<br />J
<br />
<br />'..
<br />
<br />u
<br />
<br />564
<br />
<br />M. G, FOLEY
<br />
<br />draulic radius because it gives maximum-flow conditions in the
<br />channel - not just aVCr.lge flow conditions for the cross section.
<br />Kennedy (1961) urged the use of local now conditions fot anodune
<br />calculations in narural streams, where maximum-flow conditions
<br />may vary greacly from the mean because of irregular channel cross
<br />sections. The difference becwecn maximum depth and hydraulic
<br />radius for the January 1974 runoff evenr is only 1.4%, so the dis.
<br />tinction is not significanr. The transverse wavelength of stationary
<br />waves is raken as the width of the largest trough formed by scour,
<br />which was nearest the south bar (south channel margin in cross
<br />seCtion AN, Fig. 4a),
<br />Decembet 1974 Runoff Event. Calculated now behavior for
<br />December is the same as for January (Table 1), if the initial channel
<br />cross-section geometry is used. Maximum bed reworking from ini-
<br />tial bed elevation in cross seaion AA' (Fig. 4b) was 82 cm. At the
<br />maximum antidune amplitude of the January event, this amount of
<br />reworking implies an additional 21 em of mean-bed scour. How-
<br />ever, net scour of the center of the channel ocrurred during me De-
<br />cember event and resulted. in a posdlood bed elevation there mat
<br />was 17 cm lower man me preflood bed elevation (Fig. 4b, cross
<br />section AA'). Peak-flow calculations are sensitive to flow depth, so
<br />that it is imporranr to determine whether the net scour occurred be-
<br />fore or after peak flow.
<br />If the observed ner scour had occurred before or during peak
<br />flow, che antidune amplitude for the December event should have
<br />been in the range of 44 to 92 em (Table 1), enough to account for
<br />all the bed tewotking, Net scour and net fill of the channel as a
<br />whole support chis interpretation. Net fill, mostly fine sand, oc-
<br />curred at the edges of the scoured channel (Fig. 4b, cross section
<br />AA'). Scour-cords showed no reworking of the initial bed under
<br />chis fill, which suggests mat filling occurred early in the runoff event
<br />and was nOt disturbed by subsequem antidune development near
<br />che banks at peak flow. This asymmetry of channel reworking with
<br />net scour in one part and fill in another contrasrs wich channel be-
<br />havior in the January 1974 runoff event. It appears that the Dc-
<br />cember flow was modifying the channel toward a smaller width
<br />depth rano. The smaller widlh depth ratio implies that the De-
<br />cember flow carried a larger proportion of fine material than the
<br />January now (Schumm, 1960), and the telaovely fine-grained na-
<br />ture of the channel-edge deposits supports this implication. It is
<br />more likely that this modification occurred at peak flow when the
<br />sedimenr.transport rare was much higher, than during rhe waning
<br />flow; therefore, the flow conditions shown in Table 1 for December
<br />ptobably prevailed.
<br />Comparison between calrulated and observed maximum scour
<br />supporcs this conclusion. The January flow inferred from scour-
<br />cord measurements is ar the lower end of the calculated range for
<br />the friction factor, whereas the December flow is in the middle of
<br />the calculated range (Table 1). This suggests chat the January flow,
<br />if bankfull,. had a higher friction facror rario than the Decembe~
<br />now and hence a coatser bed load and sinaller proportion of silt
<br />and day. These variations in load material can be caused by rainfall
<br />over different tributaries, which then provide different sediment
<br />sizes. However, if mOSt bed reworking is by anridunes as is
<br />hypothesized here, gross sediment transport is roo small to allow
<br />tributary input effects to be felt very far downstream during the
<br />same flow event. Only the wash load and finest fraaions of salta.
<br />ring bed load would be expeCted to move completely through the
<br />drainage system. In this case, channel cross-seaion modification is
<br />either a re'sult of transportation and deposition of only the finest
<br />fraction of the bed load or, as W. W. Emmett (1976, written com-
<br />mun.) has suggested, a result of passage of more slowly moving
<br />slugs of tributary sediment derived from previous runoff eVents.
<br />
<br />Resolution of this question will require many more flood observa.
<br />tions than the two reported here.
<br />It is possible [Q account for the asymmetrical bed reworking in
<br />the December 1974 runoff event entirely by antidune development,
<br />if it is assumed that the transported load was more fine-grained
<br />than usual; nevertheless, a remote possibility is that the December
<br />flew asymmetry developed because part of the Stream bed was fro-
<br />zen. The southern half of the stream bed is shaded by a diff, and in
<br />episodes of cold winter weather it escapes diurnal thaw. Under such
<br />conditions a midday runoff eVent would scour the northern part of
<br />the channel more easily than the southern parr. However, the flow
<br />would probably thaw the thin (less than 5 em thick) frozen layer on
<br />the southern part of the channel before the end of peak flow, and an
<br />initially partly frozen bed does not explain the fine-grained fill on
<br />both the northern and southern channel edges. The assumption of
<br />line-grained load is thus the preferred explanation. . ~.
<br />
<br />lABORATORY EXPERIMENT
<br />
<br />i:.
<br />J ~~i"
<br />
<br />:l~
<br />A series of laboratory experiments on scour and fill was con-.';
<br />ducted in the W. M. Keck Laboratory of Hydraulics and Waret ae.:.~,
<br />sources, California Institute of Technology. These were sand-bed,~
<br />experiments in a nonrecircularing. rigid-wall flume with rim~::
<br />varying rates of water and sediment input patterned on rhose of1
<br />floods in ephemeral streams. The purpose was to observe scour and ~_
<br />fill caused by noods in a steep, sand,bed stream in a controlled 1a1>-1
<br />orarory environment. These experiments were intended as"~
<br />analogues of the field environmem, not scale models of any parriai~!
<br />larnarural stream. .'":J),
<br />i~~
<br />Equipment and Procedure _:.-~
<br />-.,j,
<br />ExperimentS were conducted. in a rigid-wall channel18.J m long, ~
<br />26.7 em wide, and 25.4 em deep, installed in a tilting steel flume.;~
<br />Flwne slope was adjustable from 0% to 1.5%. Time-varying walte~.. ".
<br />and sediment input was conrrolled by prc::programmed moror: '
<br />driven cams that controUed a pneumatic .......ater inlet valve and a
<br />variable.speed wet sand feeder. Water supply for the flume was
<br />constant-head tank supplied. from a large reservoir. . ~
<br />Flume oudet was a frec-overfaU into a settling tank, where sari '.
<br />was separated for recycling before the water was returned to the i
<br />reservoir. Total sediment discharge was determined at the outl1l
<br />using a travetsing vertical-slot sampler, Water-surface and bed e1e:Jiii
<br />vacions were determined using a point gage mounted on a carriag~
<br />that could be moved the length of the nume on tails mounted at~PI',
<br />the flume walls. The point gage permitted location of any point In. t
<br />the flume to within 0.5 mm horizontaUy and 0.05 mm vertically: .
<br />Determinations of water-surface and sand-bed elevations reproduc.lt
<br />ible within 0.05 mm could only be made for srill water, or for sand '
<br />saturated with water but with no standing water. Sand.surface de::.
<br />vation was rhus determined from the elevation of the water he.ld
<br />between the grains by surface tension. Watet-surface clevatio
<br />were recorded continuously at seven locations along the flume ~.,
<br />pressure transducers attached to piezometric taps in the flume ho~- ;it
<br />tom. Sttip-chatt rerordings of transducer output were used to d
<br />termine mean water.surface elevation and slope during unstead
<br />flows with violent surfao:-wave activity. Pressure.transducec d~
<br />terminations of water~surface elevations were accurate co wich~
<br />0.2mm.
<br />Bed and input load material used in these experiments w,as-.
<br />natural sand wich geometric mean size d, = 0.28 mm and geom~~
<br />ric standard deviation cr = 1.42. This sand consisted almost e~~,~.
<br />tirely of subcounded to well-rounded quartZ grains, wich traces -~-
<br />magnetite. ~
<br />
<br />
|