formation by fluvial and glacial

David R. Montgomery Department of and Space Sciences, University of , Seattle, Washington 98195, USA

ABSTRACT Cross-valley pro®les from the west slope of the , Washington, are used to investigate the relative effects of ¯uvial and glacial erosion on valley formation. Unlike most ranges where and sequentially occupied the same valleys, neighboring valleys in the Olympic developed in similar lithologies but were subject to different degrees of glaciation, allowing comparison of the net effect of glacial and ¯uvial processes integrated over many glacial cycles. Upslope drainage area was used to normalize comparisons of valley width, -crest-to-valley-bottom relief, and valley cross-sectional area as measures of net differences in the mass of excavated from below ridgelines for 131 valley-spanning transects. Valley width, relief, and cross-sectional areas are similar for glaciated, partly glaciated, and unglaciated (¯uvial) valleys with drainage areas of Ͻ10 km2, but diverge downslope. Glaciated valleys draining Ͼ50 km2 reach two to four times the cross-sectional area and have to 500 m greater relief than comparable ¯uvial valleys; partly glaciated valleys have intermediate dimensions. At dis- tances of Ͼ5 km from valley heads, the cumulative upstream volume of rock removed to form valleys is two to four times greater in glacially incised valleys than in ¯uvial valleys. The ®nding of strong differences in the net result of valley excavation by ¯uvial and glacial erosion supports the interpretation that alpine glaciers are more effective erosional agents than are rivers and implies that large alpine valleys deepened and enlarged in response to glaciation.

Keywords: glacial, ¯uvial, erosion, valley formation.

INTRODUCTION et al., 1996), recent studies in areas of rapid (Augustinus, 1992b), but there has been no There is no longer serious debate about uplift document incision rates compara- direct quantitative comparison of valley mor- whether glaciers can erode their beds and ble to high erosion rates in glaciated regions phology in comparable glaciated and ungla- sculpt topography, but debate continues over (e.g., Burbank et al., 1996). Moreover, sedi- ciated basins. Here I report evidence for sub- whether glaciers are more erosive than rivers. ment yield data are dif®cult to translate di- stantial differences in the geomorphic Some workers ®nd evidence for greater rates rectly into erosion rates owing to long-term expression of ¯uvial and glacial erosion in the of glacial erosion than ¯uvial erosion (e.g., storage in ¯uvial systems (Dunne et , Washington. Clague, 1986; Braun, 1989; Harbor and War- al., 1998; Goodbred and Keuhl, 1998) and the burton, 1992, 1993; Clayton, 1996; Hallet et potential for lag times between erosion and STUDY AREA al., 1996; Kirkbride and Mathews, 1997), sediment delivery to exceed the duration of The Olympic Mountains of western Wash- whereas others ®nd evidence for low rates of typical glacial cycles (Church and Slaymaker, ington comprise an accretionary wedge that glacial erosion or little difference between ¯u- 1989). rose from the Paci®c Ocean in the late Mio- vial and glacial erosion rates (e.g., Sugden, Of central importance to understanding cene in response to convergence of the Juan 1976, 1978; LindstroÈm, 1988; Hicks et al., feedback among change, erosion, and de Fuca plate and North America (Tabor and 1990; Summer®eld and Kirkbride, 1992; Heb- in alpine areas is whether glaciation Cady, 1978). Rock-uplift rates across the don et al., 1997; Lidmar-BergstroÈm, 1997). leads to more erosion and greater relief than range have been steady since ca. 14 Ma, and However, comparisons of ¯uvial and glacial can be attributed to ¯uvial processes. Whipple the range is thought to have been in topo- erosion rates face the fundamental problem et al. (1999) analyzed river longitudinal pro- graphic steady state for most of that time that glaciers and rivers sequentially occupied ®les and argued that any in¯uence of higher (Brandon et al., 1998). The Olympic Moun- the same valleys, making the signatures of erosion rates in glacial periods is likely to tains are just south of the maximum extent of glacial and ¯uvial processes on valley size dif- have only limited impact on relief for equilib- Pleistocene ice sheets but were the site of re- ®cult to deconvolve (Roberts and Rood, rium river pro®les in systems where peated episodes of alpine glaciation. Geologic 1984). Global comparisons are further com- development of threshold bedrock slopes pre- mapping of the western Olympics shows that plicated by the fact that data come from wide- vents attainment of steeper valley sides some valleys in the range repeatedly hosted ly different settings, such as cold-based con- (Schmidt and Montgomery, 1995). Physically large Pleistocene valley glaciers, whereas oth- tinental ice sheets, warm-based alpine based models have been proposed for the de- ers had either no glaciers or smaller glaciers glaciers, and temperate, tropical, and arid riv- velopment of U-shaped glacial valley forms generally restricted to their headwaters (Wash- ers in a broad range of tectonic contexts. Al- from initially V-shaped ¯uvial valleys (Harbor ington Division of Geology and Earth Re- though the most comprehensive comparison et al., 1988; Hirano and Aniya, 1988; Harbor, sources Staff, 2001). The west side of the indicates that sediment yields are higher in 1992), and the in¯uence of rock strength on Olympic Mountains presents an unusual op- glaciated than in unglaciated regions (Hallet glacial valley form also has been investigated portunity in that a series of neighboring val-

᭧ 2002 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. Geology; November 2002; v. 30; no. 11; p. 1047±1050; 4 ®gures. 1047 Figure 1. Shaded relief map of west slope of Olympic Mountains, Wash- ington, showing locations of basins studied, cross sections (white bars), and (black are- as) as portrayed on 1: 100 000 scale geologic maps (Washington Divi- sion of Geology and Earth Resources Staff, 2001).

leys with similar geology had strikingly dif- were limited to drainage areas of 0.1 km2 to ferent glacial histories but shared the same 500 km2, because ®ner scale features are not general climate variability through time. As well resolved and glaciers did not extend the range has been in a topographic steady much beyond this range in valley size. state since the late Miocene, a well-developed Comparisons of valley morphometry were ¯uvial valley network must have evolved prior normalized by the drainage area upslope of to onset of Pleistocene glaciation, which mod- each cross section. Mean slope and precipita- i®ed an initial ¯uvially incised valley network. tion also were calculated for the drainage ba- River systems exhibit systematic scaling in sin upslope of each cross section. Valleys that which and valley size increase with head highest in the range developed large val- drainage area or river (Leopold and ley glaciers in the glacial climate, but the Maddock, 1953), and size similarly dominantly ¯uvial valleys on the western scales with drainage area or ice discharge slope of the Olympic Mountains also extend (Roberts and Rood, 1984; Augustinus, 1992a). to high elevations where precipitation reaches Such relationships suggest that systematic dif- 4±5 m´yrϪ1. Modern drainage divides do not ferences in the morphometry of valleys on the necessarily coincide with Pleistocene ice di- western slope of the Olympic Peninsula be- vides, but the two should be similar in the tween those formed primarily by ¯uvial ero- high-relief valleys in the core of the range, and sion and those sequentially incised by both any glacial spillover would account for a pro- glacial and ¯uvial erosion can be used to gressively smaller proportion of the drainage gauge long-term morphologic effects of gla- area downstream through the valley network. ciation on valley form. The relief given by the difference between the present maximum and minimum elevations on METHODS each valley-spanning topographic cross sec- The spatial distribution of glacial moraines tion provides only a minimum constraint on was used to assess the relative degree of gla- the local bedrock relief, because postglacial cial in¯uence on nine valleys (Fig. 1). Fully alluvial valley ®lls now occupy valley ¯oors. glaciated valleys (Hoh, South Fork Hoh, and The methodology I employ does not evaluate Figure 2. Comparison of valley morphometry Queets) had large, valley-spanning Fraser and erosion rates directly, but allows assessment vs. drainage area for (A) valley width; (B) ridge-crest-to-valley-bottom relief; and (C) pre-Fraser age moraines. Partially glaciated of the integrated long-term signature of ero- valley cross-sectional area. Black squares valleys (Bogachiel, Sams, and Tshletshy) had sional processes on the volume of material re- represent valleys with major alpine glaciers less extensive glacial in¯uences with small moved from between valley walls. Hence the (Hoh, South Fork Hoh, and Queets), trian- Fraser age moraines located partway down the approach allows examination of the relative ef- gles represent partly glaciated valleys (Bo- gachiel, Tshletshy, and Sams), and circles valley. Unglaciated valleys had no mapped ®cacy of ¯uvial and glacial processes as valley- represent unglaciated ¯uvial valleys (Clear- moraines (Clearwater, North Fork Bogachiel, forming agents. , North Fork Bogachiel, and Solleks). and Solleks). A total of 131 valley-spanning, ridgetop-to-ridgetop cross sections were de- RESULTS rived from a 10-m-grid digital elevation model Valleys carved by alpine glaciers on the much as twice as wide as ¯uvial valleys. This (DEM) compiled for the range; 54 transects western slope of the Olympic Mountains are greater cross-valley ridge-to-ridge distance in- were from fully glaciated basins, 42 were generally wider than ¯uvially carved valleys, dicates valley excavation well beyond that ex- from partially glaciated basins, and 35 were although the difference is less clear at small pected to result simply from conversion of an from unglaciated basins. Cross sections were drainage areas (Fig. 2A). The width of glaci- initially V-shaped valley to a U-shaped valley, oriented orthogonal to the valley centerline ated valleys increases faster downslope than which can be done with no change in overall and located to avoid the complicating in¯u- the width of ¯uvial valleys such that at drain- valley width (Harbor et al., 1988; Harbor, ence of valleys. In addition, data age areas of Ͼ10 km2, glacial valleys are as 1992).

1048 GEOLOGY, November 2002 Ridge-crest-to-valley-bottom relief for all three types of valley initially increases at low drainage areas and then decreases at higher drainage areas (Fig. 2B). There is little differ- ence in the relief of ¯uvial and glacial valleys for drainage areas of Ͻ10 km2. At larger drainage areas, the relief of glacial valleys progressively exceeds that of ¯uvial valleys; differences in relief increase to ϳ500 m for drainage areas of Ͼ50 km2. The drainage area associated with the maximum cross-valley re- lief is offset, with glacial ex- tending farther downslope than the maximum local relief in ¯uvial valleys. Hence, the great- Figure 4. Mean basin slope (left) and mean est relief enhancement by alpine glaciers is fo- basin precipitation (right) for drainage areas cused signi®cantly downvalley from of 1, 10, and 100 km2 along glaciated valleys source areas. (black squares), partially glaciated valleys (triangles), and unglaciated ¯uvial valleys The cross-sectional areas of glaciated, part- (circles). Mean basin slope derived from ly glaciated, and ¯uvial valleys vary within U.S. Geological Survey 10 m digital eleva- about a factor of 2 around their underlying tion models; mean basin precipitation trends (Fig. 2C). Although cross-sectional ar- based on 4 km grid of mean annual precip- itation from 1961±1990 (Daly et al., 1994). eas of glaciated valleys are comparable to those of ¯uvial valleys at drainage areas of times greater rock mass in carving their val- Ͻ10 km2, the difference increases to two to leys in the same geologic and climatic setting. four times for drainage areas of Ͼ50 km2. This greater removal of rock through glacial Hence, there is little difference in the amount widening and deepening provides an estimate of material differentially eroded to form val- of erosion associated with conversion of ¯u- leys between ¯uvial and glacial valleys for vial valleys to glacial-valley form. However, small drainage areas at high elevation, but dif- changes in valley shape or dimensions do not ferences between the two valley types increase address directly rates of erosion, because a dramatically for larger drainage areas. change in valley shape may trigger only a The cumulative upstream volume of rock transient increase in erosion rates during the removed to form valleys (determined from the period of adjustment from one equilibrium valley cross-sectional area and the distance form to another. downvalley between successive cross sec- Differences in the size of ¯uvial and glacial tions) is substantially different for ¯uvial and valleys are not uniformly distributed along the glacial valleys (Fig. 3). At distances of Ͻ5km Figure 3. A: Cumulative volume of rock re- valley system, and there is little difference in from the valley head there is no compelling moved from upstream to form valley as the morphometry of headwater valleys. The difference, but at greater distances downval- function of distance down valley. B: Enlarge- progressively larger downvalley differences ley, the cumulative valley volume increases ment of uppermost 20 km of each valley sys- imply that a nonlinear glacial erosion law de- much more rapidly for glaciated valleys than tem. HÐHoh, H؅ÐSouth Fork Hoh, QÐ scribes valley excavation in glaciated terrain for ¯uvial valleys, and partly glaciated valleys Queets, TÐTshletshy, BÐBogachiel, SÐ Sams, CÐClearwater, B؅ÐNorth Fork and leads to formation of hanging valleys in are intermediate. At distances of 10 km -Ͼ Bogachiel, and S؅ÐSolleks. Black squaresÐ valley systems by overexcavating large val from valley heads, glaciated valleys represent glaciated valleys; trianglesÐpartially glaci- leys, as predicted by a physically based model net removal of two to four times the mass of ated valleys; circlesÐunglaciated ¯uvial valleys. of glacial-valley longitudinal-pro®le evolution rock than ¯uvial valleys. (MacGregor et al., 2000). In the eastern Sierra Mean basin slope varies by only a few de- Nevada, Brocklehurst and Whipple (2002) grees between the three types of valleys for for basins occupied by alpine glaciers parallel also found that glaciers increased relief by 2 basins Ͼ10 km (Fig. 4). The headwaters of prior reports that fjord cross sections scale Ͻ100 m in basins with drainage areas of 3± two of the glaciated valleys (Hoh and South with drainage area and therefore ice ¯ux (Rob- 36 km2 and lengths of 3.4±16.4 km. The lack Fork Hoh) receive substantially more precip- erts and Rood, 1984; Augustinus, 1992a). The of relief enhancement in glacial source areas itation than the other valleys (including the relatively small differences in mean basin in both the Olympics and the Sierra Nevada glaciated Queets). However, in basins with a slope and precipitation between the three val- suggests that signi®cant relief enhancement is 2 drainage area of 100 km the mean basin pre- ley types for large basins (i.e., 100 km2) in- limited to large alpine glaciers. cipitation for ¯uvial and partly glaciated val- dicate that morphometric differences noted Previous analyses of the Olympic Moun- leys ranges from 77% to 97% and from 86% here primarily re¯ect differences in the pro- tains concluded that threshold slopes were de- to 104%, respectively, of the values for gla- cesses of valley development. On the west veloped throughout the core of the range; ciated valleys. slope of the Olympic Mountains, valleys with mean slopes are relatively uniform, even extensive glaciation are larger than those that though there are large gradients in rock-uplift DISCUSSION AND CONCLUSIONS underwent minor glaciation, which are in turn rate across the range (Montgomery, 2001). Systematic increases in valley size with in- larger than unglaciated valleys. Compared to The greater relief observed in glacial valleys creasing drainage area (and therefore ice ¯ux) rivers, alpine glaciers removed two to four could be reconciled with development of

GEOLOGY, November 2002 1049 threshold slopes by the greater width of gla- Augustinus, P.C., 1992b, The in¯uence of rock mass A numerical model of development ciated valleys. Close examination of the DEM strength on glacial valley cross-pro®le mor- by glacial erosion: Nature, v. 333, p. 347±349. phometry: A case study from the Southern Hebdon, N.J., Atkinson, T.C., Lawson, T.J., and of the study area suggests that the glaciated Alps, New Zealand: Earth Surface Processes Young, I.R., 1997, Rate of glacial valley deep- valleys of the Hoh and Queets Rivers expand- and , v. 17, p. 39±51. ening during the late Quaternary in Assynt, ed at the expense of tributary valleys. In par- Brandon, M.T., Roden-Tice, M.K., and Garver, J.I., : Earth Surface Processes and Land- ticular, the drainage pattern suggests that what 1998, Late Cenozoic exhumation of the Cas- forms, v. 22, p. 307±315. Hicks, D.M., McSaveney, M.J., and Chinn, T.J.H., appears to have once been the headwaters of cadia accretionary wedge in the Olympic Mountains, northwest Washington State: Geo- 1990, Sedimentation in proglacial Ivory Lake, the Solleks River has been captured by wid- logical Society of America Bulletin, v. 110, Southern Alps, New Zealand: Arctic and Al- ening of the valley. p. 985±1009. pine Research, v. 22, p. 26±42. An intriguing implication of the greater ex- Braun, D.D., 1989, Glacial and periglacial erosion Hirano, M., and Aniya, M., 1988, A rational expla- of the Appalachians: , v. 2, nation of cross-pro®le morphology for glacial cavation of rock to form larger valleys during valleys and of glacial valley development: the Pleistocene is the potential in¯uence on p. 233±256. Brocklehurst, S.H., and Whipple, K.X., 2002, Gla- Earth Surface Processes and Landforms, v. 13, elevations in the core of the range. The sub- cial erosion and relief production in the East- p. 707±716. stantial widening and deepening of valleys in ern Sierra Nevada, California: Geomorpholo- Kirkbride, M., and Mathews, D., 1997, The role of gy, v. 42, p. 1±14. ¯uvial and glacial erosion in landscape evo- response to glaciation implies the potential for lution: The Ben Ohau Range, New Zealand: isostatic rebound to raise surrounding peaks Brozovic, N., Burbank, D.W., and Meigs, A.J., 1997, Climatic limits on landscape develop- Earth Surface Processes and Landforms, v. 22, p. 317±327. and suggests that some of the elevation of the ment in the northwestern Himalaya: Science, Leopold, L.B., and Maddock, T., 1953, The hydrau- highest peak in the range (Mount Olympus) v. 276, p. 571±574. lic geometry of channels and some may re¯ect Pleistocene valley widening in the Burbank, D.W., Leland, J., Fielding, E., Anderson, physiographic implications: U.S. Geological R.S., Brozovic, N., Reid, M.R., and Duncan, surrounding moat of large valleys. Montgom- Survey Professional Paper 252, 57 p. ery and Greenberg (2000) calculated that ex- C., 1996, Bedrock incision, rock uplift and Lidmar-BergstroÈm, K., 1997, A long-term perspec- threshold hillslopes in the northwestern Him- tive on glacial erosion: Earth Surface Process- cavation of valleys in the core of the range alayas: Nature, v. 379, p. 505±510. could have resulted in as much as 500±750 m es and Landforms, v. 22, p. 297±306. Church, M., and Ryder, J.M., 1972, Paraglacial sed- LindstroÈm, E., 1988, Are roches mountoneÂs mainly of isostatically induced rock uplift at Mount imentation: A consideration of ¯uvial processes preglacial forms?: Geogra®ska Annaler, Olympus. If glaciers more than doubled the conditioned by glaciation: Geological Society v. 70A, p. 323±332. preglacial valley volume in the core of the of America Bulletin, v. 83, p. 3059±3072. MacGregor, K.R., Anderson, R.S., Anderson, S.P., Church, M., and Slaymaker, O., 1989, Disequilib- range, as implied by this analysis, then onset and Waddington, E.D., 2000, Numerical sim- rium of Holocene sediment yield in glaciated ulations of glacial-valley longitudinal pro®le of Pleistocene glaciation could have raised British Columbia: Nature, v. 337, p. 452±454. evolution: Geology, v. 28, p. 1031±1034. Mount Olympus by several hundred meters. Clague, J.J., 1986, The Quaternary stratigraphic re- Montgomery, D.R., 2001, Slope distributions, Offset of maximum glaciogenic relief well cord of British ColumbiaÐEvidence for epi- threshold hillslopes and steady-state topogra- downstream of glacier source areas is consis- sodic sedimentation and erosion controlled by phy: American Journal of Science, v. 301, glaciation: Canadian Journal of Earth Scienc- p. 432±454. tent with the potential for positive feedback in es, v. 23, p. 885±894. Montgomery, D.R., and Greenberg, H., 2000, Local which widening and deepening of large gla- Clayton, K., 1996, Quanti®cation of the impact of relief and the height of Mount Olympus: Earth ciated valleys enhance rock uplift in glacier glacial erosion on the British Isles: Institute of Surface Processes and Landforms, v. 25, source areas, leading to higher elevations, British Geographers Transactions, v. 21, p. 386±396. greater snow accumulation, and larger glaciers p. 124±140. Roberts, M.C., and Rood, K.M., 1984, The role of the ice contributing area in the morphology of (Brozovic et al., 1997). Daly, C., Neilson, R.P., and Phillips, D.L., 1994, A statistical-topographical model for mapping transverse , British Columbia: Geograf- The generally greater width of glaciated climatological precipitation over mountainous iska Annaler, v. 66A, p. 381±393. valleys in the Olympic Mountains demon- terrain: Journal of Applied Meteorology, Schmidt, K.M., and Montgomery, D.R., 1995, Lim- strates the importance of valley widening as v. 33, p. 140±158. its to relief: Science, v. 270, p. 617±620. Sugden, D.E., 1976, A case against deep erosion of Dunne, T., Mertes, L.A.K., Meade, R.H., Richey, part of geomorphic response to climate shields by ice sheets: Geology, v. 4, J.E., and Forsbergh, B.R., 1998, Exchanges of change. The effects of glacial erosion on the p. 580±582. sediment between the ¯ood and channel Sugden, D.E., 1978, Glacial erosion by the Lauren- net excavation of rock from the valley system of the in Brazil: Geological So- of the western Olympics demonstrate the po- tide ice sheet: Journal of , v. 20, ciety of America Bulletin, v. 110, p. 450±467. p. 367±391. tential for glaciation to increase valley size Goodbred, S.L., Jr., and Keuhl, S.A., 1998, - Summer®eld, M.A., and Kirkbride, M.P., 1992, Cli- and relief in alpine topography. Although rates plain processes in the Bengal Basin and the mate and landscape response: Nature, v. 355, of erosion may be greatest under an oscillating storage of - sedi- p. 306. ment: An accretion study using 137Cs and dominance of glacial and ¯uvial processes Tabor, R.W., and Cady, W.M., 1978, The structure 210Pb geochronology: Sedimentary Geology, of the Olympic Mountains, WashingtonÐ (Church and Ryder, 1972), an implication of v. 121, p. 239±258. Analysis of a subduction zone: U.S. Geologi- the ®ndings reported here is that the onset of Hallet, B., Hunter, L., and Bogen, J., 1996, Rates cal Survey Professional Paper 1033, 38 p. Pleistocene glaciation should have resulted in of erosion and sediment yield by glaciers: A Washington Division of Geology and Earth Re- a pulse of sediment yield in areas subject to review of ®eld data and their implications: sources Staff, 2001, Digital geologic maps of Global and Planetary Change, v. 12, the 1:100 000 quadrangles of Washington: alpine glaciation during the period while val- p. 213±235. Washington Division of Geology and Earth leys adjusted to new forms. Harbor, J.M., 1992, Numerical modeling of the de- Resources Digital Report 2, version 1 (CD- velopment of U-shaped valleys by glacial ero- ROM). ACKNOWLEDGMENTS sion: Geological Society of America Bulletin, Whipple, K.X., Kirby, E., and Brocklehurst, S.H., Liila Woods prepared cross sections and Harvey v. 104, p. 1364±1375. 1999, Geomorphic limits to climate-induced Greenberg created Figure 1. A constructive review Harbor, J., and Warburton, J., 1992, Glaciation and increases in topographic relief: Nature, v. 401, by Jonathan Tompkin is also appreciated. rates: Nature, v. 356, p. 751. p. 39±43. Harbor, J., and Warburton, J., 1993, Relative rates Manuscript received January 23, 2002 REFERENCES CITED of glacial and nonglacial erosion in alpine en- Revised manuscript received July 24, 2002 Augustinus, P.C., 1992a, Outlet glacier trough sizeÐ vironments: Arctic and Alpine Research, Manuscript accepted July 26, 2002 Drainage area relationships, Fiordland, New v. 25, p. 1±7. Zealand: Geomorphology, v. 4, p. 347±361. Harbor, J.M., Hallet, B., and Raymond, C.F., 1988, Printed in USA

1050 GEOLOGY, November 2002