Blikra and Nemec 1998 Postglacial colluvium in wertern Norway.pdf

Postglacial colluvium in western Norway:depositional processes, facies and palaeoclimatic record

LARS H. BLIKRA* and W. NEMEC*Geological Survey of Norway, PO Box 3006 Lade, N-7002 Trondheim, Norway(E-mail: [email protected])Geological Institute, University of Bergen, N-5007 Bergen, Norway (E-mail: [email protected])

ABSTRACT

The postglacial Quaternary colluvial systems in western Norway are arrays of steep

fans, often coalescing into aprons, developed along the slopes of valley sides and fjord

margins. The coarse debris, derived from weathered gneissic bedrock and its glacial-

till mantle, varies from highly immature to mature. The depositional processes are

mainly avalanches, ranging from rockfalls and debrisows to snowows, but include

also waterow and debris creep. The mechanics and sedimentary products of these

processes are discussed, with special emphasis on snow avalanches, whose role as an

agent of debris transport is little-known to sedimentologists. The subsequent analysis

of sedimentary successions is focused on colluvial-fan deltas, which are very specic,

yet little-studied, coastal depositional systems. The stratigraphic variation and

depositional architecture of the colluvial facies assemblages, constrained by abundant

radiometric dates, are used to decipher the signal of regional climatic changes from

the sedimentary record. The stratigraphic data from two dozen local colluvial

successions are compiled and further compared with other types of regional

palaeoclimatic proxy record. The analysis suggests that the colluvial systems,

although dependent upon local geomorphic conditions, have acted as highly sensitive

recorders of regional climatic changes. The study as a whole demonstrates that

colluvial depositional systems are an interesting and important frontier of clastic

sedimentology.

INTRODUCTION

This paper discusses the sedimentology of thepostglacial Quaternary colluvium in westernNorway, based on case studies from two dozenlocalities in the Mre-Romsdal region (Fig. 1).The colluvium comprises arrays of steep, gravellyfans developed along the valley sides and fjordmargins. The depositional processes include awhole range of avalanches, and the paper putsspecial emphasis on snowows, whose mechan-ics and products are little-known to sedimentolo-gists. The paper further focuses on the faciesanatomy, chronostratigraphy and depositionalhistory of the colluvium, with emphasis oncoastal colluvial systems, which represent a

specic type of fan delta. In the nal part, thepalaeoclimatic record of the colluvial successionsis discussed and compared with other types ofregional palaeoclimatic data.

TERMINOLOGY

Colluvium is a general term for clastic slope-wastematerial, typically coarse grained and immature,deposited in the lower part and foot zone of amountain slope or other topographic escarpment,and brought there chiey by sediment-gravityprocesses (Holmes, 1965; Bates & Jackson, 1987;Blikra & Nemec, 1993a). In geomorphologicalliterature, colluvium is also referred to as `talus',

Sedimentology (1998) 45, 909959

1998 International Association of Sedimentologists 909

`scree', `debris slope', `slope-waste deposits' and`hillslope (or hillside) deposits'; bedded col-luvium is often called `stratied debris slope'.Colluvial depositional systems have typically theform of relatively steep and short fans, or `cones',often coalescing into aprons. Modern colluvialfans are readily distinguishable from alluvial fans(Fig. 2), also due to the characteristic products ofavalanche processes.

Avalanches are rapid gravitational movementsof a wet or dry rock debris, snow, or their mixture,occurring on steep slopes. These mass move-ments are also called `catastrophic', because rapidoften means faster than an escaping human canrun and the speed of large avalanches can be ashigh as 5080 m s1 (Voight, 1978). Colluvialavalanches include rockfalls, debrisfalls, debris-ows, snowows, and possibly also some slidesof rock or debris. The individual processes are

dened and discussed further in the text. (Thispaper suggests that the terms `debrisfall' and`debrisow' be written as single words, forconvenience and in semantic analogy to suchterms as rockfall, mudow and snowow.) Adebrisow or a snowow rapidly descending asteep slope is said to be avalanching, because themassow behaviour in such conditions tends tobe different than on a gentle slope.

The term `grainow' refers to a cohesionlessdebrisow, possibly dry, whose movement ischaracterized by pervasive shear and ubiquitousgrain collisions (Bagnold, 1954; Lowe, 1976). Notevery cohesionless debrisow must necessarilybe a grainow (cf. Lowe, 1982), for the phenom-enon of grain collisions the dening feature of agrainow may be of minor volumetricimportance in many such frictional massows(Nemec, 1990b). This notion is particularly true

Fig. 1. Locality map of the Mre-Romsdal region, western Norway, showing the main sites of detailed case studiesand the position of the Scandinavian ice-sheet margin in the Younger Dryas time. The localities are: (1) Gardvik,rstafjorden; (2) Flaskjr; (3) Nordre Vartdal; (4) Tverrelva, Barstadalen; (5) Litlekoppen, Ytre Standal; (6) Try-ttesvora and Seljesvora, Hjrundfjorden; (7) Vellesterdalen; (8) Sunnylvsmoldskreddalen; (9) Korsbrekke, Hellesylt;(10) Fiksdal; (11) Fiksdalstrand; (12) Tomrefjorden; (13) Skorgedalen; (14) Midsund and Stormyr, Oterya; (15)Eikesdalsvatnet; (16) Sunndalsra; and (17) Gravem, Sunndalen.

910 L. H. Blikra and W. Nemec

1998 International Association of Sedimentologists, Sedimentology, 45, 909959

for relatively thick, arenitic sandows, whichoften carry òating' cobbles or boulders. Aspointed out by Middleton & Southard (1978,chapt. 8), `the idea that grainow alone isresponsible for the emplacement of thick, massivebeds of sand can only have been based on amisunderstanding of Bagnold's theory'.

Some authors have considered àvalanche' to bea specic type of mass movement, different fromrockfall, debrisow or slide (e.g. Selby, 1982,1994; Cas & Wright, 1987). However, the deni-tions offered are quite odd, suggesting a mecha-nism analogous to Bagnoldian grainow anddepositional features similar to those of rockfallsand many cohesive debrisows. Likewise, someauthors have oddly restricted the term àvalanche'to snowows only. The reader should also beaware that many colluvial avalanches, includingrockfalls and debrisows, have been describedunder the misleading label `landslide' which ingeomorphology means little more than a massmovement.

The descriptive terminology for gravel charac-teristics, including clast fabric notation, used inthe present study is after Harms et al. (1975) andCollinson & Thompson (1982). The fabric nota-tion uses symbols a and b to denote the long andintermediate axes of the clast, indices (t) and (p)

refer to the axes orientation transverse or parallelto ow direction, respectively, and index (i)indicates axis imbrication.

DEPOSITIONAL SETTING

The Mre-Romsdal region (Fig. 1) is representa-tive of the postglacial development and geomor-phic conditions in western Norway. The regionhas a rugged topography, with mountain peaks ofup to 1800 m, rolling plateaux, glacial cirques,large valleys and deep fjords. The bedrock, chieyPrecambrian gneisses, is strongly fractured andweathered, but has been eroded by glaciers andits primary weathering cover is thus thin andirregular. The NE topographic trend of the mainvalleys and fjords is related to the Mre-Trnde-lag Fault Zone, an early Mesozoic offshore linea-ment that impinges onto the Scandinavian land-mass in the region (Gabrielsen et al., 1984). Otherfjords and valleys, trending NW or NNW, arerelated to an older fault system that has beenreactivated in Mesozoic times and parallels theJan Mayen Fracture Zone (Aanstad et al., 1981).

The region was covered by the Scandinavianice-sheet during the last-glacial (Late Weichsel-ian) maximum, when the oors and slopes of the

Fig. 2. A comparison of the distinctive features of colluvial fans and alluvial fans.

Postglacial colluvium in Norway 911


mountain valleys were mantled with glacial till. Itis uncertain whether the highest peaks protrudedabove the ice-sheet, as nunataks (Sollid & Srbel,1979; Nesje et al., 1987; Follestad, 1990; Larsenet al., 1995). The region was deglaciated, from theouter coast progressively to the south-east, be-tween c. 13 000 and 10 000 years BP (Mangerudet al., 1979; Mangerud, 1980; Fareth, 1987; Larsenet al., 1991). Since then, the glacial till and localkame terraces, perched on the valley sides, havebeen subject to erosion and intense resedimenta-

tion by gravitational processess, resulting inarrays of colluvial fans, solitary (Fig. 3) or co-alescing into aprons (Figs 4 and 5). Many of thesehave prograded into deep-water fjords, formingcolluvial-fan deltas (Figs 3 and 5D).

Studies of the Holocene glaciers in westernNorway have indicated that the postglacial cli-mate uctuated considerably (Nesje & Kvamme,1991; Matthews & Karlen, 1992; Nesje & Dahl,1993; Dahl & Nesje, 1994; Nesje et al., 1995). Forexample, a major climatic deterioration occurrednear the end of the deglaciation phase, in theYounger Dryas time (c. 11 00010 000 years BP),when the shrinking ice-sheet re-advanced (Fig. 1)and local alpine glaciers developed (Reite, 1967;Larsen et al., 1984). The Younger Dryas climatewas cold-oceanic (Larsen et al., 1984) and causeda marked increase in avalanche processes (Blikra& Nemec, 1993a; Nesje et al., 1994a; Blikra &Nesje, 1997).

The present-day annual temperatures at sealevel range between 12 and 14C (July) and 1C(February), averaging 67C. On higher mountainslopes, the mean annual temperatures are lower,ranging between 2C and 4C, respectively. Themean annual precipitation is between 1500 and2200 mm at sea level and even greater at higheraltitudes. The snow accumulation potential isgreatest on mountain slopes facing the SE, E andNE, due to the prevalent winds in these direc-tions. Modern avalanches abound, many reachingthe shoreline and causing considerable damage tothe local vegetation and roads (Lied, 1989), withoccasional loss of human life.

The region has a well-established curve ofpostglacial relative sea-level changes (Svendsen& Mangerud, 1987, 1990), which is used as apalaeogeographic reference framework in thepresent study. The postglacial isostatic uplifthas considerably exceeded the eustatic sea-levelrise, resulting in a relative sea-level fall by 5080 m and gradual emergence of the colluvial-fandeltas (Figs 3 and 5D). Modern human activity,with numerous roadcuts and gravel pits, hascreated good outcrops of the colluvium.

COLLUVIAL PROCESSES AND FACIES

The colluvium is associated with bedrock moun-tain slopes of 35 to 50. The slopes have variedmorphometric proles, with local inclinationsfrom 2530 to as much as 7090. The colluvialfans and aprons that abut these slopes are steep,although their surface inclination depends upon

Fig. 3. Map showing the geomorphic setting and mo-rphometry of two colluvial-fan deltas at Gardvik (lo-cality 1 in Fig. 1). Note that the successive lobes (I-III)of the now-emerged fan deltas were built at progres-sively lower relative sea level. The fan surfaces shownumerous debrisow lobes and levees, and some freshsnowow deposits (asterisks). Letters A-E denote out-crop sections in the gravel pits.



Fig. 4. Map showing the geomor-phic setting and morphometry of asubaerial colluvial apron in NordreVartdal (locality 3 in Fig. 1). Notethat this valley-side apron is an ar-ray of coalesced colluvial fans.

Fig. 5. Examples of colluvial systems in the study area: (A) coalescent colluvial fans dominated by rockfalls, inOterya; (B) a valley-side colluvial apron formed by debrisows, rockfalls and subordinate snowows, at Litlekop-pen; (C) coastal colluvial apron dominated by snowows, in fjord-head area at Korsbrekke; and (D) coalescentcolluvial-fan deltas dominated by snowows, at Eikesdalsvatnet. Photographs from late spring 1992; for localities,see Fig. 1.



the depositional processes. Fans dominated byrockfalls are generally much steeper and shorterthan those formed by debrisows, whereas fansdominated by snowows have intermediate gra-dients. Fan slope depends also upon the preva-lent avalanche size and headwall characteristics(Blikra, 1994). The steepest fans have slopes of4345 (apex) to 2030 (toe) and the least steepfans, 3036 to 515, respectively. Fans domi-nated by waterow processes, or long-runoutsnow avalanches, are no steeper than 1315and morphometrically transitional to commonalluvial fans.

The main depositional processes responsiblefor the development of the colluvial systems aresummarized in Fig. 6 and discussed in thefollowing sections. The review combines generalknowledge with our eld observations and givesthe genetic facies criteria used in the presentstudy.

Rockfall avalanches

Rockfall is a gravitational movement of a mass offragmented bedrock liberated abruptly from a cliff

or steep headwall, whose components tumblefreely downslope by rolling, bouncing and slid-ing; or it may be an analogous movement of asolitary rock fragment. The rock debris is charac-teristically angular, very immature. Large blocksoften disintegrate upon impact, spawning smallerclasts. Rockfalls have been studied extensively(Kent, 1965; Bjerrum & Jrstad, 1968; Carson &Kirkby, 1972; Whalley, 1984; Statham & Francis,1986; Selby, 1994).

In a rockfall, clasts move through a series ofimpacts on the colluvial slope, which may retardthe clast, accelerate it, or stop it instantly. Factorsthat control the runout of a falling clast includeclast size (weight) and shape and the gradient androughness of the slope surface (Parsons & Abra-hams, 1987). Slope roughness is considered interms of the ratio of the diameter of the fallingclast to the diameter of the clasts resting on theslope. Clasts roll down easier on relativelysmooth, low-roughness slopes.

The mobility of a rockfall avalanche dependsalso upon the amount of the debris involved.Colluvial fans formed by larger rockfalls arelonger and less steep (Statham & Francis, 1986).

Fig. 6. Summary of the main depositional processes and facies of colluvial fans/aprons, with special reference to thepostglacial colluvium in western Norway.



In a voluminous rockfall, the interacting claststend to exchange and share their momentum, in amanner similar to that prevailing in a grainow(Bagnold, 1954; Lowe, 1976; Campbell, 1990;Campbell & Brennen, 1983, 1985). However, theconcentration and collision frequency of clasts ina rockfall are lower than in a grainow, whichrenders the larger clasts relatively free to movedownslope and segregate according to sizes(Campbell & Gong, 1986; Savage & Hutter, 1991).The heavier the clast, the greater its momentum,whereby large clasts tend to move faster andoutpace the smaller ones. The tongues of rockfallgravel thus show a marked downslope coarseningand a corresponding increase in surface rough-ness (Figs 6 and 7). The downslope part of arockfall deposit is typically openwork, composedof boulders and large cobbles, often with someblocks scattered further downslope as `out-run-ners'. The upslope part is thinner, comprised ofner gravel and often inltrated with granulesand.

In a rockfall avalanche, the faster-moving largeclasts usually come to rest rst, and the trailingmass of ner debris then overrides this frontaldeposit, resulting in an apparent normal grading(Fig. 6). When a series of rockfall avalanchesdescend the slope along the same track, one afterthe other, the effective runout of the successiveavalanches may be shortened, resulting in anoverall upward coarsening (Statham & Francis,1986; Parsons & Abrahams, 1987; Nemec, 1990b,g. 11).

The role of clast interactions in a rockfallavalanche increases with the volumetric concen-tration. The collisional stress component be-comes dominant when the concentration of clasts

exceeds c. 18 vol.% (Campbell, 1989a), at whichpoint the rockfall regime turns into the shear-owregime of an avalanching grainow (Nemec,1990b, g. 9). This type of dynamic transforma-tion is common in the ner-grained `tails' ofrockfall avalanches (Bolt et al., 1975; Hsu , 1975;Wasson, 1979; Cruden & Hungr, 1986; Grimstad &Nesdal, 1991; Nemec & Kazanc, 1999).

Rockfall debris is often subject to transientaccumulation in a ravine or other slope recess,and subsequently tumbles down as a loose massonce the threshold of frictional yield has beenexceeded. Likewise, rockfall gravel that has ac-cumulated in the apical part of a colluvial fanoften loses stability and avalanches further down-slope as a secondary rockfall, or a debrisow. Thelatter may be a grainow, or a cohesive ow if thefan-head gravel has been illuviated with mud(Nemec & Kazanc, 1999).

Deposits. Colluvial rockfall deposits range fromscattered or randomly clustered boulders andcobbles, to distinct, tongue-shaped beds of im-mature coarse gravel characterized by markedupslope ning, common normal grading andmainly openwork texture (Figs 6 and 7). Theopenwork texture may be well-preserved (Nemec& Kazanc, 1999), but the gravel in the presentcase is more often illuviated with sand and mud(Fig. 8). This secondary inll includes soil mate-rial rich in plant detritus, derived by contempo-raneous slopewash processes. Where emplacedsubaqueously, the rockfall gravel is commonlylled with a wave-derived silt or ne sand,possibly bearing fauna shells or their crumbs.

The clast fabric of rockfall deposits is varied.Boulders and cobbles may show a `rolling' fabric,a(t) or a(t)b(i), when emplaced solitarily or as the

Fig. 7. An openwork, boulderygravel lobe formed by severalmodern rockfall avalanches, withnumerous `out-runner' blocks, up to45 m long, in the forefront; notethe slushow tongue to the left.Valley-side colluvial apron inGlomsdalen. Photograph by P.A.Hole, from summer 1985.



frontal elements of an avalanche on a low-roughness slope. However, rockfall clasts gener-ally tend to adjust their resting positions to thestatic slope debris, are often subject to minorsecondary sliding and are vulnerable to re-orien-tation by subsequent avalanches. Many claststhus show an a(p) or random orientation, and theoverall fabric is disorderly. The fabric in somecases may show weak downslope alignment(Statham, 1973; Perez, 1989; Bertran et al.,1997), and an aligned fabric a(p) or a(p)a(i) iscommon in the rockfall `tails', where the colli-sional stresses often prevail over the free-rollingmotion of the individual clasts.

A typical example of a local colluvial succes-sion relatively rich in rockfall deposits is shownin Fig. 9.

Debrisfall avalanches

Debrisfall (Holmes, 1965) is a process mechani-cally analogous to rockfall, but involving anolder, resedimented gravel, rather than freshlyfragmented bedrock, which means debris that isrelatively mature, subrounded to well-rounded(Fig. 6). Examples include uvioglacial gravelfalling from a perched kame terrace down thevalley-side slope, or a river-derived gravel fallingdown the steep subaqueous slope of a Gilbert-type delta (Nemec, 1990b). The term `rockfall' insuch cases would be inappropriate and mislead-ing. Although debrisfall is not a separate processcategory, its distinction from rockfall may becrucial to a sedimentologist, because the differ-ence in maturity has important implications as to

the debris provenance, mobility and runoutpotential.

In the present case, debrisfall processes arethought to have played a role in the resedimen-tation of mature glacigenic gravel from steepmountain slopes. However, the importance ofdebrisfall deposits in most of the colluvial fansand aprons is minor, compared to that of eitherrockfall or debrisow deposits. The relativescarcity of debrisfalls is attributed to the matrix-rich texture of the glacial till, preventing even thecoarsest gravel from rolling freely down the slope.The few debrisfall deposits that have been recog-nized are thought to indicate the local slopeconditions where the glacial mantle included akame terrace or has been strongly eluviated byslopewash. Apart from the roundness and high

Fig. 8. Detail of a rockfall deposit in vertical outcropsection. The large basal interstices of this immature,openwork gravel have been lled with a stratiedmuddy sand derived by sheetwash. The lens cap(centre) is 5 cm. Colluvial fan at Midsund (locality 14in Fig. 1).

Fig. 9. Portion of outcrop section and detailed log of acolluvial fan dominated by rockfall deposits, at Stor-myr (locality 14 in Fig. 1). The log is normal to thebedding. Letter code in the log: CS clast-supportedgravel texture; OW openwork gravel texture.



apparent maturity of debris, the characteristics ofthese sporadic debrisfall deposits are similar tothose of rockfall deposits (see previous sectionand Fig. 6).

Debrisow avalanches

Debrisows are a common and important agent ofcolluvial sedimentation in a wide range of cli-matic zones (Rapp, 1960, 1963, 1987; Voight,1978; Wasson, 1979; Rapp & Nyberg, 1981, 1988;Keefer & Johnson, 1983; Eisbacher & Clague, 1984;Williams, 1984; Nyberg, 1985; Cruden & Hungr,1986; Van Steijn, 1988, 1996; Blikra et al., 1989;Brabb & Harrod, 1989; Jibson, 1989; Kotarba,1991; Nicoletti & Sorriso-Valvo, 1991; Luckman,1992; Rickenmann & Zimmermann, 1993; Nemec& Kazanc, 1999). In the geomorphological liter-ature, debrisows are referred to also as `debristorrents', `debris slides', `debris streams', `debrisavalanches', `grainows', `earthows', `owingdebris masses', `lahars', `landslides', `mudows',`mudspates', `rock ows', `rocky mudows',`sandows', `sand runs' and possibly othernames. Both subaerial and subaqueous debris-ows have long been a subject of laboratory andeld studies, and the existing knowledge isconsiderable (Johnson, 1970; Hampton, 1975,1979; Lowe, 1976, 1982; Brunsden, 1979; Lawson,1982; McTigue, 1982; Haff, 1983; Savage,1983; Johnson & Rodine, 1984; Nemec & Steel,1984; Campbell & Brennen, 1985; Campbell,1986, 1990; Pierson, 1986; Costa & Wieczorek,1987; Van Steijn & Filippo, 1987; Van Steijn &Coutard, 1989; Savage & Hutter, 1989, 1991;Nemec, 1990b; Takahashi, 1991).

Debrisow is a type of sediment-gravity ow,dened as a gravitational movement of a shearing,highly concentrated, yet relatively mobile, mix-ture of debris and water. In some climatic zones,debrisows may involve an admixture of snow orslush, as is common on the mountain slopes inNorway. Dry debrisows are rare in the presentcase, but common in other climatic settings(Melton, 1965; Oberlander, 1989; Nemec & Ka-zanc, 1999). A debrisow may be turbulent,although a subaqueous turbulent massow is tobe classied as a turbidity current (Lowe, 1982).

In rheological terms, a debrisow is a simple- topure-shear plastic ow, which means that thegranular material tends to spread, behaves like asingle-phase uid and has a nite yield strength.The shear strength derives from a combination ofcohesive forces, due to the electrostatic bondsbetween clay particles, and frictional forces due

to particle interlocking and the frictional resis-tance against interparticle slip (cohesion of clay-free silt and ne sand is very low, highlydependent upon water content and consideredto be negligible; Kezdi, 1979.) Because one of thetwo forces usually predominates, debrisowshave been categorized as cohesive and cohesion-less (frictional), respectively (Nemec & Steel,1984), with the classic mudow and grainowas end-member models (Lowe, 1982). This rheo-logical distinction corresponds with the conven-tional engineering classication of `soils', ornatural clastic materials, into two analogouscategories (Kezdi, 1979; Keedwell, 1984).

When the shear stress, or the downslopecomponent of the material's weight, exceeds theyield strength, the material begins to shear andow, behaving like a dense, viscous uid. Theapparent viscosity of a debrisow is the `bulk'dynamic viscosity of the shearing mass, whichmeans the dynamic resistance to shear derivingfrom the interparticle uid and the sedimentparticles themselves. For example, a grainowbehaves like a high-viscosity uid even when thegranular material is dry (Lowe, 1976; McTigue,1982; Melosh, 1983). In mechanical terms, theapparent viscosity is a dynamic coefcient trans-lating the applied shear stress into the shear-strain rate, and thus controlling the mobility andinternal shear regime of a debrisow, includingthe possible development of turbulence.

The dynamic viscosity of a debrisow is nec-essarily much higher than that of water, but mayvary greatly from one ow to another; it may be ashigh as c. 103 Pa s (Sharp & Nobles, 1953) or aslow as c. 10 Pa s (Pierson, 1986). The apparentviscosity of both cohesive and cohesionless de-brisows varies, depending upon the sedimenttexture and the content of water, or possiblysnow. Further, the viscosity in some debrisowsis roughly constant and independent of the shear-strain rate, like in Newtonian uids, whereas inothers it varies with the strain rate like in non-Newtonian uids. The former case represents theBingham plastic category and applies reasonablywell to cohesive debrisows (Johnson, 1970;Hampton, 1975, 1979; Brunsden, 1979; Johnson& Rodine, 1984), whereas the latter case repre-sents non-Bingham plastics and applies to cohe-sionless debrisows (McTigue, 1982; Haff, 1983;Campbell, 1990), as well as to those waterydebrisows in which the collisional stresses ingravel fraction predominate despite the presenceof a uidal muddy matrix (Bagnold, 1954; Taka-hashi, 1991).



As the frictional debrisow accelerates, itdilates and its apparent viscosity increases. Theviscosity of such a `shear-thickening' debrisowincreases typically as a quadratic function of theshear-strain rate (McTigue, 1982; Campbell &Brennen, 1983, 1985; Haff, 1983; Savage, 1983;Takahashi, 1991). A sandow or sand-rich de-brisow, particularly subaqueous, may com-mence its movement as a `shear-thinning' lique-ed ow, in which the shear strain makes thesand expel its pore water and becomes lubricatedby it, and the apparent viscosity is thus lower athigher strain rate. However, the liquefaction ofloose sand is relatively rapid, and the subaqueousmassow may then: (1) cease to move after a fewseconds or so; (2) accelerate and turn into a high-density turbidity current (Middleton & Southard,1984); or (3) turn into a owslide a debrisowthat glides on a thin, shearing basal layer (Camp-bell, 1989b; Nemec, 1990b; see also Bolt et al.,1975; Lang & Dent, 1983; Lang et al., 1989). Theundrained conditions required for sand liquefac-tion are rare on subaerial mountain slopes, butmany of the cohesive debrisows in our studyarea were probably triggered by a spontaneousliquefaction of the glacial till.

The conceptual rheological framework outlinedabove has been the basis for the interpretation ofdebrisow deposits in the present study. Ourobservations indicate that the colluvial debris-ows, although derived mainly from glacial till,have shown considerable variation. The glaci-genic mantle was likely heterogeneous, and itswater content probably varied greatly on both alocal and temporal basis. The surcial part(c. 1 m) of the mantle was affected by vegetationand weathering, and the debrisow sources alsoincluded the weathered bedrock and upper-slopecolluvium, and possibly local snowpacks.

The debrisow deposits are pebbly to boulderygravel beds ranging from matrix- to clast-support-ed. Matrix varies from a muddy, poorly sorted sandto nearly arenitic sand or sandy granule gravel. Therelationship between the bed thicknesses andmaximum clast sizes (Fig. 10) indicates that thematrix-rich debrisows can generally be classiedas cohesive and the matrix-poor ones as cohesion-less (cf. Nemec & Steel, 1984). The high apparentcompetence of the colluvial debrisows (see theb-values in Fig. 10) can be attributed to two factors.Firstly, the debrisows, particularly those poorerin matrix, are likely to have borne an admixture ofsnow. A dense snow would increase the owcompetence, and as the snow melted, the gravelbed thickness would decrease, resulting in an even

higher maximum clast size/bed thickness (MCS/BTh) ratio (the data in Fig. 10A are, in fact, fromdebrisow deposits of the snowy Younger Dryasperiod.) Secondly, the colluvial debrisows de-scended very steep slopes and were rapid. Thedownslope component of clast weight on a steepslope would be higher than the slope-normalcomponent, which might effectively delay theclast settling and allow the ow to maintainrelatively large clasts over the short descent time.

Furthermore, the geometry of the depositsindicates marked variation in the spreading modeof the debrisows, even on the same colluvial fanor similar fan surfaces. Some debrisows havespread as relatively broad lobes, whereas others

Fig. 10. Relationship between the maximum clast size(MCS) and bed thickness (BTh) of debrisow deposits,based on data from three colluvial fans. Letter symbols:n number of data; r coefcient of linear correla-tion; b regression coefcient (gradient of the least-squares regression line).



have formed highly elongate tongues, passingupslope into furrows with levees (Figs 11, 12 and13). This variation is apparently shown by bothcohesive and cohesionless debrisows, and isthought to represent ows of high and lowviscosity, respectively (Fig. 6). The low-viscositybehaviour is attributed to a relatively high contentof water or lubricating slush, whereas the high-viscosity behaviour is attributed to a low watercontent or an admixture of dense drier snow.

The rheological behaviour of a debrisow isknown to be highly sensitive to relatively minorchanges in water content (Pierson, 1980, 1981,1986; Johnson & Rodine, 1984). When a debris-ow is relatively rich in water and/or slush, itseffective shear strength is considerably reduced.For example, Johnson & Rodine (1984; p. 286)noted that the strength of a debrisow comprising

silty mud and pebble gravel had changed fromc. 320 Pa at a water content of 14 wt.% to c. 40 Paat a water content of 165 wt.%. The shearstrength decreased by an order of magnitude withan increase of water content by merely 25 wt.%.The density of the mixture had changed from2200 to 2130 kg m3, respectively, which is achange of c. 3% only. The driving shear stress ofthe debrisow on a given slope would thus havechanged insignicantly, while the apparent vis-cosity, mobility and internal regime of the owwould be affected quite drastically. On the otherhand, a signicant admixture of a dry or dampsnow, further densied by the load and motionsof the host debris, would render the apparentviscosity of a debrisow very high and result inhigh apparent cohesive strength (Tusima, 1973;Fukue, 1979; Salm, 1982).

Fig. 11. Coalescing debrisow lobeson the surface of a valley-side col-luvial fan in Grasdalen. Note thatsome of the debrisows have spreadas relatively broad lobes, whereasothers have formed elongate ton-gues. The slope catchment at thepresent time is prone to snow ava-lanches, to which the scattered largeblocks on the fan surface are as-cribed. The short dimension of thephotograph is c. 350 m and thegeographical north is downwards.Aerial photograph by FjellangerWidere A/S.



Deposits of high-viscosity debrisows

The deposits attributed to high-viscosity debris-ows (Fig. 6) are beds of matrix-rich to clast-supported gravel, 0417 m thick, which arerelatively extensive and tabular on a gravel-pitoutcrop scale. They contain `oating' cobbles orboulders and commonly show inverse grading.The grading in matrix-supported beds is of`coarse-tail' type, limited to the coarse gravel,and can be attributed to the presence of a `rigidplug' above the shearing lower part of thedebrisow (Johnson, 1970; Naylor, 1980). Thegrading in clast-supported beds is more of a`distribution' type, involving a wider range ofclast sizes, and is attributed to clast collisions,with the upward displacement of larger clasts dueto dispersive pressure (Bagnold, 1954; Lowe,1976; Walton, 1983) combined with kinematicsieving (Middleton, 1970; Scott & Bridgwater,1975; Savage & Lun, 1988; Jullien et al., 1992).The beds of subaqueous debrisows often showalso a weak normal grading at the top (Fig. 14A),probably due to the shear at the ow/waterinterface (Hampton, 1972). The debrisow bedsin Gilbert-type delta foresets tend to be thinnerand have more distinct boundaries (Fig. 15). TheMCS/BTh data (Fig. 16) indicate that the compe-tence of subaqueously emplaced debrisows wasgenerally lower, probably because these moremobile ows were more watery or incorporatedwater when plunging into the sea (cf. Nemec &Steel, 1984 g. 23).

The lobate geometry of the debrisow depositsof this category (Fig. 6) is most apparent on thefresh surfaces of many colluvial fans (Fig. 11).Some of the debrisow lobes can be traced nearlyto the fan apex, where they are thinner, muchnarrower and often have bouldery levees. Otherlobes are more `detached' from the apex, pinchingout upslope in a mid-fan zone. The latter debris-ows have apparently bypassed the steepest fanslope with little or no deposition. Their bypasstracks are shallow scours, with or without levees.Another trace of a bypassing debrisow are`debris horns': cusps of debrisow material accre-ted onto the upslope sides of large blocks pro-truding above the fan surface. The high-reliefobstacles apparently cause local plastic `freezing'of a bypassing debrisow, if not too watery.

The clast fabric in the debrisow beds of thiscategory is mainly a(p) or a(p)a(i), at least for thelarger clasts (Figs 14 and 15). The steep fronts ofmodern debrisow lobes commonly show an a(t)fabric, with large boulders that have apparently

Fig. 12. Fresh deposits of low-viscosity, slush-rich de-brisows on vegetated colluvial aprons in Norangs-dalen (near locality 8 in Fig. 1). (A) Note the `digitated'shape of this elongate lobe, due to the lateral splays; theslope height is c. 100 m. (B) Note the elongate, tongue-like geometry of this lobe, with a relatively thick,coarser `head' and thinner upslope `tail'; the latterpasses upslope into an erosional furrow with levees;the slope height is c. 50 m.



rolled down the front or been `bulldozed' by it.These frontal features, however, are seldomrecognizable in the outcrop sections of olderdebrisow deposits. Likewise, the individualbeds show little or no downslope change in clastsize within the limits of a large gravel-pit outcrop,although the deposits of apparently similar deb-risows on modern fan surfaces often showdownslope thinning and ning (Mears, 1980;Statham & Francis, 1986). It is possible that thelatter trend occurs on a lateral scale larger thanthat of a gravel pit, for the debrisow beds are, infact, often thinner and ner grained (in terms ofMCS) in the downfan outcrops. However, thisapparent trend may reect the greater spreading(smaller thickness) and lesser competence of themore mobile debrisows.

Deposits of low-viscosity debrisows

The deposits attributed to low-viscosity debris-ows (Fig. 6) typically include an erosive, fur-row-like track with levees and a highly elongate,

coarse-fronted gravel lobe in the downslopedepositional zone (Figs 12, 13, 14B, 17 and 18).The watery or slushy debrisows are more sensi-tive to the slope topography, and their tracksoften show bends. The tongue-shaped gravellobes are characterized by a relatively thick,clast-supported, bouldery to cobbly `head' thatpasses upslope into a thinner, ne cobbly topebbly, clast- to matrix-supported `tail'. There islittle visible difference between the subaerial andsubaqueous deposits of this category. In the ow-parallel outcrop sections, these debrisow lobesare recognizable as long, but highly asymmetricallenses, either solitary or multiple. The multiplelenses are often stacked upon one another in an`imbricate' fashion, dipping upslope, or show amore complex style of stacking, with the succes-sive debrisow lobes having overridden andoverstepped one another to a variable extent(Figs 6 and 14B). The head of the lobe typicallyshows a tractional a(t) fabric (Figs 12B and 18),whereas the fabric in the tail is more varied; a`shear' fabric a(p) or a(p)a(i) is common, but

Fig. 13. Fresh deposits of slush-richdebrisows and snowows on veg-etated colluvial aprons in Norangs-dalen. (A) Note the `digitated' lobeof a slushy debrisow to the left; theleveed narrow tracks and small,`digitated' debris lobes of slushowsto the right; and the scattered and`patchy' debris deposited by snow-ows in the lower middle; this col-luvial apron's height is c. 50 m. (B)Frontal part of a lateral `spill-over'lobe of a watery debrisow, showingmainly `rolling' clast fabric andcommon imbrication (ow towardsthe viewer); the main lobe hasspread to the right (see in the back-ground).



`rolling' fabric occurs as well, particularly ingravel beds showing normal grading.

Our observations from active fans support thenotion that the watery debrisows are triggeredby a rapid melting of slope snowpack (Harris &Gustafson, 1988) or episodes of heavy rainfall(Caine, 1980; Lawson, 1982). The reported veloc-ities of such debrisows are of the order of a fewmetres per second. As the watery debrisowdescends a steep slope, the ow accelerates anddilates rapidly, the dispersion of its larger clastsincreases and its two component phases thecoarse gravel and the uidal matrix tend tobehave almost independently. The dynamic re-gime of the avalanching debrisow thus ap-proaches that of a debrisfall (Campbell, 1989a;

Nemec, 1990b, g. 9), with the clast-size distri-bution and the mean free distance between thelarge clasts becoming important controlling pa-rameters. The large clasts tend to move down-slope according to their own momentum, pushingthemselves, or `streaming' in engineering par-lance, through the uidal ner material to thefront of the ow (Suwa, 1988; Nemec, 1990b). Abouldery to cobbly `head' develops, whose grow-ing thickness, increasing clast concentration andgreater internal friction render its speed lowerthan that of the following `tail'. The latter isusually turbulent, so long as it retains an abund-ant interstitial uid; otherwise, the apparentviscosity rises and suppresses the turbulence.The faster-moving tail continually feeds the head

Fig. 14. Debrisow deposits in out-crop sections. (A) Portion of rela-tively thick, tabular bed of unsorted,matrix-supported gravel attributedto a high-viscosity debrisow em-placed on the subaqueous slope of aconical colluvial-fan delta; note thecoarse-tail inverse to normal grad-ing, with `oating' cobbles andboulders in the middle part. (B)`Imbricate', lenticular gravel bedsattributed to low-viscosity debris-ows emplaced on the subaqueousslope of a Gilbert-type fan delta;note that the debrisow tongueshave clast-supported, bouldery tocobbly `heads' and a thinner, peb-bly, clast-to matrix-supported `tails';note also the thicker, more tabulardeposits of high-viscosity debris-ows in the lower part. The whiteruler is 23 cm. Examples from thecolluvial fan-delta complex atTomrefjorden (locality 12 in Fig. 1).



with the lubricating uid, which percolates intothe frontal part due to gravity (in conditionspromoting such debrisows, the colluvial slope isusually water-logged; otherwise, the uid wouldescape into the porous substratum.) The percola-tion rate depends upon the size distribution ofdebris, especially its ner fractions. If the perco-lation is not too rapid and the head lubrication iseffective, the debrisow may arrive on the lowerslope without much change in the internalregime, and the runout may be considerable.When the head of the ow spreads and comes to ahalt, its nal inlling by the watery tail occurs,often with a thin sheet or multiple ngers ofpebbly sand deposited in the forefront by theescaping turbid water. The fronts of slushy

debrisows are more `digitated' due to thepostdepositional melting and secondary mobili-zation (Figs 12 and 13).

The onset of turbulence in a cohesive debris-ow, approximated as a Bingham plastic, occurswhen the Hampton number of the ow exceeds1000, which may mean ow velocity in excess ofc. 3 m s1 (Middleton & Southard, 1984). Suchvelocities can be attained by a debrisow, even ifbriey, on the steep colluvial slope. The morewatery debrisows have lower viscosities andbecome turbulent much easier, but they are alsogenerally thinner. Vigorous turbulent churningmay occur when the Froude number of the ow

Fig. 15. Details of the subaqueous foreset of Gilbert-type fan delta, Tomrefjorden (locality 12 in Fig. 1). (A)Diamict, mud-rich bed with `oating' cobbles andboulders, attributed to cohesive, high-viscosity debris-ow. (B) Two beds of matrix-supported gravel,attributed to high-viscosity debrisows, underlain by aclast-supported gravel unit attributed to subaqueouslyemplaced, debris-rich snowows; note the coarse-tailinverse grading in the lower (lighter-shade) debrisowbed. Transport direction is to the left. The beds areinclined at c. 25, and the photographs are parallel tobedding. The ruler is 23 cm.

Fig. 16. Relationship between the maximum clast size(MCS) and bed thickness (BTh) of the subaerial andsubaqueous debrisow deposits in a colluvial-fan deltaat Eikesdalsvatnet (locality 15 in Fig. 1). Letter sym-bols: n number of data; r coefcient oflinear correlation; b regression coefcient. Note thatthe b-value in the lower diagram is c. 25% smaller andthe regression line intersects the MCS axis near the zeropoint, which suggests less competent and cohesionlessdebrisows (for methodological details, see Nemec &Steel, 1984).



signicantly exceeds 1 (Pierson, 1986), and suchsupercritical conditions can easily be reached ona steep slope. However, a subaerial debrisow isunlikely to become fully turbulent, like a New-

tonian uid, with the large eddies breakingcontinually down into progressively smallervortices. Unlike powder snowows and somesubaqueous debrisows, the subaerial debris-ows do not burst into fully turbulent owsdownslope. A watery subaerial debrisow 115 m thick may show vigorous turbulence, butcan hardly suspend clasts larger than smallpebbles (Pierson, 1986), due to the limited scaleand highly dissipative character of the eddies.The turbulence may be mixing the debris con-tinuously, but the larger clasts are lifted less,settle faster and segregate from the small ones.The signatures of turbulence in a debrisowdeposit are thus normal grading and possibly a`rolling' fabric, which are features commonlyrecognizable in the tails of the colluvial debris-ows. It should be noted that subaqueousdeposits with such features are to be classiedas deposits of high-density turbidity current of(Lowe, 1982).

When the uidal matrix of the debrisowpercolates into the head and/or substratum toorapidly and the slope is very steep, the tail turnsinto a frictional debrisow, possibly still slightlyturbulent, where Bagnoldian dispersive stressesmay prevail. The energy and frequency of clastscollisions (or the `granular temperature' of theow) may be high in a fast-moving, levee-con-ned ow, whereby the coarser clasts will tend tobe pushed inwards and upwards, and be furthertransferred to the head of the ow in a conveyor-belt fashion (Hirano & Iwamoto, 1981; Johnson &Rodine, 1984; Pierson, 1986; Takahashi, 1991).The tail gravel will be clast-supported, inverselygraded and have an a(p) or a(p)a(i) fabric. Thehead gravel will have an a(t) fabric in the frontaland upper part.

Fig. 17. Elongate, tongue-shapedgravel lobes passing upslope intoleveed erosive furrows, formed bywatery debrisows on a valley-sidecolluvial apron in Liadalsdalen(south of locality 2 in Fig. 1); theupslope extent of the avalanchetracks corresponds to the upperlimit of the mountain slope's glaci-genic mantle. The width of the pic-ture is c. 250 m. Photograph by O.Longva, from early summer 1991.

Fig. 18. Fresh deposit of a low-viscosity debrisow onthe surface of a colluvial fan in Glomsdalen (upslopeview). The tongue-shaped lobe has a bouldery frontwith transported tree logs, and passes upslope into anarrow, gravel-lined furrow with levees. Photograph byP.A. Hole, from summer 1985.



Some debrisow tails have a matrix-supported,nearly bimodal texture, though otherwise showsimilar characteristics as mentioned above. Thematrix is a poorly sorted sand, often quite muddy.The deposit shows coarse-tail inverse grading andfrontal concentration of large clasts, which hasbeen attributed to a combination of dispersivepressure and the conveyor-belt mechanism (John-son & Rodine, 1984; Takahashi, 1991). However,Bagnold's (1954) theory predicts that the colli-sions of sand grains result in relatively smalldispersive pressure, and the latter also tends to besuppressed by cohesive forces. It is more likelythat the large clasts in a shearing sand move awayfrom the margins and lower boundary of theconned debrisow due to the so-called Magnuseffect: a lift force acting normal to the shearboundary and related to the rotation of the largeclasts by the local shear-velocity gradient. Thiseffect is thought to occur, for example, in somesnowow avalanches (Hopnger, 1983).

The low-viscosity colluvial debrisows arecomparable to the watery debrisows describedfrom mountain ravines and narrow gorges (Okudaet al., 1980; Suwa & Okuda, 1983; Johnson &Rodine, 1984; Pierson, 1980, 1986; Takahashi,1991). However, the elongate debrisow lobes inthe present case have been deposited on opencolluvial slopes of 30 to 515, whereas theravine-conned debrisows have reportedly beenturbulent and erosive on slopes as gentle as 58.This difference may have two reasons: (1) thecolluvial debrisows are due to slumping ofwater-logged sediment, rather than to sedimententrainment by ood-water surges; and (2) themechanism of ow-head lubrication by uidalmatrix may be less effective, or shorter-lived, onvery steep, rough and porous colluvial slopes.

The debrisow tracks are erosive furrows(Figs 12, 13, 17 and 18), widening and shallow-ing-out in the downslope direction. Their widths,excluding levees, are several times their depth.The depth decreases from 50 to 70 cm in theupper reaches to less than 510 cm in the apicalpart of the associated gravel lobe, which is mainlynon-erosive. The furrows are probably scoured byturbulence, but their lower reaches may be ruts ofnonturbulent massow. The levees are due to theplastic `freezing' of the lateral margins of the ow,where the shear rate is at a minimum (Johnson &Rodine, 1984). These `dead zones' include cob-bles and boulders that have been shoulderedaside by the debrisow front (Pierson, 1986).

The morphology of modern deposits shows thatthe turbulent tail of a low-viscosity debrisow

may locally breach the levee and form a smaller,secondary lobe of gravel outside the main track,particularly at a local bend (Figs 6, 12 and 13).The debouching may deplete the parental debris-ow of its lubricating uidal part and reduce therunout. Our observations further indicate thatwhen the head of a debrisow `freezes' and jamsthe steep descent path, the tail may movesideways, develop a new head and form asecondary lobe further downslope; or the tailmay override the halted head and form a new lobedirectly in front of it. These secondary lobes aregravelly and their heads may be nearly as coarseas the primary ones (Fig. 13B). When formed on alower slope, at a relatively late stage of debrisowevolution, the secondary lobe is relatively thin,composed of sandy pebble gravel, and its head ispoorly developed. These characteristics may helpto distinguish between the primary and second-ary lobes of debrisows, although the distinctionin outcrop sections is not easy.

The subaqueous foresets of colluvial-fan deltasabound in deposits of both high- and low-viscos-ity debrisows, associated with the graded bedsof sand and pebbly sand attributed to high-density turbidity currents. Turbidity currentscan be generated by subaerial debrisow ava-lanches plunging into the water (Hampton, 1972;Weirich, 1989). However, the turbidites are muchmore abundant in the foresets of Gilbert-type fandeltas, whose formation involved streamowprocesses, which suggests that the latter wereprobably more important in generating turbiditycurrents, by hyperpycnal outow or mouth-barcollapsing (Nemec, 1990b).

Some illustrative examples of local colluvialsuccessions relatively rich in debrisow depositsare shown in Figs 19 and 20.

Snowow avalanches

Snowows (Fig. 5C,D), also referred to as `snowavalanches', have been a subject of intenseresearch by glaciologists, geomorphologists andengineers, primarily because these avalanches area serious hazard in mountainous terrains in manyparts of the world. Surprisingly, snowows havedrawn little sedimentological interest, althoughthey are known to carry often abundant rockdebris.

The literature on snow avalanches is extensive,but widely scattered in the glaciological andtechnical journals, special reports and symposiavolumes. There are several monographs (Mantis,1951; Kingery, 1963; Oura, 1967; Perla &



Martinelli, 1975; Voitkovskiy, 1977; Fukue, 1979;Washburn, 1979; Colbeck, 1980; Glen et al., 1980;Gray & Male, 1981; Ramsli, 1981); a number ofreviews (Luckman, 1971, 1977, 1978; LaChapelle,1977; Mellor, 1978; Perla, 1980; Schaerer, 1981;Salm, 1982; Hopnger, 1983); and a wealth ofempirical studies (e.g. Dent & Lang, 1980, 1982;Lied & Bakkehi, 1980; Mears, 1980; Narita, 1980;

Schaerer & Salway, 1980; Gubler, 1982; Lang &Dent, 1983; McClung & Schaerer, 1983, 1985;Lang et al., 1985; Salm & Gubler, 1985; Hutteret al., 1989; Hermann & Hutter, 1991; McClung &Tweedy, 1993) and various theoretical consider-ations (e.g. Perla et al., 1980; Bakkehi et al.,1981, 1983; Dent & Lang, 1983; Norem et al.,1985, 1987, 1989; McClung & Lied, 1988; Gubler,1989; Gubler & Bader, 1989; Lackinger, 1989;Nishimura & Maeno, 1989; Nishimura et al.,1989). Some publications are focused specicallyon slushows (Washburn & Goldthwaite, 1958;

Fig. 19. Portion of an outcrop section and detailed logof the subaerial part of a fan delta rich in debrisowdeposits, Tomrefjorden (locality 12 in Fig. 1). Thedownslope direction is to the left, and the log is normalto bedding.

Fig. 20. Portion of an outcrop section and detailed logof the subaerial part of a colluvial-fan delta showing analternation of debrisow and snowow deposits,Flaskjr (locality 2 in Fig. 1). The downslope directionis to the right, and the log is normal to bedding. Notethe radiocarbon dates in the log. Texture code CS andOW as in the caption to Fig. 9.



Nobles, 1966; Washburn, 1979; Hestnes, 1985;Nyberg, 1985; Conway & Raymond, 1993).

The literature on snowows seems to exceedthat on debrisows, but bears less consensus. Thescientic language of the publications varies fromthe formal terminology of physics to the descrip-tive parlance of geomorphology, and there seemsto be relatively little exchange of results and ideasbetween the two respective groups of researchers.The following short review attempts to synthesizethe diverse views and classications into areasonably simple and coherent conceptualframework, which has been the basis of our studyof colluvial snowow deposits. The review refersto the classication in Fig. 21 and focuses onthose aspects of snowow processes that arerelevant to a sedimentologist.

Snow rheology

Snow is a peculiar material, with a highlycomplicated rheological behaviour, althoughsnowows are generally akin to sediment-gravityows. On a microscopic scale, snow consists of aframework of ice particles, which may be singlecrystals or their larger aggregates. The mechanicalproperties of snow vary with the particle charac-teristics (size, shape, packing), interparticlebonds (number and type of contacts, pore-wallarea) and the spatial arrangement of bondedparticles (fabric). Snowpacks are commonly lay-

ered (bedded) and texturally heterogeneous dueto their incremental accumulation under variableweather conditions. A fresh powdery snow maycontain c. 90 vol.% air, whereas an older, com-pacted and recrystallized snow may be nearly asdense as ice.

In broad terms, snow can be regarded as aplastic material, whose shear strength ranges frommainly frictional to mainly cohesive. However,snow has a high and irreversible compressibilityand an equally high thermodynamic instability.The shear deformation of snow is volumetric,rather than purely deviatoric (Salm, 1982), andthe content of free water depends on the ambienttemperature (Ambach & Howorka, 1966), whichrenders snow rheologically quite different fromall natural clastic materials.

Snow has a high capacity to dissipate stress, bydensication. For example, the compaction ofsnow from an initial density of 300 kg m3 to adensity of 900 kg m3 absorbs three times asmuch energy as its compaction from 600 to900 kg m3 (Brown, 1980a). Further, the contentof water is very important. The free water spreadsthrough the pores by capillary forces, but maybegin to percolate through the snow, due togravity, when the saturation exceeds c. 1 vol.%(Langham, 1981). At low, `pendular' saturations(< c. 7 vol.%), the interparticle water lm isdiscontinuous, the snow abounds in air bubblesand is relatively strong, as is common with early

Fig. 21. Tentative classication ofthe relevant owing media in termsof the relative proportion of rockdebris, snow and water involved.The terminology pertains to subae-rial ows. For subaqueous condi-tions, the solid-laden, high-densitywaterow and the turbulent waterydebrisow would be classied asturbidity currents. The classboundaries are based partly onBeverage & Culbertson (1964) andFukue (1979).



spring snowpacks (Male, 1980). At higher, `fun-icular' saturations (> c. 712 vol.%), the waterlm becomes continuous and the snow is rela-tively weak. When the water saturation exceedsc. 25 vol.%, the snow turns into slush (Fig. 21).Saturations of up to 3550 vol.% have beenreported from watery slushows (Nobles, 1966;Mellor, 1978). In undrained laboratory tests, slushreaches a maximum water saturation of 58 vol.%(Nyberg, 1985).

Frictional interparticle forces dominate in freshpowdery snow and dry granular snow, althoughcohesive forces are also active in the former:down to a temperature of 5C in dry conditionsand down to 30C in regelational `wet' condi-tions; below these temperatures, a granular snowdevelops in either case. Cohesive forces predom-inate in damp/wet snow, due to the adhesiveeffect of free water; in semibonded snow, due tothe interparticle water lms and crystallinebonds; and in bonded (sintered) snow, due tothe crystalline bonds alone. Sintering occurs attemperatures below the melting point, but thebonds form faster as the temperature nears thelatter (Fukue, 1979). The shear strength variesgreatly with the snow density, from c. 102 Pa for avery low-density snow (q 100 kg m3) to c. 106

Pa for a high-density snow (q 600 kg m3). Thedensity of a slushy or densied snow may reach c.700 kg m3, and a strongly sintered snow ap-proaches the density of ice (917 kg m3). How-ever, a high water content renders snow weaker(Salm, 1982), hence the strength of slush is verylow.

As the cohesive strength prevails in some snowvarieties and the frictional strength dominates inothers, it has been suggested that snowows beclassied into cohesive and cohesionless (orfrictional) (Nyberg, 1985). Notably, this rheolog-ical distinction is similar to the classicationsuggested for debrisows (Nemec & Steel, 1984).

The static angle of internal friction for snow ishigh, typically in the range of 3546, and also thedynamic angle of friction for a owing loose snowis high, 2532 (Fukue, 1979), comparable to theangle for classic grainows (Bagnold, 1954).However, the dynamic angle depends stronglyupon the shear-strain rate (Fukue, 1979). Thedynamic viscosity increases almost exponentiallywith the density, much stronger for dry than fordamp or wet snow (Salm, 1982, g. 15). Forexample, laboratory tests show that the dynamicviscosity of a wet snow increases from c. 103 toc. 106 Pa s due to a change in density from 700 to900 kg m3, whereas a dry snow shows a similar

change in viscosity due to a density increase ofless than 10 kg m3. Rapid rebonding of particlesoccurs in the failing framework of dry snow,whereas extensive bond slips and particle growthoccur in wet snow. The rate of particle growth atan ambient temperature of + 20C is nearly 50%higher than at 0C, and the increasing size anddecreasing number of the particles and bondsrender wet snow considerably weaker (Salm,1982). Further, the effective stress at the particlecontacts in wet snow is signicantly reduced bythe repulsive electrostatic forces due to thedouble-charged layers at each solidliquid inter-face of the water lm separating the particles.These repulsive forces are stronger at low watersalinities (Colbeck et al., 1978).

The constitutive relationship between the shearstress and shear-strain rate for snow is generallyconsidered to be nonlinear, with a rate-dependentapparent viscosity, but depends strongly uponthe strain rate itself (Mellor, 1978; Fukue, 1979;Salm & Gubler, 1985). The relationship is nearlylinear at very high rates (Fukue, 1979), particu-larly for powdery and granular snows, which isprobably why turbulent snowows are akin toturbidity currents. A dry loose snow at high strainrate shows shear-softening and strong dilation,after the initial collapse of the particle framework(McClung, 1979; Narita, 1980). At lower strainrates, the relationship varies. At a very low rate,snow creeps due to the rupture of interparticlebonds and the localized melting and slip withinthe ice crystals and their aggregates. The creep isoften approximated as a linear viscous deforma-tion (Brown & Lang, 1973; Salm, 1982), althoughthe strain rate decreases with depth, roughlylinearly, as the viscosity increases linearly withthe normal stress (Desrues et al., 1980; McClung,1980; Watanabe, 1980). The constitutive relation-ship is nonlinear once the snow mass hasdetached itself and ows at a low rate. Theviscosity then increases almost exponentiallywith the strain rate, much faster than the snow'sYoung modulus, whose increase is nearly linear(Kry, 1975). This is the state of `brittle' owage(Fukue, 1979), with the strain breaking andthoroughly reworking the particle framework.Importantly, the shear strain does not dilate theloose snow, but rather increases its density. Thesnow thus densies, yet its rheological behaviouris similar to that of dilational grainow (LaCha-pelle, 1977). The shear stress/rate relationshiphas been modelled as a third-order dilation,proportional to the cube of the rate (Brown &Lang, 1973).



The apparent viscosity increases with the strainrate until the microstructure of the snow has beenpervasively reworked. The viscosity then shifts toa lower level and remains approximately con-stant, almost like in a Newtonian uid. This is thestate of `ductile' owage (Fukue, 1979), whichcauses snow hardening in laboratory shear-boxtests. On a mountain slope, the owing snowmass at this point accelerates and its furtherbehaviour may be quite different. For example,turbulence may be ignited in a dry snowow,which will then rapidly dilate into suspension.Unfortunately, the available empirical data aresparse and often inconclusive, and the existingconstitutive models show large discrepancies(Voellmy, 1955; Mellor, 1978; Salm & Gubler,1985).

For example, Mellor (1978) reported that theshear rate in a purely deviatoric (nonvolumetric)snowow increased rapidly with the shear stress,which would imply a decreasing viscosity andthus a contractional non-Newtonian behaviour.However, the velocity proles measured by Salm& Gubler (1985) and Nishimura & Maeno (1989)indicate that the apparent viscosity of relativelydense snowows remains constant, as in a Bing-ham plastic ow, but increases (proportionally tothe second power or so of the shear rate) in fasterows with greater particle dispersion. This dila-tional non-Newtonian behaviour would be akin tothat of a grainow.

It would thus appear that the rheologicalbehaviour of snow varies greatly, depending onthe snow type and the shear stress level. There isno universal constitutive equation for snow, butrather equations for its behaviour under particularconditions (Salm, 1982). As a convenient simpli-cation, the ow of a relatively dense snow(> 300 kg m3) is often considered to be contrac-tional (`shear-thinning'), and the ow of a lower-density snow to be dilational (`shear-thickening'),although these behaviours may change or combinein an evolving snowow (Brown, 1979, 1980b).

Snowow types

As snow accumulates on a mountain slope, thenormal stress in the snowpack increases with theincreasing thickness, or load. Concurrently, theyield strength of the snowpack increases, due todensication. Strong winds may increase thesnowpack density, and a wind-blown snow isalmost invariably denser than one deposited byfreefall. When the stress increases faster than thestrength, a gravitational failure occurs. Other

triggering mechanisms include falling cornicesand snowpack destabilization by creep. The winddrift of snow is very important, for it affects thesnowpack thickness, builds cornices on the leeside of mountain ridges, and lls the mountainravines and gullies with thick snowpacks thatmay later obstruct runoff.

Snow typically fails on slopes of 3040, but awet snow can fail on a surface of 10, if relativelysmooth or icy. The snowpack may slough by aseries of small retrogressive failures, or collapseen masse as an avalanche (Perla, 1978). Twotypes of snowow avalanche can be distin-guished, with possible intermediate varieties(Hopnger, 1983).

Dense snowows. Cohesive or other densesnow is usually subject to a line failure, leadingto a translational slide. A broad slab of shearingsnow then descends the slope, leaving well-dened crown and ank scarps in the snowpack.The result is a dense snowow, also called`owing avalanche [Ger. Fliesslawine]' or `granu-lar-snow gravity ow' (Hopnger, 1983). Thesesnowows may be cohesive or cohesionless, andrange from pseudolaminar to turbulent. The snowinvolved may be a dry granular or sintered snow,a damp granular or semibonded snow, a wetgranular snow, or a slush. A hard snow slab(sintered or semibonded) may resist fragmenta-tion, and the avalanche may carry large snowblocks to the depositional area (Fig. 22C). If fullyfragmented, the avalanche will consist of gravel-sized hard clods mixed with a granular orpowdery snow (Fig. 22B, 3A). Cohesive snowallows `snowrollers' (snow balls) to form andbreak repeatedly.

Dense snowows are comparable to debris-ows, and are rheologically modelled as cohesiveor cohesionless plastics (Salm & Gubler, 1985;Lang et al., 1989; Nishimura & Maeno, 1989), or aviscoelastic material (Voellmy, 1955; Mellor,1978; Salm, 1982) in the case of very slowmovement. Their steep-fronted depositional lobesand leveed tracks (Figs 22A & B, 23A and 24) aresimilar to those of debrisows. In fact, slushowwas originally dened as `a mudow-like owageof water-saturated snow' (Washburn & Goldth-waite, 1958). However, the snowows have shear-dependent viscosity and behave more like non-Bingham plastics (Lang & Dent, 1983; Lang et al.,1985). In a ow of roughly constant density (say,c. 400 kg m3), the snow hardness may range overan order of magnitude and the kinematic viscositymay vary from c. 400 to c. 2500 m2 s1 (Maenoet al., 1980; Lang et al., 1985).



The dense snowows on mountain slopes havetypical velocities of 1030 m s1 and are mainlysupercritical (Fr > 1). The mean density rangesfrom c. 150 kg m3 for damp snow to c. 350 kg m3

for wet or hard snow (Mellor, 1978), and the snowparticle concentration is typically in the range of3344 vol.%. These snowows are `groundavalanches' (Hopnger, 1983), with the lowerboundary cut at the ground surface or deeply inthe snowpack.

Dense snowows often arrive on the lowerslope by turning into owslides: the ow `freez-es' plastically from the top downwards, but

keeps gliding on a thin, shearing basal layer;and when the front brakes, the snow massdeforms by means of discrete listric thrusts andlongitudinal shear zones (Salm & Gubler, 1985).The resulting longitudinal and transverse shears/ridges (Fig. 22D, 23A; Mears, 1980, gs 6,7 & 8)resemble those observed in debrisows that havecompleted their movement in a sliding fashion(Nemec, 1990b, gs 28, 29 & 30). Slushow canmove on a smooth slope of less than 2 (Nobles,1966), but stop on a rough, bouldery slope ofmore than 70 (Luckman, 1971) (Fig. 24).

Fig. 22. Features of fresh snowow deposits. (A) Dense snowow lobes on a colluvial fan at Gravem (locality 17 inFig. 1), spring 1993. (B) Snowow deposit composed of pebble- to boulder-sized, hard snow clods mixed with apowdery snow entrained en route by the avalanche; note the levee to the right (downslope view, shovel for scale);Gravem, spring 1993. (C) Large, hard snowballs at the surface of a melting snowow lobe; Gravem, spring 1993.(D) Longitudinal shear zone, rich in meltout debris, in the lateral part of a dense snowow lobe; Hjrundfjorden(locality 6 in Fig. 1), spring 1988.



Powder snowows. A fresh, powdery snow or aloose older snow that has lost cohesion (drygranular snow) is usually subject to a pointfailure, which quickly expands. The term `pow-der snowow' is used here broadly for such low-density snowows, composed partly or entirely ofpowdery snow and most often turbulent. Thesesnowows come to rest on slopes ranging from34 to horizontal, but most commonly on those of1728 (McClung & Schaerer, 1983). The turbu-lent snowows, also referred to as `airborne snowavalanches' (Hopnger, 1983), have generallylonger runouts.

A turbulent cloud typically develops at the topof a powder snowow and grows at the expenseof the latter. The entrainment of dry snow by theambient air begins when their relative velocityreaches c. 7 m s1, and the turbulent cloudbecomes fully developed when the velocity ex-eeds c. 10 m s1. The trailing cloud acceleratesand turns into an overriding surge, which mayattain speeds of 3070 m s1, possibly up to125 m s1, thus outrunning and overtaking thedepleted parental ow. The thickness of theresulting turbulent avalanche may vary from lessthan 10 m to more than 100 m, depending on thesnow volume involved and the travel distance.The density may vary from c. 1020 kg m3 in thelower part to little more than 12 kg m3 at thetop of the ow (Mellor, 1978).

Many powder snowows burst into a ground-hugging, highly turbulent cloud shortly after theirrelease, probably due to an internal shock wave(Hopnger, 1983). For a snowow with meanvelocity U and thickness h, the kinematic wavethat travels through the ow has a celerityuk 15 U, whereas the corresponding long dy-namic wave has a celerity ud (qh)1/2 and effec-tively propagates with a velocity ud* U + (qh)1/2where q is the acceleration due to gravity. When theFroude number of a ow, Fr U/(qh)1/2, exceeds2, the kinematic wave moves faster than thedynamic wave, which may generate a `roll-wave',or growing shock bore. A velocity of 10 m s1, towhich an avalanche can acceleratevery quickly on asteep slope, corresponds to Fr > 2. Ultra-rapid owregime, with Froude numbers as high as 685, canbe reached by large powdersnowowson very steepslopes (Schaerer & Salway, 1980).

A turbulent powder snowow is typically a`surface avalanche' (Hopnger, 1983), whoselower boundary is cut shallowly in the snow-pack, rather than at the ground surface. Theseows are akin to turbidity currents (Hopnger,1983, g. 4), and are often modelled as such.They have mean frontal velocities of 1260 m s1,more typically 2040 m s1, and mean densitiesof 50300 kg m3 (Mellor, 1978; McClung &Schaerer, 1983). The density in the basal partmay be 24 times greater than the mean. Themaximum velocity is near the base, at a height ofroughly 1/10 of the ow thickness. For a owthickness h and a typical range of slope inclina-tions of 60 to 30, the maximum velocity zone istheoretically at a height of 21h1/2 above the base(McClung & Schaerer, 1983, 1985). The maximumvelocity of the ow body is 225 times the frontalvelocity. The concentration of snow is low,

Fig. 23. Features of fresh snowow deposits on a col-luvial fan at Bndergjerde, Skorgedalen (locality 13 inFig. 1). (A) Upslope view of the upper part of a densesnowow lobe, showing large axial furrow, levees andtransverse listric shears; the furrow was scoured byyounger, bypassing snowows. The backpack to the leftis c. 50 cm long. (B) Rock boulders, an uprooted birchtree, numerous broken branches and abundant nerdebris melting out at the surface of a dense snowowlobe; the lobe at this stage is still c. 3 m thick. The lenscap is 5 cm. Photographs from late spring 1993.



typically 525 vol.%, which renders internalfriction very low. For example, the dynamicangle of friction for a concentration of 5 vol.%is less than 1, and such a snowow will behavelike a Newtonian uid.

Some powder snowows are non-turbulent.They are typically mixtures of powdery snowand an older granular snow, and are akin tocohesionless debrisows. Gubler (1989) has com-pared the existing models for dry non-turbulentsnowows and concluded that the Salm & Gubler(1985) model of `partly uidized' snowow pre-dicts the observed velocities and runouts mostcorrectly. The model invokes a lower `uidized'layer (liquidized by the basal shear), with anexponentially convex velocity prole, overlain bya nonshearing `rigid plug'. Their relative thick-nesses depend on the total ow thickness and thegradient and roughness of the slope surface, asthe latter factors control the ow speed andrunout. The convex velocity prole, resemblingthat of a grainow (Middleton & Southard, 1984),suggests a concentration gradient and is attribut-ed to the snow texture, considered to be a mixtureof powder snow and granular snow rich in hardclods of ne pebble size. The powder snow andair, incorporated en route by the ow, act as aninterstitial uid in the basal part. The energydissipation in the lower layer is mainly due topseudolaminar shear, so long as the slope rough-ness elements have a relief/spacing ratio smallerthan 003, but is more due to the snow-clodcollisions (Bagnoldian `viscous regime') when theslope roughness is greater. The constitutive rela-tionship used in this model invokes a non-Newtonian behaviour with the power coefcient

for the strain rate taken to be 06 (contractionalow).

Snowow avalanches have an enormous impactforce and high transport competence, which ren-ders them capable of carrying huge and heavyobjects. Published measurements indicate that theimpact force may vary from 10 to 20 kPa forturbulent powder snowows to 10002000 kPa forsome large denser snowows, with a reportedmaximum of 3900 kPa. For example, Voellmy(1955) calculated an impact force of 83 kPa for asnowow that displaced a 120-t locomotive by20 m. Hopnger (1983) reported a large avalanchethat detached a three-storey barracks from itsconcrete basement and carried it across topograph-ic ridges and gorges over a distance of 15 km.

Snowow deposits

Snowow avalanches are capable of transportinglarge amounts of rock debris, including huge

Fig. 24. Multiple tongues of debris-rich slushows on a colluvial fan inSkjerdingsdalen, late spring 1988.Note that the narrow tracks ofsnowows to the left have cobblylevees, one-clast thick, but showlittle debris deposition at the termi-ni; the snowows were nonerosive(vegetation not removed), carriedlittle debris and formed the leveesmainly by shouldering aside thepre-existing slope debris. The largeblocks and smaller debris scatteredin the forefront are attributed tolarge powder snowows.

Fig. 25. Features of modern snowow deposits. (AB)Longitudinal grooves, interpreted as tool-marks formedby the dragging of large, angular clasts in snowowtraction; note the `rolling' a(t) fabric of scattered pebblesand cobbles. (C) Upslope view of a `patchy' lobe ofcoarse gravel, one-clast thick, showing clast imbricationand both a(t) and a(p) fabric; the `mixed' fabric indicatesclast reorientation by subsequent snowow avalanches.(D) Close-up view of a(p)a(i) clast fabric (depicted bythe notebook, 20-cm long) in a `patchy' gravel lobe de-posited by dense snowow. (E) Large boulders and nerdebris concentrated at the surface of a melting lobe ofdense snowow; the snow deposit at this stage is still c.45 m thick. (F) Rock debris and humic soil material richin uprooted plants, at the surface of a melting lobe ofdense snowow, which is still c. 4 m thick here.



boulders. Some snowows bear little or no debris,as is the case with many powder snowows in themid-winter time, whereas others are very rich in

clastic material (Figs 23B, 25E & F). The contentof debris in a single ow may be extremelyvariable (Fig. 22D; Luckman, 1971). Snowows



transport debris that has accumulated on thesnowpack due to rockfalls and related processes,including wind-blown nes; debris that has beenremoved from the mountain slope/ravine andincorporated en route by the ow; and debris thathas been swept by the ow from the apical part ofa colluvial fan.

Dense snowows have high shear strength, andtheir `rigid plugs' can support rock clasts as largeas boulders. Non-turbulent powder snowowscan carry cobbles in a similar way. These densersnowows thus carry rock debris in much thesame way as debrisows do, with the importantdifference that the snow `matrix' here melts outshortly after deposition and all the debris settlesto the ground. The turbulent powder snowowscarry debris similarly to rapid turbidity currents:granules, sand and ner grains are carried inturbulent suspension, and the coarser debris inbedload traction. On a steep and relativelysmooth or snow-covered slope, the bedload mayeasily include cobbles and boulders.

The sedimentary deposits of snowows, asobserved on modern colluvial slopes, range fromblankets of scattered, unsegregated debris toirregular, `patchy' lobes of unsorted debris, nomore than one-cobble or -boulder thick (Figs 6, 24and 25C). The clasts are occasionally foundresting upon one another in precarious positions,due to the passive settling from the melting snowmass. The debris is surrounded by a blanket ofwaterlain sand (Fig. 24), usually rich in negravel, but is seldom fully illuviated or buried(Fig. 26B). The sand blanket, derived by meltwa-ter runoff, has low-relief internal scours andcommonly shows planar stratication and ripplecross-lamination, even within the larger interstic-es of the buried gravel patches (Fig. 26A). Whenemplaced on a subaqueous slope and soon buriedby a debrisow, the `patchy' debris lobe mayretain its openwork texture (Fig. 26C). Otherwise,it is lled with silt and ne sand from nearshoresuspension (Fig. 15B), occasionally bearing faunashells.

The deposits of successive snowows, whenemplaced directly after one another, are amal-gamated and the indistinct boundaries render theindividual avalanche events difcult to recognizein an outcrop section. Although the discontinu-ous horizons of large clasts within a thickerblanket of waterlain sand (Fig. 6) almost certainlyrepresent separate avalanches, the actual numberof snowow events recorded by such a compositeunit is likely to be larger, because not everysnowow might necessarily carry coarse debris or

the large clasts may be few and unexposed in aparticular section. A sedimentary unit little morethan one-cobble thick may represent 515 con-secutive snowows.

Fig. 26. Details of snowow deposits in outcrop sec-tions. (A) Bouldery, clast-supported gravel whose largeinterstices have been lled with ripple cross-laminatedsand and poorer-stratied granule gravel; the knife is17 cm. (B) Debris deposited by a larger number ofsnowows in an intertidal backbeach zone, illuviatedwith nes and heavily shattered by frost action (notethe open fractures in the foliated gneiss fragments);outcrop of the Younger Dryas palaeobeach in a collu-vial-fan delta; the knife (encircled) is 17 cm. (C)Openwork gravel deposited by snowows that havedescended onto the subaqueous slope of a fan delta; theruler's arms are 23 cm each. Photographs from thecolluvial successions at Gardvik and Fiksdalstrand(localities 1 & 11 in Fig. 1).



The clast fabric of snowow deposits varies ona local scale, and is disorderly when measuredmore systematically. Boulders and cobbles de-posited from turbulent powder snowows mayoriginally have tractional a(t) fabric (Fig. 27A),but the scattered clasts on a steep and sand-lledsubstratum are highly vulnerable to reorientationby the subsequent avalanches and meltwater ow(Fig. 25C & D). The nonturbulent dense snow-ows and slushows may create an internal a(p)fabric of clasts, due to laminar shear, but thefabric loses order when the debris melts out andsettles to the ground (Fig. 25E & F).

Among the characteristic features formed bysnowows on colluvial slopes are low-relieflongitudinal grooves (Fig. 25A & B) and ribs(Fig. 27A). The occasional presence of a cobbleor boulder stuck at the downslope end of a groovesuggests that these furrows are tool-marks formedby the snowow's dragging of large angular clasts.The ribs are due to a linear accumulation ofdebris, rather than erosion (Fig. 27A), but their

formation mechanism is unclear. These longitu-dinal ribs are merely one-boulder or -cobble thickand their width/height ratio is in the range of 25.Where coupled, they may be levees made of thesurcial debris that has been shouldered aside bya dense, highly elongate snowow (Figs 13A and24). Where multiple (Fig. 27A), they may still belevees of a series of such snowows, or mayreect some peculiar, helicoidal pattern of sec-ondary ow in the turbulent boundary layer of alarge powder snowow (cf. Allen, 1984, chapt. 1).Snowows in western Norway often alternatewith heavy rainfalls, and it cannot be precludedthat the longitudinal ribs are due to strong runoff,with the sheetow power maximized by the steepslope and a frozen or water-logged substratum.

Snowows occasionally leave `debris horns' onthe upslope sides of large, immobile obstacles(Fig. 6), which is attributed to the local plastic`freezing' of a dense snowow rich in rock debris.A disorderly `meltout' fabric and an openworktexture with waterlain basal sandy inll help to

Fig. 27. (A) `Patchy' gravel lobe deposited by a turbulent, powder snowow on the grass-covered surface of acolluvial fan's ank; Sunnylvsmoldskreddalen (locality 8 in Fig. 1), spring 1993. Note the `rolling' fabric of largeclasts, the longitudinal `debris ribs', one-clast thick, and the overall scatter of debris. The slope's height is c. 60 m. (B)`Debris shadow' on the lee side of large block, left by an erosive slushow; this short ridge is a relict of the upper partof a pre-existing debrisow deposit. (A large, bouldery debrisow descended the adjoining, till-covered mountainslope of c. 40 during a heavy rainfall in the early spring 1991, and is thought to have been followed by a turbulentslushow derived from the higher slope; the latter avalanche scoured the deposit of its predecessor.) Fiksdal (locality10 in Fig. 1); photograph taken a few days after the event.



distinguish these `debris horns' from the similarfeatures formed by debrisows (see earlier sec-tion).

Snowows may also form `debris shadows' onthe downslope sides of large boulders (Fig. 6).Some of these features are depositional, compris-ing coarse debris that has apparently beendropped from a turbulent, debris-laden snowavalanche in the ow-separation zone on theobstacle's lee side. Other `debris shadows' areclearly erosional, representing the lee-side relictsof a surcial layer of collvium that has elsewherebeen wiped out by an erosive snowow (Fig. 27B).The obstacle in either case must have been deeplygrounded, or its height was much less than 1/10 ofthe ow thickness (see previous section).

The extreme hydraulic jump at the foot of a cliffmay cause a virtual crash-landing of snowowavalanches, with the snow mass splashing outand ejecting substratum debris in a blast-likefashion. A result of the powerful impacts is acharacteristic impact crater, referred to also as`plunge pool/pit' (Liestl, 1974; Corner, 1980).These oval depressions are 15 m deep and 20100 m in diameter, are often lled with meltwater(pond) and have a distinct outer rim of debrisejecta (Fig. 28). No such craters are found to berelated to the other avalanche types, althoughtheir impact on the foot-zone substratum, com-bined with the powerful thrust, can be enormous.For example, a georadar (GPR) survey in Roms-dalen indicates that some large debrisow ava-lanches have deformed the postglacial valley-oor alluvium to a depth of c. 30 m, forminghydroplastic folds, thrusts and surcial `pressureridges'.

Some illustrative examples of colluvial succes-sions rich in snowow deposits are shown inFigs 20 and 29. Importantly, these deposits oftencontain plant fragments (Fig. 25D, E & F), andalso their waterlain inll commonly includesaccumulations of humic soil rich in plant detri-tus, derived by contemporaneous slopewash.This plant material and the marine shells fromsubaqueous facies have been used for the radio-carbon dating of the colluvial deposits.

Waterow processes

Waterlain deposits are of minor volumetric im-portance in most of the colluvial systems, whoseslope catchments are typically small. However,the role played by waterow is often quitesignicant, as is shown particularly well by thecolluvial-fan deltas.

The ow of water, whether due to snow-melt orrainfall, occurs generally in two modes: as ashallow, unconned or poorly conned sheet-ow, and as a channelized streamow. Sheetowwashes the colluvial slope and hence is oftenreferred to as `sheetwash' by geomorphologists.The owing water winnows mud and sand fromthe upper slope and deposits this sedimentfurther downslope, mainly by percolatingthrough the coarse surcial colluvium and llingits interstices. Sheetow is more powerful whenthe surcial colluvium is frozen or water-logged.The shallow waterow in such conditions issupercritical, capable of transporting pebblesand dislodging isolated larger clasts. In short,sheetow transfers the ner surcial sedimentfrom the fan apex to the lower slope, alters the

Fig. 28. An impact crater formed bysnowow avalanches that havecrash-landed at the foot of a verysteep colluvial slope; Valldalen(west of locality 15 in Fig. 1), Au-gust 1996. Note the crescentic rim ofejected debris at the outer margin ofthe crater pond (c. 25 m in diame-ter).



primary texture of the surcial colluvium andtends to bury the openwork or scattered debris ofpreceding rockfalls and snowows (Fig. 29). Ananalogous process, more often called `slopewash',occurs on the adjoining mountain slope, wherethe owing water removes ne debris, includingcontemporaneous soil material, and deposits it onthe colluvial slope below.

Streamow is a more pronounced mode ofwaterow (Fig. 6). There is seldom more than onechannel active on a colluvial fan (Figs 3 and 5D),and the channel shifts due to inlling andavulsion, rather than lateral migration. Thesechannels are typically narrow gullies, developedfrom the ruts of earlier avalanches, whose trackshave been overtaken and modied by water-ow. The channels are up to 115 m deep, haveV-shaped cross-sections and generally lack lev-ees. Some channels have levees formed by earlierand/or contemporaneous debrisows, but theonly waterlain overbank deposits are thin splaysof sand and ne gravel. The modern channels aremainly ephemeral, carrying the seasonal meltwa-ter discharge and the runoff from heavy autumnrains. The isolated palaeochannels found in thecolluvial successions suggest that similar condi-tions occurred also in the past.

The waterow in the gullies is often slushy,carrying heavy sediment load and rapidly losingcompetence in the middle to lower reaches. Thesediment concentration is high, and the water mayalso bear slush at the beginning of a snow-meltseason. A `hyperconcentrated', high-density wa-terow (Fig. 21) is thought to be characterized byhigh apparent viscosity and a rheological behav-iour intermediate between plastic and Newtonianuid ow (Beverage & Culbertson, 1964; Nemec &Muszynski, 1982; Rickenmann, 1990, 1991); thiswould mean a non-Newtonian pseudoplasticuid, which has no limiting plastic strength, butwhose viscosity coefcient increases rapidly withthe decreasing turbulent shear stresses. The tur-bulence in such conditions tends to be suppressedand the deposition of debris occurs mainly byrapid dumping, with little or no tractional segre-gation.

Where the slope catchment involves a glacier orsemipermanent snowpack and the water dis-charge is relatively high and more perennial, thestreamow on a colluvial fan assumes the form ofa relatively wide and shallow channel, oftenbraided (Fig. 30). These channels may be a fewmetres wide, less than 05 m deep, and showmedial gravel bars. The ow is rapid and thesediment transport is mainly tractional. The bars

Fig. 29. Detailed log from a valley-side colluvial aprondominated by snowow avalanches, Korsbrekke (lo-cality 9 in Fig. 1). The colluvial system has prograded,obliquely, onto the submarine front of a Gilbert-typedelta formed by the valley-

Blikra and Nemec 1998 Postglacial colluvium in wertern Norway.pdf

Documents

Transcript of Blikra and Nemec 1998 Postglacial colluvium in wertern Norway.pdf