies around a path enclosingthe isotropic point. In one case the trajectory variesin a prograde sense and the isotropic point is calleda ‘monstar’, whereas in the other case the trajectoryvaries in a retrograde sense and the isotropic pointis called a ‘star’ (Figure 5).

The two categories of isotropic points can beidentiRed in the model strain-rate trajectories, ‘stars’occurring near input glaciers and ‘monstars’ near icefronts if they are in a stable conRguration. Icebergscalving off Filchner Ice Shelf in 1986 moved theposition of the ‘monstar’ up to the newly formed icefront, whereas on Ronne Ice Shelf the ‘monstar’ isabout 50 km inland of the ice front. This suggeststhat the existence of a monstar can be used asa ‘weak’ indicator of a stable ice front. Calving canremove a monstar even if the subsequent ice frontposition is not necessarily an unstable one and mayreadvance.

How Vulnerable are the LargeIce Shelves and How Stableare Grounding Lines?

The two largest ice shelves (Ross and Filchner}Ronne) are too far south to be attacked by theatmospheric warming that is predicted for thetwenty-Rrst century. Basal melting will increase ifwarmer water intrudes onto the continental shelf,but the effects of a warming trend could be counter-intuitive. If the warming was sufRcient to reduce therate of sea ice formation in the Weddell Sea, thenthe production of High Salinity Shelf Water (HSSW)would reduce as well. This would affect the subiceshelf circulation, replacing the relatively warmHSSW under the Filchner}Ronne Ice Shelf witha colder Ice Shelf Water which would reduce themelting and thus thicken the ice shelf. Furtherclimate warming would restore warmer watersand eventually thin the ice shelf by increasing themelting rates, possibly to values of around15 m year�1 as seen under Pine Island Glacier. Thelikelihood of this is discussed in Sub Ice-shelfCirculation and Processes.

The supplementary question is what effect, if any,would the collapse of the ice shelves have on the



65°S

66°S 62°W

66°S

60°W

Scale

20 0Kilometers

Figure 5 Modeled strain-rate trajectories on Larsen Ice Shelf superimposed on a Radarsat image taken on 21 March 1997. Thetrajectory of the smallest principal strain-rate (�� 2) is negative (compressing flow, blue) over nearly all the Larsen Ice Shelf in its post-iceberg calving configuration (January 1995) but is positive (red) in the region where the iceberg calved. The pattern shows isotropicpoints (‘stars,’ green) near where glaciers enter the ice shelf and, for the ‘iceberg’ area a ‘monstar’ (yellow cross) near the ice front.‘Stars’ are seen for all the different ice front configurations, indicating that no fundamental change is occurring near the groundingline due to retreat of the ice front. A ‘monstar’ is only seen for Larsen B before the iceberg calved in 1995. The disappearance of theisotropic point is perhaps a ‘weak’ indication that the iceberg calving event was greater than expected for normal calving froma stable ice shelf and suggests further retreat may be expected.

ice sheet, especially the West Antarctic Ice Sheet(WAIS). It has been a tenet of the latter part of thetwentieth century that the WAIS, known as a mar-ine ice sheet because much of its bed is below sealevel, is potentially unstable and may undergo disin-tegration if the fringing ice shelves were to disap-pear. However, the theoretical foundations of thisbelief are shaky and more careful consideration ofthe relevant dynamics suggests that the WAIS is nomore vulnerable than any other part of the ice sheet.

The problem lies in how to model the transitionzone between ice sheet and ice shelf. If the transitionis sharp, where the basal traction varies over lengthsof the order of the ice thickness, then it can be

considered as a passive boundary layer and does notaffect the mechanics of the sheet or shelf to Rrstorder. Mass conservation must be respected butthere is no need to impose further constraints. Re-quiring the ice sheet to have the same thickness asthe ice shelf at the grounding line permits only twostable grounding line positions for a bed geometrywhich deepens inland. Depending on the depth ofthe bed below sea level, the ice sheet will eithershrink in size until it disappears (or, if the bedshallows again, the grounding line is Rxed on a bednear sea level), or grow to the edge of the continentalshelf. The lack of intermediate stable grounding linepositions in this kind of model has been used to



support the idea of instability of marine ice sheets.However, permitting a jump in ice thickness at thetransition zone means the ice sheet system can be inneutral equilibrium, with an inRnite number ofsteady-state proRles.

Another kind of transition zone is a smooth one,where the basal traction varies gradually, forexample, along an ice stream. The equilibrium dy-namics have not been worked out for a full three-dimensional Sow, but it seems that the presence ofice streams destroys the neutral equilibrium andhelps to stabilize marine ice sheets. This emphasizesthe importance of understanding the dynamics of icestreams and their role in the marine ice sheet system.

Attempts to include moving grounding lines inwhole ice sheet models suffer from incomplete speci-Rcation of the problem. Assumptions built into themodels predispose the results to be either too stableor too unstable. Thus there are no reliable modelsthat can analyze the glaciological history of theAntarctic Ice Sheet. Predictive models rely on lin-earizations to provide acceptable accuracy for thenear future but become progressively less accuratethe longer the timescale.

See also

Bottom Water Formation. Icebergs. Ice+Ocean Inter-action. Sub Ice-shelf Circulation and Processes.

Further ReadingDoake CSM and Vaughan DG (1991) Rapid disintegra-

tion of Wordie Ice Shelf in response to atmosphericwarming. Nature 350: 328}330.

Doake CSM, Corr HFJ, Rott H, Skvarca P and YoungN (1998) Breakup and conditions for stability of thenorthern Larsen Ice Shelf, Antarctica. Nature 391:778}780.

Hindmarsh RCA (1993) Qualitative dynamics of marineice sheets. In: Peltier WR (ed.) Ice in the ClimateSystem, NATO ASI Series, vol. 12, pp. 67}99. Berlin:Springer-Verlag.

Kellogg TB and Kellogg DE (1987) Recent glacial historyand rapid ice stream retreat in the Amundsen Sea.Journal of Geophysical Research 92: 8859}8864.

Robin G de Q (1979) Formation, Sow and disintegrationof ice shelves. Journal of Glaciology 24(90):259}271.

Scambos TA, Hulbe C, Fahnestock M and BohlanderJ (2000) The link between climate warming and break-up of ice shelves in the Antarctic Peninsula. Journal ofGlaciology 46(154): 516}530.

van der Veen CJ (ed.) (1997) Calving Glaciers: Report ofa Workshop 28 February}2 March 1997. BPRC ReportNo. 15. Columbus, Ohio: Byrd Polar Research Center,The Ohio State University.

van der Veen CJ (1999) Fundamentals of Glacier Dynam-ics. Rotterdam: A.A. Balkema.

Vaughan DG and Doake CSM (1996) Recent atmosphericwarming and retreat of ice shelves on the AntarcticPeninsula. Nature 379: 328}331.

IGNEOUS PROVINCES

M. F. Cof\n, University of Texas at Austin,Austin, TX, USAO. Eldholm, University of Oslo, Oslo, Norway

Copyright ^ 2001 Academic Press

doi:10.1006/rwos.2001.0463

Introduction

Large igneous provinces (LIPs) are massive crustalemplacements of predominantly Fe- and Mg-rich(maRc) rock that form by processes other than nor-mal seaSoor spreading. LIP rocks are readily distin-guishable from the products of the two other majortypes of magmatism, midocean ridge and arc, on theEarth’s surface on the basis of petrologic, geochemi-cal, geochronologic, geophysical, and physical vol-canological data. LIPs occur on both the continentsand oceans, and include continental Sood basalts,volcanic passive margins, oceanic plateaus, submar-ine ridges, seamounts, and ocean basin Sood basalts(Figure 1, Table 1). LIPs and their small-scale

analogs, hot spots, are commonly attributed todecompression melting of hot, low density mantlematerial known as mantle plumes. This type ofmagmatism currently represent&10% of the massand energy Sux from the Earth’s deep interior to itscrust. The Sux may have been higher in the past,but is episodic over geological time, in contrast tothe relatively steady-state activity at seaSoor spread-ing centers. Such episodicity reveals dynamic, non-steady-state circulation within the Earth’s mantle,and suggests a strong potential for LIP emplace-ments to contribute to, if not instigate, major envir-onmental changes.

Composition, Physical Volcanology,Crustal Structure, and Mantle Roots

LIPs are deRned by the characteristics of theirdominantly Fe- and Mg-rich (maRc) extrusiverocks; these most typically consist of subhorizontal,subaerial basalt Sows. Individual Sows can extend

1290 IGNEOUS PROVINCES


Documents

ICE-SHELF STABILITY - Judith Currycurry.eas.gatech.edu/.../Chapter10/Ency_Oceans/Ice-shelf_Stability.pdf · future. What are Ice Shelves? Physical and geographical setting An ice