channel avulsion on alluvial fans in southern arizona

Ž .Geomorphology 37 2001 93–104www.elsevier.nlrlocatergeomorph

Channel avulsion on alluvial fans in southern Arizona

John Field)

Green Mountain College, One College Circle, Poultney, VT 05764, USA

Received 6 January 1999; received in revised form 1 September 2000; accepted 4 September 2000

Abstract

Historical aerial photographs and field observations on five fluvially dominated alluvial fans in southern Arizonademonstrate that channel avulsion invariably occurs where bank heights are low and often at channel bends. Channelabandonment occurs through stream capture when overland flow from the main channel accelerates and directs headwarderosion of smaller channels heading on the fan surface. Five distinct channel morphologies observed on the fans are relatedto different stages of the avulsion process and can be used to identify areas on a fan surface that are prone to avulsion. Adescriptive model of channel avulsion illustrates how the morphology of a single channel reach will evolve through time asit captures the main flow path and is itself eventually abandoned. Immediately following avulsions, small preexistingchannels that capture flow from the main channel will typically experience three fold or greater increases in channel width.Subsequent large floods can be stably conveyed through these high-capacity reaches. An uninterrupted sequence ofsediment-charged small flows, however, will eventually begin to back-fill the wide channels as vegetation growth stabilizesthe banks. The stabilized and back-filled channels are now prone to abandonment during large floods because the decrease inthe channel’s capacity leads to the generation of overland flow beyond the margins of the shallowed channels. The action ofthe small aggrading floods is critical in the avulsion process since the greatest amount of overland flow is generated wherebank heights are lowest. As a result, both small and large floods are effective agents of landscape change on the fans.Channel avulsions on the five fans are not completely random events in space and time because their occurrence iscontrolled by the relative positioning of low banks along the main channel and smaller channels draining the fan surface.Consequently, the location and timing of future channel avulsions can potentially be anticipated in an effort to improve floodhazard assessment on fluvial fans in the rapidly urbanizing southwestern United States. q 2001 Elsevier Science B.V. Allrights reserved.

Keywords: avulsion; alluvial fan; Arizona; geomorphic effectiveness

1. Introduction

The importance of channel migration on alluvial-Žfan development has long been recognized Drew,

. Ž 3 4 .1873 . Over long time periods 10 to 10 years ,

) Tel.: q1-802-287-8270.Ž .E-mail address: [email protected] J. Field .

channels must migrate over the entire surface toensure the establishment and maintenance of fanform. When considering the long-term aggradationof alluvial fans, the exact location of channels throughtime is indeterminate and channel migration can be

Žmodeled as a stochastic process Price, 1974; Hooke.and Rohrer, 1979 . However, short-term processes of

channel migration on alluvial fans are of greater

0169-555Xr01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.Ž .PII: S0169-555X 00 00064-7

( )J. FieldrGeomorphology 37 2001 93–10494

Fig. 1. Map of Arizona showing major cities and the location ofŽ .A Ruelas, Wild Burro, and Cottonwood Fans on the Tortolita

Ž . Ž .Mountains piedmont; B White Tank Fan; and C Tiger WashFan. See Table 1 for UTM coordinates of fan apices.

interest to engineers and geologists studying floodhazards and the effect of individual floods on fanmorphology. Mathematical models used by the Fed-eral Emergency Management Agency to establishflood hazard zones on alluvial fans assume random-ness in the location of channel avulsions during a

Žsingle flood Federal Emergency ManagementŽ . .Agency FEMA , 1995 . However, a recent report by

Žthe National Research Council National ResearchŽ . .Council NRC , 1996 calls into question whether

the position of alluvial-fan channels during a singleevent should be considered indeterminate. Previousstudies have examined short-term channel avulsion

Žprocesses on debris-flow fans Beaty, 1963; Whipple. Žand Dunne, 1992 and fluvial fans alike Kesel and

.Lowe, 1987; Wells and Dorr, 1987 , but many ques-

tions related to the National Research Council reportŽ .NRC, 1996 remain unanswered. What role dofloods of different magnitude play in the process ofchannel avulsion? How do pre-existing fan morphol-ogy and drainage patterns control the avulsion pro-cess? In addition, should the locations of future

Žchannels on shorter time scales 10s to 100s of.years be considered indeterminate as they are on

longer time scales? This paper addresses these ques-tions by examining the influence of both large- andsmall-scale floods on channel morphology and chan-nel avulsion processes on five fluvial fans in south-

Ž .ern Arizona Fig. 1; Table 1 . Documentation ofrecent channel changes on the five fans is used toconstruct a model of channel avulsion that linkschannel morphology to different stages of channeldevelopment and the avulsion process. Channel mor-phology can thus be used to identify areas on a fansurface prone to avulsion and target areas for morequantitative hydraulic analysis of flood hazards.

2. Site description

The five alluvial fans are located in a tectonicallyinactive portion of the Basin-and-Range province insouthern Arizona with source areas largely com-

Ž .posed of felsic intrusive rocks Table 1 . The climateis semi-arid with mean annual rainfall ranging from15 to 28 cm, increasing to the southeast. The five

Ž .fans are active secondary fans Blissenbach, 1954forming at the downstream termini of fanheadtrenches passing through abandoned Pleistocene sur-faces higher on the respective piedmonts. Fan de-

Table 1Selected drainage-basin characteristics for the five southern Arizona study fans

Alluvial fan Location of fan apex Total drainage Distance fan apex Dominant Average annual2X Ž . Ž . Ž .area km to mtn front km lithology rainfall cmUSGS 7.5 Quad UTM Coordinates

35 4Ruelas Fan Ruelas Canyon 91 200r 92 200 9.3 1.8 granite, granodiorite 2835 4Wild Burro Fan Ruelas Canyon 91 250r 90 500 18.5 4.3 granite, granodiorite 2835 4Cottonwood Fan Marana 92 600r 82 700 34.5 10.4 granite, granodiorite 2837 3White Tank Fan White Tank Mts. 09 750r 50 400 14.6 4 granite, gneiss 2037 2Tiger Wash Fan Weldon Hill 27 000r 85 050 249.6 6.5 Mix 15

( )J. FieldrGeomorphology 37 2001 93–104 95

Fig. 2. Schematic plan view and longitudinal profile of a discontinuous ephemeral-stream system showing relationship between sheetfloodzones, depositional and erosional channels, overland flow zones, and bank heights. Not to scale.

Žposits are predominately fine grained sand through.clay fraction and boulders are uncommon.

Detailed studies of surficial processes have beenŽcompleted on all five fans House et al., 1991, 1992;

.Maricopa County, 1992; Field, 1994 . Secondaryfans throughout southern Arizona are dominated byfluvial processes associated with discontinuousephemeral streams, a distinctive stream pattern char-acterized by alternating erosional and depositional

Žreaches Fig. 2; Schumm and Hadley, 1957; Bull,.1997 . Overland flow, a critical component of the

channel avulsion process described below, emanatesfrom the margins of the sheetflood zones and aggrad-

Ž .ing downstream ends of channels Fig. 2 . Channelbackfilling caused by the headward migration ofaggradational reaches can transform a deep channelinto an area of sheetflooding over periods of tens to

Ž .hundreds of years Bull, 1997 . The five alluvial fansstudied in this report show no evidence of debris-flowactivity; modern rates of weathering in the mountainranges of southern Arizona are insufficient to pro-duce debris flows with enough sediment to reach

Ždistant fan apices Table 1; Melton, 1965; Bull,.1991 .

3. Evidence of past channel avulsions on the allu-vial fans

Historical aerial photographs extending over 60years were used to establish the timing of recentchannel avulsions on the five alluvial fans by notingchanges in flow paths between pairs of aerial pho-

Ž .tographs taken on different years Fig. 3 . In addi-tion, aerial photographs and field reconnaissancewere used to document changes in channel morphol-ogy associated with each avulsion and to identifychannel reaches abandoned prior to the earliest pho-tographs. Avulsion is defined here as the diversion ofa majority of flow from one channel into another,leading to a total or partial abandonment of theprevious flow path. Avulsions that occurred at ornear the fan apex and which appeared to have di-verted more than 50% of the fan’s total discharge


Ž . Ž .Fig. 3. Aerial photographs of Ruelas Fan: a 1936 and b 1988 showing the avulsion that occurred between 1949 and 1956. The arrow, inthe same position on both photos, highlights the channel reach into which flow was diverted. The letter AaB, indicating the fan apex, is in thesame position on both photos. Note the bend created in the main channel when flow was diverted into the narrow preexisting channel.


were termed major avulsions; all others were consid-ered minor avulsions. All five fans show evidence ofat least one minor avulsion with evidence of a major

Ž .avulsion on four of the fans Table 2 . Channelsabandoned by major avulsions prior to the photo-graphic record are identified by vegetation growth inand minor incision of the former channel bottoms.

The following observations are critical for under-standing the avulsion mechanisms on the five allu-vial fans.

Ž .i All of the avulsions, major or minor, happenedalong aggrading channel reaches or sheetflood zonesŽ .Fig. 2; Table 2 . The location of the avulsions alongthese discontinuous ephemeral stream systems corre-sponds to areas where channel banks were low or

Ž .nonexistent Table 2 .Ž .ii Channel avulsions preferentially occurred on

Ž .the outside bends of channels Fig. 4; Table 2 , but

some avulsions actually resulted in the creation ofŽ .bends along previously straight reaches Fig. 3 .

Ž .iii Following an avulsion, the new flow paths, inall cases, followed the preexisting course of a smalltributary channel draining only a small portion of the

Ž .fan surface itself Figs. 3 and 4; Table 2 .Ž .iv The increases in discharge, as the channels

receiving the diverted flow were incorporated intothe main flow path, resulted in greater than three-foldincreases in channel width along some channel

Ž .reaches Figs. 3 and 5 .Ž .v The bed elevation of the new flow path fol-

lowing the avulsion was generally lower than, but ina few cases the same as, the bed elevation of the

Ž .abandoned reach Table 2 .Ž .vi Wide channel reaches abandoned by avul-

sions sometimes experienced minor incision with anarrower channel inset into the old channel bed.

Table 2Characteristics of past avulsions on the five alluvial fans

c Ž .Avulsion Year Avulsion type Preexisting Avulsion Bank height cm Change in beda b d Ž .channel? at bend? elevation cmMajor Minor At avulsion Max. on fan

Ruelas fanRF-1 1949–1956 X Yes Yes 8 123 6RF-2 Before 1936 X Unknown Yes 21 123 12RF-3 Before 1936 X Unknown No 0 123 5

Wild Burro fanWB-1 Before 1936 X Unknown No 19 136 70

Cottonwood fanCF-1 1962 X Yes No 32 194 70CF-2 1949–1956 X Yes No 0 194 4CF-3 1936–1949 X Yes Yes 24 194 40CF-4 1936–1949 X Yes Yes 0 194 0CF-5 Before 1936 X Yes No 0 194 36

White Tank fanWT-1 1951 X Yes Yes 9 54 10

Tiger Wash faneŽ .TF-1 1900 ? –1953 X Unknown No 0 142 4

TF-2 Before 1953 X Unknown No 12 142 91

a Indicates whether flow was diverted into an already existing channel. For diversions prior to photographic record, this can bedetermined by continuity of new channel with a tributary channel upstream of avulsion site.

b Indicates whether diversion occurred on the outside of a channel bend.cComparison of bank height at point of avulsion to maximum bank height on fan to illustrate how avulsions occur where bank heights

are low.d Taken as the difference in bed elevation between the new channel and abandoned channel at the point of avulsion.eTin cans circa 1900 transported in now abandoned channel.


Ž . Ž .Fig. 4. Channel network on White Tank Fan in a 1942 and b1992. Channel patterns were traced from aerial photographs and

Ž .the designation of channel types e.g., active channels based ongeomorphic maps constructed in the field. Note that the majoravulsion at the fan apex in 1951 occurred on the outside bend ofthe channel and activated large portions of the western half of thefan surface. Arrows point to small tributary channels heading onthe fan surface that were incorporated into the main flow pathfollowing the 1951 avulsion.

Fig. 5. Graph showing the increase in channel width following anavulsion along channel reaches capturing flow from the mainchannel. Points plotting to the left of the 2= and 3= linerepresent channels that more than doubled and tripled in width,respectively. Avulsion names labeled on each point refer to theavulsions listed in Table 2. Note that a single avulsion can causewidening of several channel reaches. The CF-1 avulsion did notcause widening of the reach because flow was diverted into alarge channel already connected to the main flow path.

Vegetation growing in the abandoned reaches givesrise to increasing gray tones on subsequent aerial

Ž .photographs Fig. 3 .Ž .vii Five distinctive channel morphologies are

present on the alluvial fans and are associated withdifferent stages of channel development preceding

Ž .and following an avulsion Fig. 6 .Except for a short gaging record of the Tiger

Wash Fan drainage basin, no gaged records of dis-charges exist for the five alluvial fans. Consequently,the record of floods since the beginning of thephotographic record is spotty and pieced togetherfrom field evidence, eyewitness accounts, and a few

Ž .published reports Table 3 . Paleohydrologic analy-ses in the Wild Burro Fan and White Tank MountainFan drainages and the gaged record from Tiger WashFan provide estimated discharges and recurrence in-

Ž .tervals for some of the known floods Table 3 .While significant floods likely precipitated all of theavulsions, this cannot be conclusively demonstratedbecause of the incomplete record. Of important note,however, is that no avulsions resulted from an ex-treme flood on Wild Burro Fan in 1988 or frommoderate floods on Cottonwood Fan in 1990 and


Fig. 6. Five channel morphologies observed on the five fluvial fans. Note that a single channel reach can go through each phase of channeldevelopment through time, and each stage of development may be observed at different places on a fan at the same time. During the activechannel phase, the channel reach is connected to the upper drainage basin. Aggrading channels are unstable and prone to avulsion whileAnewB channels and adjusted channels are more stable.

Ž .Tiger Wash Fan in 1970 Table 3 . The lack ofchange resulting from these floods is of particularinterest since small to moderate floods on fluvialfans in other regions have caused major avulsionsŽGriffiths and McSaveney, 1986; Kesel and Lowe,

.1987; Wells and Dorr, 1987 .

4. Processes of channel avulsion

Based on the observations described above, amodel has been developed to illustrate the processes

preceding, accompanying, and following channel di-Ž .version on the five fans Fig. 7 . The active distribu-

tary channels conveying flow from the upper drainagebasin occupy only a small portion of a fan surface atany one time, so numerous narrow Aon-fanB chan-nels are formed that drain small portions of the

Ž .inactive fan surface Fig. 6a ; in places, an incipientŽ .dendritic drainage system develops Fig. 4a . Por-

tions of these Aon-fanB channels occasionally ap-Ž .proach the larger distributary channels Fig. 7a and

can be incorporated into the main drainage net as theŽ .result of an avulsion Figs. 3, 4, and 7b . The banks

Table 3Record of floods on the five alluvial fans

Alluvial fan Year of flood Recurrence interval Associated Type of evidence Refs.Ž .of flood years avulsion for flood occurrence

Ruelas No recordŽ .Wild Burro 1988 )100 none flood debrisr House 1991

sand sheetsŽ .Cottonwood 1962 50–100 CF-1 written and Rostvedt 1968

oral accountsŽ .1990 5–15 none flood debris Field 1994

Ž .White Tank 1951 )100 WT-1 aerial photosr Kangieser 1969written account

Ž .1992 10–25 none flood debris Maricopa County 1992Ž .Tiger Wash 1970 )10 none stream gage Maricopa County 1992


Fig. 7. Model of channel avulsion developed from observations onthe five fluvial fans. See text for description. View is lookingupstream. Note relationship with five channel morphologies illus-trated in Fig. 6.

of the smaller channel that has captured the flow arewidened and become vertical in response to thedramatic increase in discharge from the upper

Ž .drainage basin Figs. 5 and 6b . The old and newsegments of the main channel, joined at the point of

Ž .avulsion, have contrasting morphologies Fig. 7b .The channel banks along the older upstream segment

are low, rounded, and vegetated, while downstreamof the diversion point the banks are vertical. Widen-ing and perhaps deepening of the newer downstreamchannel reach will continue as floods larger than the

Ždiversion event pass through the reach Figs. 6c and.7c . An avulsion is unlikely in reaches at this stage

of channel development because the channel is ad-justed to convey large flows.

As the frequency of floods large enough to con-tinue channel enlargement decreases, smaller floodswill begin to modify the channel’s morphology. In

Ž .channels with high width:depth ratios Fig. 6c ,transmission losses are maximized and only thelargest floods can transport the imposed sedimentload through the channel. Ephemeral streams carry-ing fine-grained sediments characteristically aggradeduring smaller flows because stream discharge infil-trates into the sandy channel bottoms before reaching

Ž .the channel mouth Bull, 1979, 1997 . Channels withgreater width:depth ratios also have a greater hy-draulic roughness, further promoting channel aggra-dation. In the absence of a flood large enough toflush accumulating sediments through the channelreach, a succession of small floods will eventuallyreduce the height of the channel banks through a rise

Ž .in the bed elevation Figs. 6d and 7d . The carryingcapacity of the channel is further reduced by theslumping, rounding, and revegetation of the increas-ingly stabilized channel banks. Floods at this stage ofchannel development begin to overtop the channelbanks, generating overland flow. With continuedaggradation, sheetflood zones may ultimately de-velop; and the magnitude of the smallest flood neededto produce overland flow decreases.

Overland flow is a critical component of thechannel avulsion process on the five study fans.Streamflows overtopping the banks of a channel tapthe upper portion of the water column and are thusrelatively clear, devoid of bed load, and capable of

Žerosion Hooke and Rohrer, 1979; Slingerland and.Smith, 1998 . Any sediment carried by the overland

flow is deposited rapidly at the channel and sheet-flood zone margins in response to the flow expan-

Ž .sion Fig. 8 . Although overland flow initially di-verges and spreads out over the fan surface, it ulti-mately recollects into the existing network of small

Žnarrow channels draining the fan surface Figs. 4b.and 8 . Consequently, overland flow enters and en-


Fig. 8. Simplified geomorphic map of Wild Burro Fan showing channel network, sand sheets, and the position of a sheetflood zone in 1936and 1988. Map is drawn from 1990 aerial photograph and based on field observations and analysis of historical aerial photographs. Sandsheets were deposited due to flow expansion during the 1988 flood when the banks along the main channel were overtopped. Note how sandsheets reconverge downstream as overland flow was captured by small channels draining the fan surface. The upstream migration of thesheetflood zone between 1936 and 1988 resulted in the widening of Reach A as more overland flow was generated further upstream alongthe main channel.

larges these small channels and generates headwarderosion directed towards the aggrading active chan-

Ž .nel Fig. 7d . When the channel heading on the fansurface is eroded back to the aggrading reach, streamcapture will occur because the bed of the headwarderoding channel is generally lower than the back-

Ž .filled channel Fig. 7e; Table 2 . The capturingchannel is, in some respects, a passive partner withoverland flow from the main channel dictating whereand how rapidly headward erosion occurs. Since thebed of the previously abandoned channels remainslower than the surrounding fan surface, overlandflow often enters these reaches and incises the old

Ž .channel bed Figs. 6e and 7d–e . Previously aban-doned channels can, at times, become reactivatedinto the main flow path as at the apex of White Tank

Ž .Fan Fig. 4 .

5. Rate of channel avulsion

The number of years required for a single channelreach to progress through the channel avulsion pro-

cess illustrated in Fig. 7 is dependent on severalŽ . Ž .factors: i the initial depth of the main channel; ii

the magnitude of the largest flood occurring duringŽ . Ž .the active channel phase Figs. 7b–d ; iii the se-

Ž .quencing of flood magnitudes; and iv the locationof the Aon-fanB channels draining the fan surfacewith respect to sheetflood zones and aggrading chan-nels along the active channel.

First, the depth to which a new channel incorpo-rated into the main flow path is eroded will deter-mine in part the amount of aggradation and timerequired before overland flow is initiated. Second, anextremely large flood flowing down an AadjustedB

Ž .channel Fig. 6c could conceivably overwhelm thechannel’s capacity at any time and produce enoughoverland flow in a single event to precipitate achannel avulsion. However, an aggrading channel or

Žsheetflood zone elsewhere on the fan Figs. 2 and.6d during the same flood would be the more likely

site of a diversion since more overland flow wouldemanate from these areas.

Third, a series of moderate to large floods duringŽ .the early stages of channel development Fig. 6b–c


may actually prolong the diversion process by peri-odically flushing sediment out of a channel reachand hindering the generation of overland flow. Incontrast, considerable aggradation resulting from anuninterrupted sequence of small sediment-chargedflows will eventually lead to overland flooding dur-ing even the smallest discharges. Diversions thathave occurred during small floods on fluvial fans

Želsewhere Griffiths and McSaveney, 1986; Kesel.and Lowe, 1987; Wells and Dorr, 1987 fit within

this context and should not be considered anomalousevents. While extended periods of constant discharge

Žcan produce avulsions in experimental studies Bryant.et al., 1995 , a similar uninterrupted sequence of

small floods is unlikely in arid and semi-arid regionswhere record discharges can be hundreds of times

Ž .larger than mean annual discharge Graf, 1988 . Aseries of small floods on fluvial fans in semi-aridregions likely hasten the diversion process and in-crease the effectiveness of extreme events by raisingthe channel bed through aggradation.

Finally, overland flow is more likely to result in achannel avulsion if a nearby channel is receiving theflow. Headward erosion towards the main channelwill not occur if overland flow does not reconvergeinto a small channel draining the fan surface. As

Ž .parts of discontinuous ephemeral systems Fig. 2 ,aggrading depositional channels on the five alluvialfans migrate headward; and overland flow is gener-ated at different points along the channel throughtime. Eventually the aggrading portion of the mainchannel will approach an area where a small channeldraining the fan surface can capture overland flow.The headward migration of a sheetflood zone onWild Burro Fan between 1936 and 1988 coincideswith the widening of Reach A on Wild Burro FanŽ .Fig. 8 . As the sheetflood zone migrated upstreamand backfilled the main channel, more and moreoverland flow entered Reach A. The widening maynot have occurred without the upstream shift in thelocation of the sheetflood zone.

6. Discussion

If geomorphic effectiveness is defined as the abil-ity of an event to shape or form the landscapeŽ .Wolman and Gerson, 1978 , then channel avulsions

must be considered an effective process on alluvialfans. Large infrequent floods are widely regarded asthe effective geomorphological agents of change in

Žarid climates Wolman and Miller, 1960; Baker,.1977; Wolman and Gerson, 1978; Kochel, 1988 , but

the fine-grained nature of the five study fans meansthat small to moderate events can transport sedimentand play an active role in landscape modification.Although large flows appear to have directly causedall of the avulsions on the five fans, this should notimply that small events are ineffective agents ofchange. Channel aggradation resulting from smallflows decreases bank heights and increases theamount of overland flow produced by subsequentflows, thereby setting the stage for and, to someextent, dictating the location of future avulsions.Without the action of small aggrading flows, thegeomorphic effectiveness of large events would begreatly diminished on the five fans.

Alluvial-fan flooding is of increasing concern inthe rapidly urbanizing southwestern United States,where over 30% of the landscape is composed of

Ž .alluvial fans Antsey, 1965 . The present FederalEmergency Management Agency method of flood-hazard assessment on alluvial fans, based on a

Ž .stochastic procedure proposed by Dawdy 1979 ,assumes that channel positions move across an allu-vial fan surface at random during each flood without

Ž .regard to preexisting flow paths FEMA, 1995 .However, channel avulsions on the five study fansare not completely random events over the shortterm, since avulsions invariably occur where channelbanks are low and divert flow into preexisting chan-

Ž .nels Table 2 . The diversion of flow into preexistingchannels following an avulsion appears to be typical

Žof fluvial fans in other regions as well Kesel, 1985;Griffiths and McSaveney, 1986; Richards et al., 1987;

.Wells and Dorr, 1987; McCarthy et al., 1992 . TheFederal Emergency Management Agency methodalso does not account for the fact that large floods onfluvial fans do not always produce avulsions or

Žmajor channel adjustments Table 3; Harvey, 1984;.Ribble, 1988 . While the assumption of randomness

in the Federal Emergency Management Agencymethod for assessing flood risk on alluvial fans mayappear safely conservative by assuming the wholefan surface is equally prone to avulsion, the methodseverely underestimates flow depth and velocity


along existing channels where flow is most likely toŽoccur O’Brien and Fullerton, 1990; House et al.,

.1992; O’Brien and Fuller, 1993; Field, 1994 .Recognizing the deficiencies in the Federal Emer-

gency Management Agency method, a recent reportŽ .by the National Research Council NRC, 1996 rec-

ommends site-specific assessment of alluvial-fanflood hazards. The use of detailed hydraulic model-ing along channel reaches identified as probableavulsion sites on alluvial fans has been suggested

Ž .previously Gundlach, 1974; Richards et al., 1987 ,and recent advances in quantitative modeling of

Žavulsions along meandering rivers Slingerland and.Smith, 1998 may also prove useful on alluvial fans.

The model presented in Fig. 7 is useful for identify-ing those channel reaches most prone to an avulsionas the channel morphology of a given reach on afluvial fan indicates the susceptibility of that reach to

Ž .diversion and abandonment Fig. 6 . Site-specificmaps of fluvial fans can focus the attention of land-use planners and hydraulic modelers on specific fanareas such as aggrading reaches with low banks andnearby small Aon-fanB channels that might captureoverland flow. Through time, a channel’s morphol-ogy will evolve as aggrading reaches and sheetfloodzones in a discontinuous ephemeral stream systemmigrate headward in response to small sediment-

Ž .charged flows Fig. 8 . As site-specific studies offlood hazards on fluvial fans are undertaken, periodicmonitoring will be necessary in order to identifychanges in flood hazards brought about not only bylarge floods but also seemingly ineffective smallfloods.

7. Conclusions

A descriptive model has been developed that elu-cidates the processes of channel avulsion on fluvially

Ž .dominated alluvial fans in southern Arizona Fig. 7 .After a period of bank widening along a new channelreach, aggradation begins due to small floods that areunable to transport the imposed sediment loadthrough the stream system. The resulting decrease inbank height leads to the generation of overland flow,which accelerates and directs the headward erosionof small channels heading on the fan surface. Streamcapture ultimately occurs and the channel receiving

the diverted flow is rapidly transformed as it isincorporated into the main channel network. Themodel is similar to avulsion and meander neck cutoff

Žmodels developed for mega-fans Wells and Dorr,.1987; McCarthy et al., 1992 and semi-arid river

Ž .systems Schumann, 1989; Gay et al., 1998 , sug-gesting that the avulsion processes outlined hereoperate at different scales and in varied environmen-tal settings.

The avulsion process described here demonstrateshow small floods, under certain circumstances, cancause an avulsion while large floods often do notprecipitate changes. Small floods on the southernArizona fans have not directly caused an avulsion,but they do potentially accelerate the avulsion pro-cess culminated by large floods. The importance ofboth small and large floods on channel avulsionunderscores the need for geomorphologists to focuson how different scales of geomorphological eventswork in concert to modify the landscape rather thanon which scale of events is the most geomorphologi-cally effective.

Although channel avulsions occur suddenly andare clearly a hazard on alluvial fans, the location ofchannel avulsions on fluvial fans should no longer beregarded as unpredictable, since new channels almostinvariably follow preexisting flow paths. A carefulanalysis of all existing flow paths in relationship topotentially unstable reaches along the active channelŽ .i.e., low channel banks may help identify the loca-tion of future channel positions. The most likelypaths of future channels on the five southern Arizona

Ž .fans have been noted elsewhere Field, 1994 , andfollow-up studies will establish the accuracy of thesepredictions. The results of this study suggest thatunderstanding the interplay between large andsmall-scale geomorphic events is critical for antici-pating the timing and location of future avulsions onsome, if not most, fluvially dominated alluvial fans.

References

Antsey, R.L., 1965. Physical Characteristics of Alluvial Fans. USArmy Natick Laboratories, Technical Report ES-20, 109 pp.

Baker, V.R., 1977. Stream-channel response to floods, with exam-ples from central Texas. Geol. Soc. Am. Bull. 88, 1057–1071.

Beaty, C.B., 1963. Origin of alluvial fans, White Mountains,California and Nevada. Ann. Assoc. Am. Geogr. 53, 516–535.


Blissenbach, E., 1954. Geology of alluvial fans in semi-aridregions. Geol. Soc. of Am. Bull. 65, 175–190.

Bryant, M., Falk, P., Paola, C., 1995. Experimental study ofavulsion frequency and rate of deposition. Geology 23, 365–368.

Bull, W.B., 1979. Threshold of critical power in streams. Geol.Soc. Am. Bull. 90, 453–464.

Bull, W.B., 1991. Geomorphological Response to Climate Change.Oxford Univ. Press, NY, 326 pp.

Bull, W.B., 1997. Discontinuous ephemeral streams. Geomorphol-ogy 19, 227–276.

Dawdy, D.R., 1979. Flood frequency estimates on alluvial fans.ASCE J. Hydraul. Div. 105, 1407–1413.

Drew, F., 1873. Alluvial and lacustrine deposits and glacialrecords of the Upper Indus basin. Q.J. Geol. Soc. London 29,441–471.

Ž .Federal Emergency Management Agency FEMA , 1995. Ap-pendix 5: Studies of alluvial fan flooding. Guidelines andspecifications for study contractors. Document No. 37, Wash-ington DC.

Field, J.J., 1994. Surficial processes, channel change, and geologi-cal methods of flood hazard assessment on fluvially dominatedalluvial fans in Arizona, PhD Thesis, Univ. of Arizona, Tuc-son.

Gay, G.R., Gay, H.H., Gay, W.H., Martinson, H.A., Meade, R.H.,Moody, J.A., 1998. Evolution of cutoffs across meander necksin Powder River, Montana, USA. Earth Surf. Processes Land-forms 23, 651–662.

Graf, W.L., 1988. Fluvial Processes in Dryland Rivers. Springer-Verlag, New York, 346 pp.

Griffiths, G.A., McSaveney, M.J., 1986. Sedimentation and rivercontainment on Waitangitaona alluvial fan — South Westland,New Zealand. Z. Geomorphol. N.F. 30, 215–230.

w xGundlach, D., 1974. Flood plain analysis for alluvial fans abs. .Am. Geophys. Union Trans. 56, 1122.

Harvey, A.M., 1984. Geomorphological response to an extremeflood: a case from southeast Spain. Earth Surf. ProcessesLandforms 9, 267–279.

Hooke, R.L., Rohrer, W.L., 1979. Geometry of alluvial fans:effect of discharge and sediment size. Earth Surf. Processes 4,147–166.

House, P.K., 1991. Paleoflood hydrology of the principal canyonsof the Southern Tortolita Mountains, southeastern Arizona.MS Thesis Prepublication Manuscript, Univ. of Arizona, Tuc-son.

House, P.K., Pearthree, P.A., Vincent, K.R., 1991. Flow patterns,flow hydraulics, and flood-hazard implications of a recent

w xextreme alluvial-fan flood in southern Arizona abs. . Geol.Soc. Am. Abs. Programs 23, A121.

House, P.K., Pearthree, P.A., Vincent, K.R., 1992. Geologicalinsights into deficiencies in federal procedures for flood haz-

w xard assessment on alluvial fans abs. . Geol. Soc. Am. Abs.Programs 24, A284.

Kangieser, P.C., 1969. Major rainstorms and snowstorms in Ari-zona 1897–1969. Unpublished Report, Arizona Weather Bu-reau, Tempe, AZ, 25 pp.

Kesel, R.H., 1985. Alluvial fan systems in a wet-tropical environ-ment, Costa Rica. Natl. Geogr. Res. 1, 450–469.

Kesel, R.H., Lowe, D.R., 1987. Geomorphology and sedimentol-ogy of the Toro Amarillo alluvial fan in a humid tropicalenvironment, Costa Rica. Geografis. Ann. 69A, 85–99.

Kochel, R.C., 1988. Geomorphological impact on large floods:review and new perspectives on magnitude and frequency. In:

Ž .Baker, V.R., Kochel, R.C., Patton, P.C. Eds. , Flood Geomor-phology. Wiley, New York, pp. 169–188.

Maricopa County, 1992. Maricopa County alluvial fan data collec-tion and monitoring study, 1992. Flood Control District ofMaricopa County, Phoenix, AZ.

McCarthy, T.S., Ellery, W.N., Stanistreet, I.G., 1992. Avulsionmechanisms on the Okavango fan, Botswana: the control of afluvial system by vegetation. Sedimenttology 39, 779–795.

Melton, M.A., 1965. The geomorphological and paleoclimaticsignificance of alluvial deposits in southern Arizona. J. Geol.73, 1–38.

Ž .National Research Council NRC , 1996. Alluvial Fan Flooding.National Academy Press, Washington, DC, 172 pp.

O’Brien, J.S., Fuller, J.E., 1993. Discussion on APreferred direc-tions of flow on alluvial fansB. J. Hydraul. Eng. 119, 1316–1318.

O’Brien, J.S., Fullerton, W.T., 1990. Two dimensional modelingŽ .of alluvial fan flows. In: French, R.H. Ed. , HydraulicsrHy-

drology of Arid Lands. American Society of Civil Engineers,New York, pp. 262–267.

Price Jr., W.E., 1974. Simulation of alluvial fan deposition by arandom walk model. Water Resour. Res. 10, 263–274.

Ribble, G.E., 1988. Scodie Canyon flash floods of 1984. In: Abt,Ž .S.R., Gessler, J. Eds. , Hydraulic Engineering. American

Society of Civil Engineers, New York, pp. 258–263.Richards, K.S., Brunsden, D., Jones, D.K.C., McCraig, M., 1987.

Applied fluvial geomorphology: river engineering project ap-Ž .praisal in its geomorphological context. In: Richards, K. Ed. ,

River Channels: Environment and Process. Blackwell, Basil,Switzerland, pp. 352–382.

Rostvedt, J.O., 1968. Summary of floods in the United Statesduring 1962. US Geological Survey Water-Supply Paper 1820.

Schumann, R.R., 1989. Morphology of Red Creek Wyoming, anarid-region anastomosing channel system. Earth Surf. Pro-cesses Landforms 14, 277–288.

Schumm, S.A., Hadley, R.F., 1957. Arroyos and the semiaridcycle of erosion. Am. J. Sci. 255, 161–174.

Slingerland, S., Smith, N.D., 1998. Necessary conditions for ameandering-river avulsion. Geology 26, 435–438.

Wells, N.A., Dorr Jr., J.A., 1987. Shifting of the Kosi River,northern India. Geology 15, 204–207.

Whipple, K.X., Dunne, T., 1992. The influence of debris-flowrheology on fan morphology, Owens Valley, California. Geol.Soc. Am. Bull. 104, 887–900.

Wolman, M.G., Gerson, R., 1978. Relative scales of time andeffectiveness of climate in watershed geomorphology. EarthSurf. Processes 3, 189–208.

Wolman, M.G., Miller, J.P., 1960. Magnitude and frequency offorces in geomorphological processes. J. Geol. 68, 54–74.

channel avulsion on alluvial fans in southern arizona

Documents