in Location-dependent sorting sediment gravelly megaripples fromtheRioGrande, centralNewMexico byDavidW.Love,JamesT. Socorro, NM Boyle, Resources, Sylveen Robinson, andMarkHemingway, NewMexico Bureau o{ MinesandMineral Introduction An ongoing study of modern fluvial sediments in New Mexico has revealedthat grainsizesorting in streamsis dependenton sample locationwithin portions of bedformsand on bedform position within downstream progressionsof similar forms. Theseresults have many potential geologicapplications. One of the sets of bedforms studied is straight-crested,gravelly megaripplesof the Rio Grande conveyancechannel near Socorro, New Mexico. Megaripplesare rippleshapedbedformshaving wavelengthslarger than 60 cm and a range in amplitudes larger than ripples. In streams,megaripplesform under lower-flow-regimeconditionsand rhigratedownstreamby progressivedeposition along their slip faces(Fig. 1). The purposes of this paper are to relate grain-sizedistributions to the form and location of megaripples and troughs left by waning flow along t h e c o n v e y a n c ec h a n n e l , t o d e t e r m i n e whether a megaripples'sposition within a progressionmay be locatedby analyzing its grain-size distribution, and to determine s o r t i n g m e c h a n i s m st a k i n g p l a c e a l o n g megaripples. Description of channel conditions The conveyancechanneleastof Socorrois about 23 m wide at the top and 4 m deep, with an averagegradient of 0.016 (Fig. 2). During the week of April 8, 1984,flow averagedan estimated86 m'/sec(basedon flow estimationtechniquesof Williams, 1978).On April 14,flow in the conveyancechannelwas diverted back into the Rio Grande channel upstream. During the next four days, the flow dropped to an estimated 3 m3/sec,so that the channelwas approfmately 10m wide and 0.2 m deep.Flow was maintainedin part by seepagefrom the banks. On April 19, the z o n eo f n o d i ff u s i o n water was clear and the tops of some of the megarippleswere near the surfaceof the water when surveying and sampling were conducted (Fig. 3). Little sediment was in suspension. Sand was transported only where megarippleswere erodedby flow in the thalweg. Most parts of the megarippleswere no longer migrating downstream. Straight-crestedmegaripples with wavelengthsaveraging2.8 m were measuredand sampledalong a 30 m long reach where reworking by a low-flow thalweg had not altered the bedform configuration. After samplingwas completed,the reachwas surveyed using a Leitz TM 20 theodolite and a rod measuring to the nearesthundredth of a foot (about 3 mm). The megarippleshad long, low-angle stoss sides, steep lee sides (at the angle of repose),and crestsabout 12 cm above the adjacenttroughs downstream (Fig. a). No ripples occurred on the stoss flow velocity oce toce c r es t 6i6s\de interior t r o ug h r e v e r s ef l o w m o b i l eb e d o v o l o n c h er e v e r s e flow FIGURE l-Anatomy of megaripples, water flow (in blue), and sediment movements (in black) involved from Jopling (1963) and Reineck and Singh (1980). Index map shows approximate study location. FIGURE 2-Low flow in conveyancechannel of Rio Grande near Socorro on April 19, l9&1. Note tops of megaripplesnear water surface. Channel is approximately10 m wide. May 1985 Nao MexicoGeolog" in sorting processes Modified FIGURE 3-Megaripples in conveyancechannel of Rio Grande. Thalweg in lower middle portion of picture is reworking sand and obliterating meganPPles. N-. tlow direction J I EV E I O 5m ffi 2 X v e r l i c o le x o g g e r o l i o n FIGURE 4-Longitudinal profile of megaripples and water surface sidesof the megaripples,exceptin the thalweg/ where the megarippleswere beginning to be reworked and fine sand was being transported.The exact timing of depositioi of the megaripples is not known, but probably the formation and preservation of the bedformslaggedbehind flow conditions (cf. Allen, 1982).The slip faces probably were depositedjust beforeflow waned to the point of no sediment transport, whereas the interior portions of the megarippleswere deposited during active transport conditions. Although the timing and conditionsfor deposition of the megaripples remain partially undetermined, the fact that they were preserved during waning flow suggestsihat similar sedimentary structures in geologic contextsmay be examined and interpreted in a similar manner. Sample treatment Sampleswere taken from the megaripples by sampling portions of bedforms progressively upstream so that the samples remainedundisturbedand uncontaminatedby sedimentsagitatedduring the samplingprocess.Thus, troughs downstream from each megaripplewere sampledfirst, then the lower slip faces,upper slip faces,the armor at the crestof the megaripples,and, finally, the interior portion of each megaripple. Not all megarippleswithin the sequencewere sampled, primarily becausethey were all very similarin overallcharacteristics. We were not convinced that differenceswould be found between adjacent megaripples, and, therefore, we expectedonly slight differencesbetween the upstream and downstream ends of the sequence.Where the megarippleswere dominatedby gravel, about 1 kg sampleswere taken in an areaof 200-400cm'. The uppermost few cm were scoopedbv hand into a la,rgeplastic bag. The interiors of megaripples were sampled by clearingaway a[l surface clasts from a broader area and then scooping the sample from the megaripple by hand. The dominantly sandy slip faceswere sampledusing film canistersthat hold about 40 g of sand when full; one to two canisters of sand were taken from each of the upper and lower slip faces. Sedimentsforming the megaripples and troughswere extremelylooselypacked.\A/hile walking acrossthe channel, a 55-kg (143lb) person could sink up to 60 cm into the uncompactedsandy sediments.Although slip faceswere present on the lee sides of the megaripples,no internal sedimentarystructures were seen during sampling, probably becauseslumping occuied immediaietyafter sampleswere taken, and becauseof the lack of differentiation of grain sizes into discrete laminae.Unfortunately, time and expensedid not allow more sophisticatedtechniquesfor s a m p l i n g c r o s s s e c t i o n so f s e d i m e n t a r y structures(such as taking epoxy box cores) to be used. In the lab, the samples were dried and sievedusing sievesizeslisted in Table1. The entire sample was sieved to a size of 1 mm in order to have an adequatesampleof large clasts. Sampleswith abundant portions of sand less than 1 mm were sDlit to about 50 g and sieved.Then the proportion of the total was calculated.Duplicatesplits of some samnleswere sievedto test the adeouacvof this procedure. Mass-frequency diagrams were plotted following Bagnold's (1941)procedures.In the diagrams that illustrate grain-size distributions (Figs. 5-9), both axes are logarithmic becauseof the wide rangesin grain size and masspercentages.The irregularintervalsbetween sieves are standardized by dividing the mass percent in each size classby the differencebetween logarithms of the diametersof the largestand smallestgrainsin the size class.The diagramsshow the mass frequencyof different-sizegrains in such a way that proportions of one population may remain similar even though other populations are added or subtractedfrom the total in other parts of the distribution. In the olots that iollow, some sievescaughtconsistentlyless mass than the sieves adiacent to them becauseof deviationsin aperturesize.For example, the 0.5-mm sieve nearly always had less sand in it than did the 0.43-mm sieve, indicatingthat someaperturesof the 0.5-mm sieve are too large and let too many grains through to the 0.43-mmsieve.Theseinconsistencieshave not been correctedin the plots, but are clearly apparent as a V-shaped depressionin the grain-sizecurvesat 0.5 mm. Somecurvesare discontinuousat the coarse end becauseno clastswere caught on some sieves. TABLE l-Sieve aperture sizes in mm used in sediment analyies. 38.05 26.67 18.35 IJ. JJ 9.52 8.00 6.30 5.66 4.76 4.00 3.35 2.83 2.36 2.00 7.68 1.40 1.18 1.00 0.85 0.71, 0.605 0.500 0.430 0.355 0.303 0.250 0.272 0.180 0.150 0.725 0.707 0.090 0-075 0.063 0.04s from unstream to downstream to illustrate any lorigitudinal shifts. More subtle trends are revealedby comparing adjacentparts of megaripples/troughs for different populations/line slopes and by examining relative percentagesof different parts of the populations. These differences and trends are summarizedbelow. Distributions from analogousportions of megaripples As shown in Fig. 5, troughs show a broad, skewed peak between 0.25 and 8-10 mm, a decreasein grains larger than about L0 mm, a paucity of grains between 0.075and 0.180 mm, and a secondpeak at 0.063mm. Grains between 0.85 and I mm increasein mass percent downstream, whereas grains between 0.355and 0.85 mm decreasein mass percent downstream. Grains armoring the crest of megaripples have distributions somewhat similar to troughs, becausethey have a generalbroad peak from 0.25to 8 mm and a small peak at about 0.063 mm (Fig. 5). The broad peak greaterthan 0.25mm tends to be flatter (more of a plateau) than the broad peak of the Results troughs, and it tends to show slightly lesser T h e g r a i n - s i z e d i s t r i b u t i o n s f r o m t h e amounts of the coarserclasts (curve has a megaripplesdepend on where sampleswere negative slope at coarsersizes).Grains betakenwithin eachmegarippleand associated tween 1.4 and 4.76 mm decreasein mass trough downstream, and on the location of percent downstream, whereas grains bethe megaripple and trough set within the tween 0.25and 1.4mm increasein masspersequence.The differencesin grain size are cent downstream, with the last megaripple reflectedin the overall distributions at a given having a definite peak between 0.5 and L.18 location within each megaripple, in trends mm (this peak may be due in part to samdownstream along each megaripple/trough pling the underlying megaripple interior). No sequence,and in proportions of gravel frac- trend is apparentin the amount of fine sand tions and various sand fractions within in- forming the peak at about 0.063mm. dividual megaripples and along the entire Samplesfrom the interiors of megaripples sequence.The grain-sizedata are organized show a similar broad peak from 0.25 to 6.3 in Figs. 5-9 for each location along mega- mm and a precipitousdeclinein the amount ripples/troughs,and the data are "stacked" Nm Mexico Geology May 1985 o o o o c c o a N N o o o o o I o o c o c o o o- o.l E o E o J ) roo Groin s i z ei n m m of troughsloFIGURE S-Grain-sizedistributions frommegaripples A, B, C, and cateddownstream J For clarity,curvesB, C, and J havebeenoffset additionof 0.3, 0.6, verticallyby the respective and0.9to theirY-values. of sand less than 0.25 mm (Fig. 7). In comparison with samples of crest armor, interiors have fewer clastslarger than 6.3 mm. There appears to be a decreasein fine sand fraction downstream, but secondary peaks at 0.107or 0.063 mm or both occur ln all interior samples. Upper slip faceshave a narrower range of grain sizes,commonly exhibiting a peak between 0.25and L mm (Fig. 8). From the upstreammegaripple (J) to the next to the last megaripple (B) in the sequence/the distribution shifts to a finer peak (from 0.605-0.85 to 0.303-0.43mm), but the last megaripple (A) is similar to the first. Moreover, samples from different parts of the upstream megaripple (activeand inactive) are quite different in proportions of coarseand fine grains (Fig. 8: Ja, active, and Jb, inactive). Lower slip faces are much better sorted than either the troughs or the rest of the megaripple locations. They have a single, relatively narrow peak, slightly skewed toward the fine side (Fig. 9). The peak of the distribution shifts downstream from 1.181,.4to 0.7t-0.85 mm. The amount of grains between0.063and 0.71mm increasesdownstream,affectingthe slope of the fine side of the distribution, but no separatepeak of fine sand occurs. Grains between 1,.4 and 2.83 mm decreasedownstream,but the last lower slip face (A) has an increasein granulesbetween 3.35 and 4 mm. c o o o- E o o J 2 G r o i ns i z ei n m m I G r o i ns i z ei n m m FIGURE6-Grain-size distributions of crest armor of megaripples A, B, C, and J. For clarity, curves B, C, and J have been offset vertically by the respectiveaddition 0.3, 0.6, and 0.9 to their Y-values FIGURE7-Grain-size distributions of megaripple interiors. No curves have been offset. Curve A is the same as that for crest armor A because no separatesample was obtained. o a) c c o o N N j'Jb ol o o o o o o cn C o o @ 3 -r E E o ) ) -2 oor o.l r "dffi"J.:lt FIGURE 8-c,",^-,;" ro roo srip upper G r o i sn r z et n m m FIGURE9-Grain-size distributionsof lower slip A, B, C, andJ CurvesB, C, facesof megaripples A, B, C, andJ. CurvesA, B, facesof megaripples C, and fa arenot offset;Jb is offsetby 0.5Y-units and J are offsetverticallyby the respectiveaddito avoidoverlap Jais from activeslip facein thal- tion of 0.2,0.4,and 0.6to theirY-values. weg;fb is frominactiveslipface2 m from thalweg. ferences downstream. The lower slip faces arenot presentin slip faces.Secondarymodes containprogressivelyless massgreaterthan at 0.063mm also decreasefrom troughs to 0.605mm downstream, but other megaripslip faces.In two of the progressionsfrom ple portions do not show significanttrends. lower slip facesto troughs, the troughs have Discussion a deficiencyof clastsin the rangeof the peak comparing grain-size merely Rather than of the lower slip faces. distributionsof adjacentsamples,discussion Other locational indicators of the above results focuseson the sorting In comparing curves, it was noted that mechanismsresponsiblefor producing the points where two curves cross(termed cross- distributions and on interpreting the distriProgressivedownstream changesin over points) may shift significantly depend- b u t i o n s b a s e d o n t h e p r o p o s e d s o r t i n g distributions from adiacentportions i n g o n l o c a t i o n w i t h i n t h e m e g a r i p p l e mechanisms. of megaripples sequence.For example,the crossoverpoints Downstream progressionsalong individ- between crest armor and adjacent trough Sorting mechanisms Numerous factors determine the grain size ual megaripplesalso revealtrends in the fol- rangedownstreamfrom 5.66mm atl to 4.76 lowing sequence:1) trough to 2) megaripple mm at C to 2.36mm at B to 1.4-1.68mm at and sedimentary structureswithin eachreach crest to 3) interior to 4) upper slip face to 5) A. Similarlv. the amount of mass in certain of a stream. In general, dominant grain size lower slip faceto 6) trough. As Figs. 5-9 show, sizeintervalsmay show trends downstream. along streambeds and characterof flow (such distributions becomebetter sorted and uni- As Table 2 shows, the mass of clasts larger as velocity, depth, stream powet and shear modal in progressionsfrom troughs to lower t h a n 2 m m i n t r o u g h s d e c r e a s e sd o w n - stress) determine bedforms. For examPle, slip faces. The coarse modes (>8 mm) of stream,but the massof all clastslarger than stream beds with a preponderance of grains troughs decreasein crestsand interiors and 0.5 mm in troughs shows no significantdif- larger than 0.5-0.7 mm do not develop ripMay 1985 Nar MexicoGeology TABLE 2-Examples of longitudinal ranges in weight percent of selectedlarge clast sizes from portions *armor and interior sampled together of megaripples along the conveyancechannel of the Rio Grande; face, here. slip represented in megarippleA; b, only Jb, inactive upper megaripple C B percent > 2 mm in troughs 75.3 75.1, 73.6 72.2 troughs percent > 0.6 mm in upper slip Iower slip 83.3 91.3 91.8 97.3 ples; rather, as flow is increased,such beds progressfrom being stationary to planar mobile beds to megaripples to upper-flow-regime plane beds, and finally to antidunes (as summarized in Reineck and Singh, 1980). Under lower-flow-regime conditions of water and sedimentdischarge,sedimenttransport and sorting mechanismsin gravelly megaripples depend primarily on 1) mobile-bed phenomenasuch as armor developmentand overpassing,and on 2) flow separition at the brink of the bedforms (Fig. 1). As a result of fluid shear stressalong the bed, mobile (i.e., all grainsmove episodically and cannot be consideredlag) gravelly beds develop an armor or pavement at the bedwater interface that results in nearly equal transport rates for all sizes of clasts in the bedload (Andrews, 1983;Parkerand Klingeman, 1982);winnowing of finer sizesby differential bedload transport apparently does not take place. Grains larger than about 4-5 times the median diameter of the bedload are most exposedto shearstressand tend to overpass thi bed (Everts, 7973; Allen, 1983; Andrews, 1983).Fine grains may also overpass, but they are more likely to become transportedin partial or full suspension.As grains are remobilized from the pavement, other grains drop into the "holes" and becomelessmobile.Andrews (1983)and Parker and Klingeman (1982)imply that grain-size distributions of pavement should reflect nearlyequalamountsof a largerangein sizes of grains.Parkerand Klingeman(1982)noted that in gravelly streams with plane beds, bedload and subpavementgrain-sizedistributions tend to be similar, while the pavement at the interface is coarser.However, with megaripple bedforms other sorting mechanismsappear to influence the grainsize distributions of the subpavement so that it is no longer similar to the bedload. FIow separationat the brink of megaripples takes some grains beyond the brink to settle through zones of differential turbulence to the trough or stossside of the next m e g a r i p p l ed o w n s t r e a m ( j o p l i n g , 1 9 6 4 ) . Grains settling beyond the brink are sorted according to forward momentum, settling velocity, and strength of turbulence. Small grains are carried away in suspension.Grains near the bed in the zone of differential turbulencemay be caught in a zone of reverse flow, being transported toward the slip face of the megaripple upstream. Other grains aredepositedat the brink and avalanchedown the slip face.Grainscaughtin avalanchesare sorted by differential shearing and relative dispersivepressure/as well as by turbulent suspensionof smaller grains. 99.8 97.7 86.4 81.9 a/.5' 72.8 28.7 52.3 armor interior 83.7 75.4 77.4 65.6 /6.J / /.2 83.0* Interpretationsof distributions based on sorting mechanisms These mechanismsof grain movement in megaripplessuggestthat the interiors of the bedforms should be made up of upper and lower slip faces,and that pavement on crests and stosssidesof megaripplesshould be derived in part from reworking megaripple interiorsalong with addition of grains that passed over previous bedforms. These implied relations appearto be substantiatedby the grainsize distributions illushated in Figs. 5-9. The interiors of megaripplesappear to be made up of sedimentsfrom the slip faces, although some clastscoarserthan those of the slip facesalso occur.Thesecoarseclasts in the interior of the megaripplesmay be due to flow conditions at an earlier time of deposition; the sampled slip facesformed later during waning flow. Megaripple interiors C andJ aresimilar to upper slip faces,whereas A and B appear to contain components of both upper and lower slip faces.In part, this may be a problem of sampling at equal depths or at equivalentpositions in adjacentmegaripples. The excellentdegreeof sorting of the lower slip facesis unexpectedconsideringthe proposed mechanismof intermittent avalanching of grainsdown the slip face.In bedforms studiedin other streams,the lower slip faces are the least well sorted locations. If avalanching was the chief sorting mechanism, coarsergrainsshould be expectedat the base of the slip face. Instead, the shape of the grain-size distributions suggests sorting during unidirectionalflow similar to that of sand in ripples and plane beds. Therefore, the sand of the lower slip face mav be resorted and accumulatedby reverse'flowin the wake of the megaripples. Under this mechanism,avalanchingcoarsegrains either may be trapped occasionallyat the base of the slip face or may roll to the trough and continue downstream. The medium to coarse sand,on the other hand, may accumulateat the baseof the slope where ihe reversecurrents begin to rise. Finer sand would continue in suspension or perhaps would be depositedon the upper slip face. Armor could have been derived either by 1) reworking megarippleinteriors, removing large amounts of finer clasts, and accumuIating coarserclastsas lag, or by 2) additions of coarserclastsin a mobile bed that moved over the troughs and up stoss sides of the megaripples.The nearly flat "plateaus" between 0.25 and 8 mm of the grain-sizedistributions of the armor (Fig. 6) could indicate eitherthat clastsin that sizerangewere equally mobilebecauseof equal shearstressto move all sizes(cf.Andrews, 1983)or that the grains were differentially exposed to shear stress depending on size and placement of surr o u n d i n g g r a i n s ( P a r k e ra n d K l i n g e m a n , 1982).The plateausof nearly equal amounts of grains are much wider (8 mm is 32 times as lig as 0.25 mm) than p,redictedby Andrewi (roughly 14 times from 0.3xd50 to 4.2xd50).This discrepancymay indicatethat the larger grains are moving under a nearly constaXtliw value of criticll dimensionless shear stress.Lower shear stressin troughs would force accumulationof otherwise mobile large grains, accounting for the slight "hump" in the massof coarsegrains (Fig. 5). If megaripplesand troughs migrate, stoss sides of megaripplesmust be progressively erodedaway as the form shifts downstream' The differences between crest armor and trough armor may be explained either by progressive accumulation(lag)of coarsegrains br 5v additions of coarsegrains in transit froni upstream. Basedon tfie results of Andrews (1983) and Parker and Klingeman (1982),armor is not due to lag. Therefore, the coarsegrainsmust come from uPstream, probably from erosion and overpassingof s t o s s s i d e s o f m e g a r i p p l e s .T h e r e l a t i v e abundanceof coarsegrains in troughs compared 'weak to crestsmay be due to the relatively overall currents in the zone of reaitachment (Fig. 1). Large grains may not be moved out of this area as readily as smaller grains. The abrupt decreasein mass percent of grains less than 0.25 mm in the armor of irests and troughs suggestsan efficientsorting processsuch as suspensionof finer grains, but secondary peaks commonly occur at 0.090-0.107and 0.063mm in the armor of the troughs and crests.This sorting of secondary modes takes place despite the fact that no ripples form along beds with average grain sizelargerthan about 0.6 mm. Bridge (t98t; explored the production of grain-size distributions with abrupt decreasesin the amountof fine grainsdue to suspension.Most of his theoreticaldistributions show an abrupt decreasein amounts of grainsfiner than u fiiu"tr size determined by-the strength of bed shear stress.For example,Bridge's calculated grain-size distribution for a mean shearstressof 4 dynes/cm'yields a curve that abruptly declinesin massof grains less than 0.21mm in diameter. Bridge (1981)proposedtwo alternativesto explainthe presenceof fine grainsin gravelly deposits.He suggestedthat some fines may be presentdue to entrapmentand infiltration between larger grains, particularly as flow wanes, and other fines may be present due to mechanicalcrushing of some clasts between larger clasts.Mechanicalcrushing to producefines in the range of the grains seen in the megaripplesappearsto be extremely unlikely. Some trapping of fines may occur becausethe bedforms are loosely packed. However, if one assumesthat the 0.25 mm grains (the smallest grains in great abundance)contTolthe sizeof intersticesbetween NezoMexicoGeology May'1985 grains,the largestspacebetweengrainswould be about 0.100mm with cubic packing and 0.039 mm for close rhombohedral packing (derived from formulae in Pettijohn, 7957). Under rhombohedral packing, void fillers could be between0.104and 0.056mm in size, but the void-filling grains could not filter between grains. In any case, smaller grains should be favored over larger grains in an entrapmentprocess,and such definitepeaks at about 0.063mm should not be expected. Therefore,somesorting processof fine sand along the bed must be taking place. The mechanismappearsto takeplaceduring mobile conditions rather than after movement of all coarsegrains has ceasedbecausesome subpavement samples exhibit secondary peaks. Fine grains may be sorted into secondary peaks by reverse flow up the slip face,but this mechanismwould not explain secondarypeaksin armor. The small modes of fine sandin armor suggestthat secondary currentssort fine sand moving between larger grains, but the nature of the sorting processesproducing secondary peaks of sand remains to be examinedfurther. The downstream trends in qrain-sizedistributions from analogonsporlio.ts of megaripples along such a relatively short reach are difficult to explain, but may be considered in terms of 1) changesin flow conditions, 2) pulses of sediment, and 3) relative mobility of grains.The actualexplanationmay be a combinationof thesefactors.Deoosition (anderosion)at eachpoint alongeachmegaripple/trough must be controlled by local flow conditions and availability of grains. The sampledreach is only a small portion of the conveyance channel containing megaripples. No obvious changes in channel geometry or size of bedforms occur along the reach,so changesin averageflow conditions along the reachappear to be unlikely causes of changesin sediment sorting. However, the reach is in a gently curving portion of the channel (Fig. 2) and flow could change along the curve. Possiblythe waning stages of deposition changed progressivelydownstreamso that deposition, particularly along slip faces,continueddownstreamafterit had ceasedupstream. The uniformity or periodicity of sediment transport cannot be assessedwithout measurementsof flow and sediment discharge during formation of the megaripples. Differentpulsesof grainsmay have reacheddifferent portions of megaripples at different times during deposition, providing more coarsegrains to upstream megaripplesthan to downstream megaripples, or providing more grains in the range of 0.85-8 mm to downstreamtroughs after passingupstream bedforms. The trough farthest downstream contains all but the coarsestclasts, so most sizes of grains have been transported through the reach.The largestgrainsmust be lessmobile than smallergrains so that movement of large grains from trough to trough is slowed. Therefore, upstream troughs should have more large claststhan downstreamtroughs, until large clastshave migrated throughout 30 May 1985 New Mexico Geology the reachto achieveequalratesof movement from bedform to bedform. The possibility of Iarger-scaleperiodicity in deposition along the conveyancechannel cannot be evaluated. Conclusions In gravelly megaripplesalong the conveyance channel of the Rio Grande, grain-size distributions and some of their d'erivative parameterscan be used to locate samples within a longitudinal sequenceof bedforms. Each part of a megaripple (crest, interior, upper slip face, lower slip face, or trough) has a diagnosticgrain-sizedistribution. Distributions from eachportion of megaripples changeslightly downstream so that locating a samplewithin a train of megaripplesis possible.Along a reach of only 10 megaripples, the crests,lower slip faces, and troughs of upstreammegaripplesare coarserthan their downstream counterparts. Similarly, the crossoverpoints between grain-sizecurves of crests and troughs shilt to finer sizes downstream.Although the mass percent of all clastslarger than 0.6 mm in troughs shows no significanttrends downstream, the mass percentof clastslarger than 2 mm does decreasedownstream. In lower slip faces,the massof clastslarger than 0.6 mm decreases downstream. Sorting mechanismsthat produce different grain-sizedistributions along megaripples apparently are related to mobile-bed formation of armor along troughs, stosssides, and crestsand to flow separation,avalanching, and reverseflow along slip faces.Suspension of fines less than 0.25 mm appears to take placefrom troughs to crestsof megaripples. Fine sand is also sorted into secondary modes that may indicatethe presenceof secondary currents between larger gravel clasts.The reasonsfor trendsin distributions downstream over a distance of only 30 m may include changes in flow conditions, pulses of sediment, and/or relative mobility of the largestgrains. The grain-size distributions illustrated aboveshow the importance of knowing the location of samples with respect to bedforms. Only the grain-sizedistributions of the lower slip facesappearto have relatively simple probability distributions. The variation in grain-size distributions along bedforms demonstratesthat many fundamentallv different distributions occur in fluvial deposits. The techniqueof relating grain-sizedistributions to locationwithin bedformssuggests severalpotential directionsfor further studv. Where grain-size data are gathered from different bedforms formed under known flow conditions, it may be possibleto relate the distributions to flow conditions more quantitatively.Alternatively,more data may demonstratethat severalsets of flow conditions can produce similar distributions or that one set of flow conditionscan produce dissimilar distributions. If quantification of flow conditions is possibleusing grain-sizedistributions from similar bedforms. it also mav be possibleto determine what entire bedforms were like from the portions that are preserved in geologic sectionsas well as their relative longitudinal position within a progressionof bedforms. ACKNOWLEDGMENTS-We are grateful to the New Mexico Bureau of Mines and Mineral Resourcesfor supporting this study. We thank Ron Broadhead,John MacMillan, and Steve Cather for their helpful comments and criticisms.We thank Cherie Pelletierfor drafting the figures. We are especiallythankful to Deb Shaw for editing and reducing the length of the manuscript, and for helping to produce the figures. References Allen, J R L , 1982, Computational models for dune time-lag----calculations using Stein'srule for dune height: SedimentaryGeology,v.20, pp 165276. Allen, J R L , 1983, Gravel overpassing on humpback bars supplied with mixed sediment-examples from the lower Old Red Sandstone,south Britain: Sedimentology, v 30, pp 285-294 Andrews, E. D, 1983,Entrainmentof qravel from naturally sorted river-bed material:Geol6gicalSocietyof America, Bulletin, v 94, pp 1225-7231 Bagnold, R A., 1941, The physics of blown sand and desertdunes:Methuen, London, 265 pp Bridge, | 5.,1981,,Hydraulic interpretation of grain-size distributions using a physical model for bedload transport: Journal of Sedimentary Petrology,v 51, pp 1109't1.24 Everts, C H , 1973,Particleoverpassingon a flat granular boundary: American Society of Civil Engineers, Proceedrngs,Journal of the Watemays, Habors, and Coastal Engineering Division, v. 99, no WW4, Paper 10125, pp 425-438 Jopling, A. V., 1953, Hydraulic studies on the origin of bedding: Sedimentology,v 2, pp 1,1,5-121 Labontory study of sorting processes Jopling,A. Y., "1954, related to flow separation:Journal of Geophysical Research,v 16, pp 3403-3418. Parker, G , and Klingeman, P. C ,7982, On why gravel bed streams are paved: Water ResourcesResearch,v 1.8,pp. 1409-1423 Pettijohn,F. I ,7957, Sedinentary rocks:Harper and Row, New York, 718 pp Reineck,H E , and Singh, I 8., 1980,Depositional sedimentary environments: Springer-Verlag, New York, 549pp Williams, C. P., 1978,Bank-full dischargeof rivers: Water ResourcesResearch,v. 1.4,pp 1,141,-1,1,54 Tl