q 2005 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or editing@geosociety.org.Geology; December 2005; v. 33; no. 12; p. 933–936; doi: 10.1130/G21957.1; 6 figures. 933

Alpine landscape evolution dominated by cirque retreatMichael Oskin* Department of Geological Sciences, University of North Carolina, Chapel Hill, North Carolina 27599-3315, USADouglas W. Burbank* Department of Earth Sciences, University of California–Santa Barbara, Santa Barbara, California 93106, USA

ABSTRACTDespite the abundance in alpine terrain of glacially dissected landscapes, the magnitude

and geometry of glacial erosion can rarely be defined. In the eastern Kyrgyz Range ofcentral Asia, a widespread unconformity exhumed as a geomorphic surface provides aregional datum with which to calibrate erosion. As tectonically driven surface uplift hasprogressively pushed this surface into the zone of ice accumulation, glacial erosion hasoverprinted the landscape. With as little as 500 m of incision into rocks underlying theunconformity, distinctive glacial valleys display their deepest incision adjacent to cirqueheadwalls. The expansion of north-facing glacial cirques at the expense of south-facingvalleys has driven the drainage divide southward at rates of as much as two to three timesthe rate of valley incision. Glacial erosion rules based on ice flux incompletely explain thisdominance of cirque retreat over valley incision. Local processes that either directly sapcirque headwalls or inhibit erosion down-glacier appear to control, at least initially, alpinelandscape evolution.

Keywords: glacial erosion, cirque retreat, Kyrgyz Range, Tien Shan.

Figure 1. Exhumed pre-Cenozoic unconformity and maximum glacial extent in east-ern Kyrgyz Range. Present-day drainage divide is shown as line of peaks (elevationsin meters) connected by heavy gray solid line. A–D are center lines of swath profilesin Figure 2. Inset is shaded relief map of central Asia with Kyrgyz Range crest ashachured line. Topography used to construct Figures 1–5 is from Shuttle Radar To-pography Mission.

INTRODUCTIONThe remarkable coincidence of glacial

snowlines with average height of high moun-tains suggests that glaciation may exert pri-mary control on mountain-range elevation, ir-respective of tectonic rates (Broecker andDenton, 1990; Brozovic et al., 1997; Mont-gomery et al., 2001; Spotila et al., 2004). Thusthe topographic evolution that accompaniesthe transformation of nascent, unglaciatedlandscapes to highly glaciated ones is a key tounraveling the erosional component of aclimate-erosion feedback system (Molnar andEngland, 1990). Comparative geomorphicstudies support contentions that glaciation en-hances valley erosion and topographic relief(Brocklehurst and Whipple, 2002; Kirkbrideand Matthews, 1997; Montgomery, 2002;Small and Anderson, 1998). Observations ofplanimetric cirque expansion (Federici andSpagnolo, 2004; Olyphant, 1981) and truncat-ed glacial valleys (Brocklehurst and Whipple,2002) indicate that cirque retreat could play acompeting role by expanding valleys horizon-tally and removing high topography. The sig-nificance of cirque retreat in limiting mountainrange height, however, has not been conclu-sively established.

For tectonically active mountain ranges un-dergoing surface uplift, their progressive pen-etration into the zone of glaciation can providea proxy for transformation of landscapes dur-ing the well-documented late Cenozoic cool-ing (Shackleton and Opdyke, 1973). Growthof the Tien Shan (Fig. 1, inset), Kyrgyzstan,when combined with its unique geologic his-tory, establishes a natural laboratory to inves-

*E-mails: oskin@unc.edu; burbank@crustal.ucsb.edu.

tigate glacial valley erosion and cirque retreat.Basement-cored uplifts of the northern TienShan exhume resistant Paleozoic plutonic andmetamorphic bedrock from beneath unconsol-idated Cenozoic nonmarine sedimentary stra-ta. A regionally extensive, pre-Cenozoic low-relief unconformity separates these units(Chediya, 1986). In the arid environment ofcentral Asia, significant areas of this exhumedunconformity form geomorphic surfaces thatappear little modified by fluvial erosion and

instead faithfully delineate the geometry ofrange-scale folding (Burbank et al., 1999).When surface uplift causes ranges to intersectthe local Pleistocene glacial equilibrium-linealtitude (ELA), however, glacial erosion maycause significant incision of the Paleozoicrocks underlying the unconformity. Recon-struction of this datum over glaciated topog-raphy thus enables reliable estimation of themagnitude of glacial bedrock erosion.

Late Cenozoic uplift of the Kyrgyz Rangeprovides a structurally well controlled exper-iment from which to measure sequential gla-cial landscape evolution. Uplift occurs viahanging-wall folding and slip on a reversefault system that emplaces resistant Paleozoicquartzite, metaconglomerate, granitic, and me-tavolcanic rocks northward over Miocene–Quaternary foreland-basin strata. Bedrockthermochronology supports erosion rates of;1 km/m.y. from 3 Ma to present in the cen-tral, highest part of the range (Bullen et al.,2001). Detrital apatite fission-track ages fromforeland-basin strata and modern rivers indi-cate progressively less exhumation in the east-ern Kyrgyz Range (Bullen et al., 2003), con-sistent with preservation of exhumed, tiltedremnants of the unconformity on .25% of the

934 GEOLOGY, December 2005

Figure 2. Topographic swath profiles acrosseasternmost Kyrgyz Range. Each profilesamples topography from a 5-km-wideswath centered on profile lines shown inFigure 1. Light gray band depicts range ofminimum and maximum topography, withmean values shown as thicker central line.Horizontal black arrow depicts extent of geo-morphic surfaces formed by exhumed rem-nants of pre-Cenozoic unconformity (Unc.).ELA—last glacial maximum equilibrium-linealtitude. White arrow is extent of last maxi-mum glaciation. Glaciated areas correspondto areas of higher relief and erosional re-moval of pre-Cenozoic unconformity atrange crest. Extent of glaciation is inverselyproportional to northward position of drain-age divide, illustrated by vertical dashedlines. Highest elevation of range crest isshown by downward arrows.

Figure 3. A: Advanced Spaceborne Thermal Emission and Reflection Radiometerscene of geomorphic surfaces formed by exhumed pre-Cenozoic unconformity onsouth slope of Kyrgyz Range. See Figure 1 for location. Surfaces are visible asmedium albedo with intricate dendritic stream network. B: Interpreted slope mapof area shown in A. Steeper slopes (darker gray) correspond to valley sides. Whitesolid lines outline surface remnants mapped from analysis of topography andsatellite imagery. Elevation contours at 100 m intervals (black lines) show minimallocal relief formed by incision of geomorphic surfaces. White dashed line showsfolded Paleozoic contact (truncated by unconformity) between metavolcanic rocks(mv) and quartzite (q).

Figure 4. Contours of reconstructed pre-Cenozoic unconformity elevation (solid lines)and slope (dashed lines) overlain on shaded relief image of eastern Kyrgyz Range.DUnconformity depicts elevations within 100 m and 200 m from modeled unconfor-mity and compares well with its mapped exhumed remnants in Figure 1. Dotted linesshow gradients of north-facing glacial valley floors. Preglaciation divide extrapolatedalong front of prominent faceted spur ridges that separate glaciated valleys.

easternmost 50 km of its south-facing slope(Fig. 1). These observations support an east-ward propagation or an eastward-decliningrate of coeval rock and surface uplift that hasprogressively elevated the range crest into thezone of glaciation above the local PleistoceneELA.

GLACIAL EROSION OF THEEASTERN KYRGYZ RANGE

Exhumation of the pre-Cenozoic unconfor-mity as a geomorphic surface underpins thequantification of erosion. Swath topographicprofiles (Fig. 2) reveal minimal topographicrelief development on its unglaciated south-facing slope. Maximum elevations of these ar-eas are defined by low-relief upland surfaces

with concordant elevation and southward tiltpreserved between larger incised drainagesthat descend from the range crest (Fig. 3).These surfaces merge with outcrops of the un-conformity beneath Cenozoic strata at the baseof the range. A fourth-order polynomial sur-face with curvature of ,18/km fit to the dis-tribution and tilt of mapped remnants definesa continuous envelope over the south-facingslope of the range (Fig. 4). This envelopecaptures the mapped unconformity within6100 m (1s), over .1500 m of structural re-lief. Tilting of the restored unconformity sur-

face increases both with elevation and fromeast to west, consistent with a gradient of slipover a curviplanar reverse fault at depth.

Comparison of three adjacent catchmentsincised into the south-facing slope of the Kyr-gyz Range reveals details of the form of initialglacial erosion and the continuity of interflu-vial geomorphic surfaces (Fig. 5). All threestreams descend at steep (.88) slopes incised,600 m into the tilted unconformity pre-served at concordant undissected interfluves.The fluvial part of each valley is analogous toa half-pipe, with uniform valley width and

GEOLOGY, December 2005 935

Figure 5. Valley profiles, incision beneath reconstructed unconfor-mity, and catchment width from three adjacent south-facing streamswith varying amounts of glaciation. See Figure 4 for catchment lo-cations. In absence of glaciation, incised valleys (S1) are shaped likehalf-pipes with uniform width. Glacial erosion corresponds to incisionmaxima at cirque floors and widening of catchments in zone of iceaccumulation, supporting valley expansion via cirque retreat.

Figure 6. Landscape evolution of eastern Kyrgyz Range with progressive rock uplift (cf. Figs. 2A–2C). Light and dark grayrepresent ridge line and valley elevation, respectively. Dashed line shows extrapolation of folded unconformity surfaceover north slope of range. 1—Range form prior to glacial erosion with divide controlled by fluvial erosion and range-scalestructural fold axis. 2—Initial glacial erosion of north-facing valley causes southward divide migration relative to fold axis.Unconformity surface is preserved on south-facing slope at range crest. 3—Further range uplift accompanied by additionalcirque retreat and expansion of north-facing valley. Note that peak of northern ridge line is higher than unconformitypreserved at drainage divide.

fairly uniform incision. In contrast, the glaci-ated parts of these catchments display signif-icant widening and deepening, with the great-est incision at their glaciated headwaters.Adjacent glaciated catchments show a similarpattern of maximum incision at their head-waters (Fig. 4).

North-facing glaciated catchments of theeastern Kyrgyz Range consist of high-elevation, gently sloped U-shaped valleyslinked to steep V-shaped canyons that descendto the range-bounding reverse fault (Fig. 1).Absence of preserved remnants of the uncon-formity near the fault limits precise estimationof incision of these catchments. To conserva-tively evaluate glacial incision by north-facingcatchments, the slopes of north-facing glaci-ated valleys are examined with respect toslope of the adjacent, south-dipping unconfor-mity surface preserved at the range crest (Fig.4). The difference of these slopes defines aratio of valley length to valley incision (0.3–0.4) that is significantly less than the averageslope [0.8 6 0.1 (1s), n 5 66] of cirque head-walls along the range crest.

The presence and extent of glaciation innorth-facing catchments corresponds tochanges of trend and elevation of the rangecrest. Outside the zone of glaciation, the rangecrest tracks ;5 km south of the trace of therange-bounding reverse fault, the range shape

reflects growth of the underlying geologicstructure (e.g., Fig. 2A), and peak elevationsclimb steadily from 2400 m near the Chu Riv-er in the east to 3400 m at the first glaciatedcatchment. Within the zone of glaciation, peakelevations stabilize at ;4000 m elevation andthe position of the range crest shifts system-atically southward relative to the range frontin proportion to the degree of late Pleistoceneice cover (e.g., Figs. 2B–2D).

DISCUSSIONThe gradient in rock uplift of the eastern

Kyrgyz Range permits along-strike space-for-time substitution of observations of incisionof the range by glaciers. Because the range isrising above an active reverse fault at ;1 km/m.y., glaciation of the eastern range tip is like-ly to have occurred sequentially during thePleistocene. Although each valley has aunique incision history, spatial trends of land-scape evolution that are shared among severaladjacent valleys may represent temporaltrends within an individual valley. Much ofthe transformation of the Kyrgyz Range intoits glacially sculpted form involves competi-tion between adjacent catchments that drivesmigration of drainage divides over time. Ex-humation of the pre-Cenozoic unconformityas a geomorphic surface datum affords aunique opportunity to quantify the degree of

glacial incision that accompanies progressivechanges in valley form.

Rapid formation of gently sloped upland U-shaped valleys accompanies initial glacial ero-sion of the eastern Kyrgyz Range. Glaciers onsouth-facing slopes accomplish this by wid-ening and eroding their valleys deeper thanadjacent fluvial valleys, evidenced by incisionand valley-width maxima near the range crest(Fig. 5). Approximately 0.5 km of incisioninto the unconformity surface forms thesecharacteristically glaciated valleys, consistentwith low erosion depths required to obtain sta-ble glacial valley cross sections in numericalsimulations (Harbor, 1992).

Systematic, relative southward migration ofthe drainage divide of the eastern KyrgyzRange shows that cirque retreat also accom-panies initial glacial erosion. Expansion ofnorth-facing valleys via glacial cirque retreatprovides a consistent explanation for thesespatial trends of landscape evolution with pro-gressive rock uplift above the local ELA (Fig.6): (1) range crest elevation stabilizes at;4000 m with the onset of erosion by gla-ciers; (2) further uplift and tilting of the rangeis accompanied by southward relative migra-tion of the drainage divide; and (3) the pre-Cenozoic unconformity is preserved as a geo-morphic surface at the range crest. Peakheights remain constant along trend, and therange crest preserves remnants of the ex-humed unconformity because the divide hasshifted southward with progressive uplift andglacial erosion (Fig. 2). In effect, during initialglacial erosion, the range crest remains moreor less pinned to the intersection of the 4000m contour line with the tilted unconformity.This explanation requires that a competitiveadvantage for cirque erosion exists for north-facing catchments over their south-facingcounterparts and that glacial erosion of theKyrgyz Range occurs, at least initially, by acombination of cirque retreat and glacial val-ley incision.

Alternative origins for southward migrationof the divide include differential fluvial inci-sion, exposure of lithologies that resist glacialerosion, or that the present divide formed inplace by range-scale folding. Range-scalefolding may be ruled out as an explanation

936 GEOLOGY, December 2005

because peaks and ridges north of the presentdivide are locally at the same elevation orhigher than the exhumed unconformity pre-served at the divide (Figs. 1, 2C, and 2D).This geometry requires that the structural axisof the Kyrgyz Range trends north of the pres-ent drainage divide in the zone of glacial ero-sion. Metasedimentary rocks that underlie theglaciated area strike transverse to valleys anddip steeply, inconsistent with lithologic con-trol of valley floor elevation. It is also unlikelythat the present low-gradient upland valleysformed entirely by glacial incision of steeperfluvial valleys (cf. Brocklehurst and Whipple,2002), because the valley heads terminate atuneroded remnants of the unconformity. Onlyheadward erosion of low-gradient glaciatedvalleys by cirque retreat could maintain thisgeometry at the range crest (Fig. 6). To esti-mate the total amount of cirque retreat andvalley incision requires additional knowledgeof the divide position that developed by range-scale folding and fluvial incision prior to gla-ciation. A tentative reconstruction of the orig-inal divide position across the easternmostfour glaciated catchments indicates between0.9 and 4.4 km of cirque retreat with only 0.4–1.3 km of valley incision (Fig. 4). Further ex-trapolation of the preglaciation divide sug-gests that .10 km of cirque retreat dominateserosion of north-facing valleys.

Evidence for substantial headward expan-sion of catchments via cirque retreat in theKyrgyz Range, the Sierra Nevada (Brockle-hurst and Whipple, 2002), and potentiallyelsewhere (Olyphant, 1981) compels a closerexamination of the resulting valley form. Theimmediate cause of cirque retreat is the col-lapse of steep headwalls at the expense of ad-jacent low-relief uplands. Processes that sapthe cirque walls at the head of the glacier ul-timately determine the retreat rate. Becausethe ratio of valley length to valley incision(0.3–0.4) in the headwaters of north-facingcatchments is two to three times less than theslope of adjacent cirque headwalls (0.8 60.1), sapping by valley incision alone is un-likely to cause the observed retreat. Rather,erosion localized at the base of cirque head-walls is required to propagate cirque retreatacross the range at a rate two to three timesgreater than the glacial valley incision rate. In-cision maxima developed at the heads ofsouth-facing glaciated valleys (Fig. 4) are con-sistent with a localized sapping process atcirque headwalls.

Because cirque headwalls form within theglacial accumulation zone, modeling cirqueerosion via ice-flux–based rules (Hallet, 1979,1996) predicts greater downstream valley in-cision than erosion of the cirque floor, a pre-diction incompatible with observations fromthe Kyrgyz Range. This may be reconciled if

subglacial water-pressure fluctuations withincirques locally enhance erosion (Hooke, 1991)or if development of glacial bedforms, such asterminal moraines and overdeepenings, re-stricts pressure fluctuations and erosion down-glacier (Alley et al., 2003). It is also possiblethat the integrated residence time of vestigialcirque glaciers under average Quaternary con-ditions (Porter, 1989) has been sufficient to tipthe balance of erosion in favor of cirque in-cision and consequent headward retreat. Sucherosion is consistent with preferred erosion ofnorth-facing catchments, where shading ofglacier surfaces lowers the local ELA by;200 m.

CONCLUSIONSAnalysis of the configuration and gradient

of glacial valleys in the eastern Kyrgyz Rangeindicates that cirque retreat dominates initialglacial erosion. This erosion is quantified inthe context of a progressively tilted pre-Cenozoic unconformity that is exhumed as ageomorphic surface. This cirque retreat maybe incompatible with existing ice-flux–basedglacial erosion rules and highlights a need tounderstand erosional processes localized with-in cirques in order to predict alpine landscapeevolution. Cirque retreat can effectively bevelacross an elevated alpine plateau and keeppace with moderate rock uplift rates, as doc-umented here for the Kyrgyz Range, whereretreat rates are at least two to three timesgreater than the accompanying valley incisionrate. Although cirque retreat may not be assignificant in all alpine settings (White, 1970),results from the Kyrgyz Range support thecontention that glacial valley incision com-bined with cirque retreat into interfluves maybe a viable mechanism whereby glaciers si-multaneously erode at high rates and limit al-pine relief.

ACKNOWLEDGMENTSThis research was supported by the National

Aeronautics and Space Administration Shuttle Ra-dar Topography Mission science program grantNAG5-13758. We thank K.Y. Abdrakmatov for lo-gistical support in Kyrgyzstan; K. Davis, M. Jakob,R. Heermance, and B. Bookhagen for assistance inthe field; and K. Whipple and J. Spotila for reviewsof earlier versions of this manuscript.

REFERENCES CITEDAlley, R.B., Lawson, A.C., Larson, G.J., Evenson,

E.B., and Baker, G.S., 2003, Stabilizing feed-backs in glacier-bed erosion: Nature, v. 424,p. 758–760.

Brocklehurst, S., and Whipple, K.X., 2002, Glacialerosion and relief production in the EasternSierra Nevada, California: Geomorphology,v. 42, p. 1–24.

Broecker, W.S., and Denton, G.H., 1990, The roleof ocean-atmosphere reorganizations in glacialcycles: Quaternary Science Reviews, v. 9,p. 305–341.

Brozovic, N., Burbank, D.R., and Meigs, A.J., 1997,Climatic limits on landscape development inthe northwestern Himalaya: Science, v. 276,p. 571–574.

Bullen, M.E., Burbank, D.W., Garver, J.I., and Ab-drakmatov, K.Y., 2001, Late Cenozoic tectonicevolution of the northwestern Tien Shan: Newage estimates for the initiation of mountainbuilding: Geological Society of America Bul-letin, v. 113, p. 1544–1559.

Bullen, M.E., Burbank, D.W., and Garver, J.I.,2003, Building the northern Tien Shan: Inte-grated thermal, structural, and topographicconstraints: Journal of Geology, v. 111,p. 149–165.

Burbank, D.W., McLean, L.J.K., Bullen, M.E., Ab-drakmatov, K.Y., and Miller, M.G., 1999, Par-titioning of intermontane basins by thrust-related folding, Tien Shan, Kyrgyzstan: BasinResearch, v. 11, p. 75–92.

Chediya, O.K., 1986, Morphostructure and neo-tectonics of the Tien Shan: Frunze, AcademiaNauk Kyrgyz CCP, 313 p.

Federici, P.R., and Spagnolo, M., 2004, Morpho-metric analysis on the size, shape, and arealdistribution of glacial cirques in the maritimeAlps (Western French–Italian Alps): Geograf-iska Annaler, v. 86, p. 235–248.

Hallet, B., 1979, A theoretical model of glacialabrasion: Journal of Glaciology, v. 23,p. 39–50.

Hallet, B., 1996, Glacial quarrying: A simple the-oretical model: Annals of Glaciology, v. 22,p. 1–8.

Harbor, J.M., 1992, Numerical modeling of the de-velopment of U-shaped valleys by glacial ero-sion: Geological Society of America Bulletin,v. 104, p. 1364–1375.

Hooke, R.L., 1991, Positive feedbacks associatedwith erosion of glacial cirques and overdee-penings: Geological Society of America Bul-letin, v. 103, p. 1104–1108.

Kirkbride, M., and Matthews, D., 1997, The role offluvial and glacial erosion in landscape evo-lution: The Ben Ohau Range, New Zealand:Earth Surface Processes and Landforms, v. 22,p. 317–327.

Molnar, P., and England, P., 1990, Late Cenozoicuplift of mountain ranges: Chicken or egg?:Nature, v. 346, p. 29–34.

Montgomery, D.R., 2002, Valley formation by flu-vial and glacial erosion: Geology, v. 30,p. 1047–1050.

Montgomery, D.R., Balco, G., and Willett, S.D.,2001, Climate, tectonics, and the morphologyof the Andes: Geology, v. 29, p. 579–582.

Olyphant, G.A., 1981, Allometry of cirque evolu-tion: Geological Society of America Bulletin,v. 92, p. 679–685.

Porter, S.C., 1989, Some geological implications ofaverage Quaternary glacial conditions: Qua-ternary Research, v. 32, p. 245–261.

Shackleton, N.J., and Opdyke, N.D., 1973, Oxygenisotope and paleomagnetic stratigraphy ofequatorial Pacific core V28-238: Oxygen iso-tope temperatures and ice volumes on a 105

and 106 year scale: Quaternary Research, v. 3,p. 39–55.

Small, E.E., and Anderson, R.S., 1998, Pleistocenerelief production in the Laramide mountainranges, western United States: Geology, v. 26,p. 123–126.

Spotila, J.A., Buscher, J.T., Meigs, A.J., and Re-iners, P.W., 2004, Long-term glacial erosionof active mountain belts: Example of theChugach-St. Elias Range, Alaska: Geology,v. 32, p. 501–504.

White, W.A., 1970, Erosion of cirques: Journal ofGeology, v. 78, p. 123–126.

Manuscript received 13 June 2005Revised manuscript received 1 August 2005Manuscript accepted 4 August 2005

Printed in USA