7.a hydrologic and hydraulic processes

77.A Hydrologic and Hydraulic Processes

• How does the stream flow and why is this understanding important? • Is streamflow perennial, ephemeral or intermittent?• What is the discharge, frequency and duration of extreme high and low flows?• How often does the stream flood?• How does roughness affect flow levels?• What is the discharge most effective in maintaining the stream channel under

equilibrium conditions?• How does one determine if equilibrium conditions exist?• What field measurements are necessary?

7.B Geomorphic Processes• How do I inventory geomorphic information on streams and use it to understand and

develop physically appropriate restoration plans?• How do I interpret the dominant channel adjustment processes active at the site?• How deep and wide should a stream be?• Is the stream stable?• Are basin-wide adjustments occurring, or is this a local problem?• Are channel banks stable, at-risk, or unstable?• What measurements are necessary?

7.C Chemical Processes• How do you measure the condition of the physical and chemical conditions within a

stream corridor?• Why is quality assurance an important component of stream corridor analysis activities?• What are some of the water quality models that can be used to evaluate water

chemistry data?

7.D Biological Characteristics• What are some important considerations in using biological indicators for analyzing

stream corridor conditions?• Which indicators have been used successfully?• What role do habitat surveys play in analyzing the biological condition of the

stream corridor?• How do you measure biological diversity in a stream corridor?• What is the role of stream classification systems in analyzing stream corridor conditions?• How can models be used to evaluate the biological condition of a stream corridor?• What are the characteristics of models that have been used to evaluate stream

corridor conditions?

7Section 7.A: Hydrologic Processes

Understanding how water flows into andthrough stream corridors is critical to de-veloping restoration initiatives. How fast,how much, how deep, how often, andwhen water flows are important basicquestions that must be answered in orderto make appropriate decisions about theimplementation of a stream corridor’srestoration.

Section 7.B: Geomorphic Processes

This section combines the basic hydrologicprocesses with the physical or geomorphicfunctions and characteristics. Water flows

through streams but is affected by thekinds of soils and alluvial features withinthe channel, in the floodplain, and in theuplands. The amount and kind of sedi-ments carried by a stream is largely a de-terminant of its equilibrium characteristics,including size, shape, and profile. Success-ful implementation of the stream corridorrestoration, whether active (requiring di-rect intervention) or passive, (removingonly disturbance factors), depends on anunderstanding of how water and sedi-ment are related to channel form andfunction, and on what processes are in-volved with channel evolution.

7.A Hydrologic Processes

7.B Geomorphic Processes

7.C Chemical Characteristics

7.D Biological Characteristics

7–2 Chapter 7: Analysis of Corridor Condition

Section 7.C: ChemicalCharacteristics

The quality of water in the streamcorridor is normally a primary ob-jective of restoration, either to im-prove it to a desired condition, orto sustain it. Restoration initiativesshould consider the physical andchemical characteristics that maynot be readily apparent but that arenonetheless critical to the functionsand processes of stream corridors.Chemical manipulation of specificcharacteristics usually involves themanagement or alteration of ele-ments in the landscape or corridor.

Section 7.D: BiologicalCharacteristics

The fish, wildlife, plants, andhuman beings that use, live in, orjust visit the stream corridor are keyelements to consider, not only interms of increasing populations orspecies diversity, but also in termsof usually being one of the primarygoals of the restoration effort. Athorough understanding of howwater flows, how sediment is trans-ported, and how geomorphic fea-tures and processes evolve isimportant. However, a prerequisiteto successful restoration is an un-derstanding of the living parts ofthe system and how the physicaland chemical processes affect thestream corridor.

Hydrologic Processes 7–3

Flow Analysis

Restoring stream structure and functionrequires knowledge of flow characteris-tics. At a minimum, it is helpful toknow whether the stream is perennial,intermittent, or ephemeral, and the rel-ative contributions of baseflow andstormflow in the annual runoff. Itmight also be helpful to know whetherstreamflow is derived primarily fromrainfall, snowmelt, or a combination ofthe two.

Other desirable information includesthe relative frequency and duration ofextreme high and low flows for the siteand the duration of certain stream flowlevels. High and low flow extremes usu-ally are described with a statistical pro-cedure called a frequency analysis, andthe amount of time that various flowlevels are present is usually describedwith a flow duration curve.

Finally, it is often desirable to estimate the channel-forming or domi-nant discharge for a stream (i.e., thedischarge that is most effective in shaping and maintaining the naturalstream channel). Channel-forming ordominant discharge is used for designwhen the restoration includes channel reconstruction.

Estimates of streamflow characteristicsneeded for restoration can be obtainedfrom stream gauge data. Procedures fordetermining flow duration characteris-tics and the magnitude and frequencyof floods and low flows at gauged sitesare described in this section. The pro-cedures are illustrated using daily mean flows and annual peak flows (the maximum discharge for each year)for the Scott River near Fort Jones, a653-square-mile watershed in northernCalifornia.

Most stream corridor restoration initia-tives are on streams or reaches that lacksystematic stream gauge data. Therefore,estimates of flow duration and the fre-quency of extreme high and low flowsmust be based on indirect methodsfrom regional hydrologic analysis. Sev-eral methods are available for indirectestimation of mean annual flow andflood characteristics; however, fewmethods have been developed for esti-mating low flows and general flow du-ration characteristics.

Users are cautioned that statisticalanalyses using historical streamflowdata need to account for watershedchanges that might have occurred dur-ing the period of record. Many basinsin the United States have experiencedsubstantial urbanization and develop-ment; construction of upstream reser-voirs, dams, and storm watermanagement structures; and construc-tion of levees or channel modifications.These features have a direct impact onthe statistical analyses of the data forpeak flows, and for low flows and flowduration curves in some instances. De-pending on basin modifications andthe analyses to be performed, this couldrequire substantial time and effort.

Flow Duration

The amount of time certain flow levelsexist in the stream is represented by aflow duration curve which depicts thepercentage of time a given streamflowwas equaled or exceeded over a givenperiod. Flow duration curves are usuallybased on daily streamflow (a recordcontaining the average flow for eachday) and describe the flow characteris-tics of a stream throughout a range ofdischarges without regard to the se-quence of occurrence. A flow duration

7.A Hydrologic Processes


curve is the cumulative histogram of theset of all daily flows. The constructionof flow duration curves is described bySearcy (1959), who recommends defin-ing the cumulative histogram of stream-flow by using 25 to 35 well-distributedclass intervals of streamflow data.

Figure 7.1 is a flow duration curve thatwas defined using 34 class intervals andsoftware documented by Lumb et al.(1990). The numerical output is pro-vided in the accompanying table.

The curve shows that a daily mean flowof 1,100 cubic feet per second (cfs) isexceeded about 20 percent of the timeor by about 20 percent of the observeddaily flows. The long-term mean dailyflow (the average flow for the period ofrecord) for this watershed was deter-mined to be 623 cfs. The duration curveshows that this flow is exceeded about38 percent of the time.

For over half the states, the USGS haspublished reports for estimating flowduration percentiles and low flows atungauged locations. Estimating flowduration characteristics at ungaugedsites usually is attempted by adjustingdata from a nearby stream gauge in ahydrologically similar basin. Flow dura-tion characteristics from the streamgauge record are expressed per unit areaof drainage basin at the gauge (i.e., incfs/mi2) and are multiplied by thedrainage area of the ungauged site toestimate flow duration characteristicsthere. The accuracy of such a procedureis directly related to the similarity of thetwo sites. Generally, the drainage area atthe stream gauge and ungauged sitesshould be fairly similar, and streamflowcharacteristics should be similar forboth sites. Additionally, mean basin ele-vation and physiography should besimilar for both sites. Such a proceduredoes not work well and should not beattempted in stream systems dominated

by local convective storm runoff orwhere land uses vary significantly be-tween the gauged and ungauged basins.

Flow Frequency Analysis

The frequency of floods and low flowsfor gauged sites is determined by ana-lyzing an annual time series of maxi-mum or minimum flow values (achronological list of the largest orsmallest flow that occurred each year).Although previously described in Chap-ter 1, flow frequency is redefined here be-cause of its relevance to the sectionsthat follow. Flow frequency is definedas the probability or percent chance ofa given flow’s being exceeded or not ex-ceeded in any given year. Flow fre-quency is often expressed in terms ofrecurrence interval or the average numberof years between exceeding or not ex-ceeding the given flows. For example, agiven flood flow that has a 100-year re-currence interval is expected to be ex-ceeded, on average, only once in any100-year period; that is, in any givenyear, the annual flood flow has a 1 per-cent chance or 0.01 probability of ex-ceeding the 100-year flood. Theexceedance probability, p, and the re-currence interval, T, are related in thatone is the reciprocal of the other (i.e.,T = 1/p). Statistical procedures for de-termining the frequency of floods andlow flows at gauged sites follow.

As mentioned earlier, most stream corri-dor restoration initiatives are onstreams or reaches lacking systematicstream gauge data; therefore, estimatesof flow duration characteristics and thefrequency of extreme high and extremelow flows must be based on indirectmethods from regional hydrologicanalysis.

Flood Frequency Analysis

Guidelines for determining the fre-quency of floods at a particular location


using streamflow records are docu-mented by the Hydrology Subcommit-tee of the Interagency AdvisoryCommittee on Water Data (IACWD1982, Bulletin 17B). The guidelines de-scribed in Bulletin 17B are used by allfederal agencies in planning activitiesinvolving water and related land re-sources. Bulletin 17B recommends fit-ting the Pearson Type III frequencydistribution to the logarithms of the an-nual peak flows using sample statistics(mean, standard deviation, and skew)to estimate the distribution parameters.Procedures for outlier detection and ad-justment, adjustment for historical data,development of generalized skew, andweighting of station and generalizedskews are provided. The station skew iscomputed from the observed peakflows, and the generalized skew is a re-gional estimate determined from esti-mates at several long-term stations inthe region. The US Army Corps of Engi-neers also has produced a user’s manualfor flood frequency analysis (Report CPD-13, 1994) that can aid in determiningflood frequency distribution parame-ters. NRCS has also produced a manual(National Engineering Handbook, Section4, Chapter 18) that can also be used indetermining flood frequency distribu-tion (USDA-SCS 1983).

Throughout the United States, flood fre-quency estimates for USGS gauging sta-tions have been correlated with certainclimatic and basin characteristics. Theresult is a set of regression equationsthat can be used to estimate flood mag-nitude for various return periods in un-gauged basins (Jennings et al. 1994).Reports outlining these equations oftenare prepared for state highway depart-ments to help them size culverts andrural road bridge openings.

Estimates of the frequency of peakflows at ungauged sites may be made byusing these regional regression equa-

River Basin

Southeastern PA

a

61

b

0.82

Upper Salmon River, ID 36 0.68

Upper Green River, WY 28 0.69

San Francisco Bay Region, CA

Qbf = aAb

53 0.93

Figure 7.1: Flow duration curve andassociated data tables.Data for the Scott River,near Fort Jones, CA,1951–1980, show thata flow of 1,100 cubicfeet per second (cfs)is exceeded about 20percent of the time.Source: Lumb et al. (1990).


Daily mean streamflow data needed for definingflow duration curves are published on a water-year (October 1 to September 30) basis for eachstate by the U.S. Geological Survey (USGS) in thereport series Water Resources Data. The data col-lected and published by the USGS are archived inthe National Water Information System (NWIS).

The USGS currently provides access to streamflowdata by means of the Internet. The USGS URLaddress for access to streamflow data ishttp://water.usgs.gov. Approximately 400,000 sta-tion years of historical daily mean flows for about18,500 stations are available through this source.The USGS data for the entire United States arealso available from commercial vendors on twoCD-ROMs, one for the eastern and one for thewestern half of the country (e.g., CD-ROMs forDOS can be obtained from Earth Info, and CD-ROMs for Windows can be obtained fromHydrosphere Data Products. Both companies arelocated in Boulder, Colorado.)

In addition to the daily mean flows, summary sta-tistics are also published for active streamflow sta-tions in the USGS annual Water Resources Datareports. Among the summary statistics are thedaily mean flows that are exceeded 10, 50, and90 percent of the time of record. These durations

are computed by ranking the observed daily meanflows from q

(1)to q

(n • 365)where n is the number of

years of record, q(1)

is the largest observation, andq

(365 • n)is the smallest observation. The ranked list

is called a set of ordered observations. The q(1)

thatare exceeded 10, 50, and 90 percent of the timeare then determined. Flow duration percentiles(quantiles) for gauged sites are also published byUSGS in reports on low flow frequency and otherstreamflow statistics (e.g., Atkins and Pearman1994, Zalants 1991, Telis 1991, and Ries 1994).

Annual peak flow data needed for flood frequen-cy analysis are also published by the USGS,archived in NWIS, and available through the inter-net at the URL address provided above. Flood fre-quency estimates at gauged sites are routinelypublished by USGS as part of cooperative studieswith state agencies to develop regional regressionequations for ungauged watersheds. Jennings etal. (1994) provide a nationwide summary of thecurrent USGS reports that summarize flood fre-quency estimates at gauged sites as well asregression equations for estimating flood peakflows for ungauged watersheds. Annual andpartial-duration (peaks-above-threshold) peak flowdata for all USGS gauges can be obtained on oneCD-ROM from commercial vendors.


tions, provided that the gauged and un-gauged sites have similar climatic andphysiographic characteristics.

Frequently the user needs only suchlimited information as mean annualprecipitation, drainage area, storage inlakes and wetlands, land use, major soiltypes, stream gradients, and a topo-graphic map to calculate flood magni-tudes at a site. Again, the accuracy ofthe procedure is directly related to thehydrologic similarity of the two sites.

Similarly, in many locations, flood fre-quency estimates from USGS gaugingstations have been correlated with cer-tain channel geometry characteristics.These correlations produce a set of re-gression equations relating some chan-nel feature, usually active channelwidth, to flood magnitudes for variousreturn periods. A review of these equa-tions is provided by Wharton (1995).Again, the standard errors of the esti-mate might be large.

Regardless of the procedure or source ofinformation chosen for obtaining floodfrequency information, estimates forthe 1.5, 2, 5, 10, 25, and (record per-mitting) 50 and 100-year flood eventsmay be plotted on standard log-probability paper, and a smooth curvemay be drawn between the points.(Note that these are flood events withprobabilities of 67, 50, 20, 10, 4, 2, and1 percent, respectively.) This plot be-comes the flood frequency relationshipfor the restoration site under considera-tion. It provides the background infor-mation for determining the frequencyof inundation of surfaces and vegeta-tion communities along the channel.

Low-Flow Frequency Analysis

Guidelines for low-flow frequency analysisare not as standardized as those forflood frequency analysis. No single fre-quency distribution or curve-fittingmethod has been generally accepted.

Vogel and Kroll (1989) provide a sum-mary of the limited number of studiesthat have evaluated frequency distribu-tions and fitting methods for low flows.The methodology used by USGS andUSEPA is described below.

The hypothetical daily hydrographshown in Figure 7.2 is typical of manyareas of the United States where the an-nual minimum flows occur in late sum-mer and early fall. The climatic year(April 1 to March 31) rather than thewater year is used in low-flow analysesso that the entire low-flow period iscontained within one year.

Data used in low-flow frequency analy-ses are typically the annual minimumaverage flow for a specified number ofconsecutive days. The annual minimum7- and 14-day low flows are illustratedin Figure 7.2. For example, the annualminimum 7-day flow is the annualminimum value of running 7-daymeans.

Flood frequency estimates also may be generated usingprecipitation data and applicable watershed runoff modelssuch as HEC-1, TR-20, and TR-55. The precipitation recordfor various return-period storm events is used by thewatershed model to generate a runoff hydrograph andpeak flow for that event. The modeled rainfall may befrom historical data or from an assumed time distributionof precipitation (e.g., a 2-year, 24-hour rainfall event). Thismethod of generating flood frequency estimates assumesthe return period of the runoff event equals the returnperiod of the precipitation event (e.g., a 2-year rainfallevent will generate a 2-year peak flow). The validity of thisassumption depends on antecedent moisture conditions,basin size, and a number of other factors.


USGS and USEPA recommend usingthe Pearson Type III distribution to thelogarithms of annual minimum d-daylow flows to obtain the flow with anonexceedance probability p (or recur-rence interval T = 1/p). The PearsonType III low-flow estimates are com-puted from the following equation:

Xd,T

= Md

– KTS

d

where:

Xd,T

= the logarithm of the annualminimum d-day low flow forwhich the flow is not exceededin 1 of T years or which has aprobability of p = 1/T of notbeing exceeded in any given year

Md

= the mean of the logarithms ofannual minimum d-day lowflows

Sd

= the standard deviation of thelogarithms of the annual mini-mum d-day low flows

KT

= the Pearson Type III frequencyfactor

The desired quantile, Qd,T

, can be ob-tained by taking the antilogarithm ofthe equation.

The 7-day, 10-year low flow (Q7,10

) isused by about half of the regulatoryagencies in the United States for man-aging water quality in receiving waters

(USEPA 1986, Riggs et al. 1980). Lowflows for other durations and frequen-cies are used in some states.

Computer software for performing low-flow analyses using a record of dailymean flows is documented by Hutchi-son (1975) and Lumb et al. (1990). Anexample of a low-flow frequency curvefor the annual minimum 7-day lowflow is given in Figure 7.3 for ScottRiver near Fort Jones, California, for thesame period (1951 to 1980) used in theflood frequency analyses above.

From Figure 7.3, one can determinethat the Q

7,10is about 20 cfs, which is

comparable to the 99th percentile(daily mean flow exceeded 99 percentof the time) of the flow duration curve(Figure 7.1). This comparison is consis-tent with findings of Fennessey andVogel (1990), who concluded that theQ

7,10from 23 rivers in Massachusetts

was approximately equal to the 99thflow duration percentile. The USGS rou-tinely publishes low flow estimates atgauged sites (Zalants 1991, Telis 1991,Atkins and Pearman 1994).

Following are discussions of differentways to look at the flows that tend toform and maintain streams. Restora-tions that include alterations of flows orchanges in the dimensions of thestream must include engineering analy-ses as described in Chapter 8.

Channel-forming Flow

The channel-forming or dominant dis-charge is a theoretical discharge that ifconstantly maintained in an alluvialstream over a long period of timewould produce the same channel geom-etry that is produced by the long-termnatural hydrograph. Channel-formingdischarge is the most commonly usedsingle independent variable that isfound to govern channel shape andform. Using a channel-forming dis-charge to design channel geometry is

August September

lowest average 14-day flow

lowest average 7-day flow

October

Dis

char

ge

(cfs

)

1

15

20

30

40

Figure 7.2: Annual hydrograph displaying lowflows. The daily mean flows on the lowest partof the annual hydrograph are averaged to givethe 7-day and 14-day low flows for that year.


not a universally accepted technique, al-though most river engineers and scien-tists agree that the concept has merit, atleast for perennial (humid and temper-ate) and perhaps ephemeral (semiarid)rivers. For arid channels, where runoff isgenerated by localized high-intensitystorms and the absence of vegetationensures that the channel will adjust toeach major flood event, the channel-forming discharge concept is generallynot applicable.

Natural alluvial rivers experience a widerange of discharges and may adjusttheir geometry to flow events of differ-ent magnitudes by mobilizing eitherbed or bank sediments. Although Wol-man and Miller (1960) noted that “it islogical to assume that the channelshape is affected by a range of flowsrather than a single discharge,” theyconcurred with the view put forwardearlier by civil engineers working on“regime theory” that the channel-forming or dominant discharge is thesteady flow that produces the samegross channel shapes and dimensions

as the natural sequence of events (Inglis1949). Wolman and Miller (1960) de-fined “moderate frequency” as eventsoccurring “at least once each year ortwo and in many cases several or moretimes per year.” They also consideredthe sediment load transported by agiven flow as a percentage of the totalamount of sediment carried by the riverduring the period of record. Their re-sults, for a variety of American rivers lo-cated in different climatic andphysiographic regions, showed that thegreater part (that is, 50 percent ormore) of the total sediment load wascarried by moderate flows rather thancatastrophic floods. Ninety percent ofthe load was carried by events with a re-turn period of less than 5 years. Theprecise form of the cumulative curve ac-tually depends on factors such as the

Flo

w C

har

acte

rist

ics

(cfs

)

Annual Nonexceedance Probability (percent)

7-day low flowLog-Pearson Type III

951

10

102

103

90 80 70 50 30 20 10 5

Figure 7.3: Annual minimum 7-day low flowfrequency curve. The Q

7,10on this graph is about

20 cfs. The annual minimum value of 7-dayrunning means for this gauge is about 10 percent.


predominant mode of transport (bedload, suspended load, or mixed load)and the flow variability, which is influ-enced by the size and hydrologic char-acteristics of the watershed. Smallwatersheds generally experience a widerrange of flows than large watersheds,and this tends to increase the propor-tion of sediment load carried by infre-quent events. Thorough reviews ofarguments about the conceptual basisof channel-forming discharge theorycan be found in textbooks by Richards(1982), Knighton (1984), and Summer-field (1991).

Researchers have used various dischargelevels to represent the channel-formingdischarge. The most common are (1)bankfull discharge, (2) a specific dis-charge recurrence interval from the an-nual peak or partial duration frequencycurves, and (3) effective discharge.These approaches are frequently usedand can produce a good approximationof the channel-forming discharge inmany situations; however, as discussedin the following paragraphs, consider-able uncertainties are involved in allthree of these approaches. Many practi-tioners are using specific approaches todetermine channel-forming dischargeand the response of stream corridors.Bibliographic information on thesemethods is available later in the document.

Because of the spatial variability withina given geographical region, the re-sponse of any particular stream corridorwithin the region can differ from thatexpected for the region as a whole. Thisis especially critical for streams drainingsmall, ungauged drainage areas. There-fore, the expected channel-forming dis-charge of ungauged areas should beestimated by more than one alternativemethod, hopefully leading to consistentestimates.

Bankfull Discharge

The bankfull discharge is the dischargethat fills a stable alluvial channel up tothe elevation of the active floodplain.In many natural channels, this is thedischarge that just fills the cross sectionwithout overtopping the banks, hencethe term “bankfull.” This discharge isconsidered to have morphological sig-nificance because it represents thebreakpoint between the processes ofchannel formation and floodplain for-mation. In stable alluvial channels,bankfull discharge corresponds closelywith effective discharge and channel-forming discharge.

The stage vs. discharge or rating curvepresented in Figure 7.4 was developedfor a hypothetical stream by computingthe discharge for different water surfaceelevations or stages. Since dischargesgreater than bankfull spread across theactive floodplain, stage increases moregradually with increasing dischargeabove bankfull than below bankfull,when flows are confined to the channel.Another method for determining thebankfull stage and discharge is to deter-mine the minimum value on a plot re-lating water surface elevation to theratio of surface width to area. The fre-quency of the bankfull discharge can bedetermined from a frequency distribu-tion plot like Figure 7.1.

Bankfull stage can also be identifiedfrom field indicators of the elevation ofthe active floodplain. The correspond-ing bankfull discharge is then deter-mined from a stage vs. dischargerelationship.

Field Indicators of Bankfull Discharge

Various field indicators can be used forestimating the elevation of the stage as-sociated with bankfull flow. Althoughthe first flat depositional surface isoften used, the identification of deposi-tional surfaces in the field can be diffi-


cult and misleading and, at the veryleast, requires trained, experienced fieldpersonnel. After an elevation is selectedas the bankfull, the stage vs. dischargecurve can be computed to determinethe magnitude of the discharge corre-sponding to that elevation.

The above relationships seldom work inincised streams. In an incised stream,the top of the bank might be a terrace(an abandoned floodplain), and indica-tors of the active floodplain might befound well below the existing top ofbank. In this situation, the elevation ofthe channel-forming discharge will bewell below the top of the bank. In addi-tion, the difference between the ordi-nary use of the term “bankfull” and thegeomorphic use of the term can causemajor communication problems.

Field identification of bankfull eleva-tion can be difficult (Williams 1978),but is usually based on a minimumwidth/depth ratio (Wolman 1955), to-gether with the recognition of some dis-continuity in the nature of the channelbanks such as a change in its sedimen-tary or vegetative characteristics. Othershave defined bankfull discharge asfollows:

■ Nixon (1959) defined the bankfullstage as the highest elevation of ariver that can be contained withinthe channel without spilling wateron the river floodplain or washlands.

■ Wolman and Leopold (1957)defined bankfull stage as the eleva-tion of the active floodplain.

■ Woodyer (1968) suggested bankfullstage as the elevation of the middlebench of rivers having several over-flow surfaces.

■ Pickup and Warner (1976) definedbankfull stage as the elevation atwhich the width/depth ratiobecomes a minimum.

Bankfull stage has also been definedusing morphologic factors, as follows:

■ Schumm (1960) defined bankfullstage as the height of the lower limitof perennial vegetation, primarilytrees.

■ Similarly, Leopold (1994) states thatbankfull stage is indicated by achange in vegetation, such as herbs,grasses, and shrubs.

■ Finally, the bankfull stage is alsodefined as the average elevation ofthe highest surface of the channelbars (Wolman and Leopold 1957).

The field identification of bankfull stageindicators is often difficult and subjec-tive and should be performed in streamreaches that are stable and alluvial(Knighton 1984). Additional guidelinesare reviewed by Wharton (1995). In un-stable streams, bankfull indicators areoften missing, embryonic, or difficult todetermine.

Direct determination of the discharge atbankfull stage is possible if a stream

Figure 7.4: Determination of bankfull stagefrom a rating curve. The discharge that corre-sponds to the elevation of the first flat deposi-tional surface is the bankfull discharge.

Stag

e (f

eet)

100

Discharge (cfs)

rating based onManning equation

bankfull stage

3

45

79

11

21

1,000 10,000


gauge is located near the reach of inter-est. Otherwise, discharge must be calcu-lated using applicable hydraulicresistance equations and, preferably,standard hydraulic backwater tech-niques. This approach typically requiresthat an estimation of channel rough-ness be made, which adds to the uncer-tainty associated with calculatedbankfull discharge.

Because of its convenience, bankfulldischarge is widely used to representchannel-forming discharge. There is nouniversally accepted definition of bank-full stage or discharge that can be consis-tently applied, has general application,and integrates the processes that createthe bankfull dimensions of the river.The reader is cautioned that the indica-tors used to define the bankfull condi-tion must be spelled out each time abankfull discharge is used in a projectplan or design.

Determining Channel-FormingDischarge from Recurrence Interval

To avoid some of the problems relatedto field determination of bankfull stage,the channel-forming discharge is often as-sumed to be represented by a specificrecurrence interval discharge. Some re-searchers consider this representativedischarge to be equivalent to the bank-full discharge. Note that “bankfull dis-charge” is used synonymously with“channel-forming discharge” in thisdocument. The earliest estimate forchannel-forming discharge was themean annual flow (Leopold and Mad-dock 1953). Wolman and Leopold(1957) suggested that the channel-forming discharge has a recurrence in-terval of 1 to 2 years. Dury (1973)concluded that the channel-formingdischarge is approximately 97 percentof the 1.58-year discharge or the mostprobable annual flood. Hey (1975)showed that for three British gravel-bed

rivers, the 1.5-year flow in an annualmaximum series passed through thescatter of bankfull discharges measuredalong the course of the rivers. Richards(1982) suggested that in a partial dura-tion series bankfull discharge equals themost probable annual flood, which hasa 1 year return period. Leopold (1994)stated that most investigations haveconcluded that the bankfull dischargerecurrence intervals ranged from 1.0 to2.5 years. Pickup and Warner (1976)determined bankfull recurrence inter-vals ranged from 4 to 10 years on theannual series.

However, there are many instanceswhere the bankfull discharge does notfall within this range. For example,Williams (1978) determined that ap-proximately 75 percent of 51 streamsthat he analyzed appeared to have recur-rence intervals for the bankfull dischargeof between 1.03 and 5.0 years. Williamsused the elevation of the active flood-plain or the valley flat, if no activefloodplain was defined at a station, asthe elevation of the bankfull surface inhis analyses. He did not establishwhether these streams were in equilib-rium, so the validity of using the top ofthe streambank as the bankfull elevationis in question, especially for those sta-tions with valley flats. This might ex-plain the wide range (1.02 to 200 years)he reported for bankfull discharge re-turn intervals for streams with valleyflats as opposed to active floodplains.The range in return intervals for 19 ofthe 28 streams with active floodplainswas from 1.01 to 32 years. Nine of the28 streams had bankfull discharge recur-rence intervals of less than 1.0 year. Itshould be noted that only 3 of those 28streams had bankfull discharge recur-rence intervals greater than 4.8 years.About one-third of the active floodplainstations had bankfull discharges nearthe 1.5-year recurrence interval.

The reader iscautioned thatthe indicatorsused to definethe bankfullcondition mustbe spelled outeach time abankfull dis-charge is usedn a project

plan or design.


Although the assumption that the chan-nel-forming flow has a recurrence inter-val of 1 to 3 years is sufficient forreconnaissance-level studies, it shouldnot be used for design until verifiedthrough inspection of reference reaches,data collection, and analysis. This is es-pecially true in highly modified streamssuch as in urban or mined areas, as wellas ephemeral streams in arid and semi-arid areas.

Effective Discharge

The effective discharge is defined as theincrement of discharge that transportsthe largest fraction of the sediment loadover a period of years (Andrews 1980).The effective discharge incorporates theprinciple prescribed by Wolman andMiller (1960) that the channel-formingdischarge is a function of both the mag-nitude of the event and its frequency ofoccurrence. An advantage of using theeffective discharge is that it is a calcu-lated rather than field-determinedvalue. The effective discharge is calcu-lated by numerically integrating the

flow duration curve (A) and the sedi-ment transport rating curve (B). Agraphical representation of the relation-ship between sediment transport, fre-quency of the transport, and theeffective discharge is shown in Figure7.5. The peak of curve C marks the dis-charge that is most effective in trans-porting sediment and, therefore, doesthe most work in forming the channel.

For stable alluvial streams, effective dis-charge has been shown to be highlycorrelated with bankfull discharge. Ofthe various discharges related to chan-nel morphology (i.e., dominant, bank-full, and effective discharges), effectivedischarge is the only one that can becomputed directly. The effective dis-charge has morphological significancesince it is the discharge that transportsthe bulk of the sediment.

The effective discharge represents thesingle flow increment that is responsi-ble for transporting the most sedimentover some time period. However, thereis a range of flows on either side of theeffective discharge that also carry a sig-nificant portion of the total annual sed-iment load.

Biedenharn and Thorne (1994) useda graphical relationship between the

Discharge

effective discharge

Freq

uenc

y (A)

Prod

uct o

f Mag

nitu

dean

dFr

equency (C)

Sediment Discharg

eRatin

gCurv

e(B

)

Figure 7.5: Effective discharge determinationfrom sediment rating and flow duration curves.The peak of curve C marks the discharge that ismost effective in transporting sediment.Source: Wolman and Miller (1960).


cumulative percentage of sedimenttransported and the water dischargeto define a range of effective dischargesresponsible for the majority of the sedi-ment transport on the Lower MississippiRiver. They found that approximately70 percent of the total sediment wasmoved in a range of flows between500,000 cfs and 1,200,000 cfs, whichcorresponds to the flow that is equaledor exceeded 40 percent of the time and3 percent of the time, respectively.Thorne et al. (1996) used a similar ap-proach to define the range of effectivedischarges on the Brahmaputra River.

A standard procedure should be usedfor the determination of the effectivedischarge to ensure that the results fordifferent sites can be compared. To bepractical, it must either be based onreadily available gauging station data orrequire only limited additional infor-mation and computational procedures.

The basic components required for cal-culation of effective discharge are (1)flow duration data and (2) sedimentload as a function of water discharge.The method most commonly adoptedfor determining the effective dischargeis to calculate the total bed materialsediment load (tons) transported byeach flow increment over a period oftime by multiplying the frequency ofoccurrence for the flow increment(number of days) by the sediment load(tons/day) transported by that flowlevel. The flow increment with thelargest product is the effective discharge.Although this approach has the merit ofsimplicity, the accuracy of the estimateof the effective discharge is clearly de-pendent on the calculation procedureadopted.

Values of mean daily discharges areusually used to compute the flow dura-tion curve, as discussed above and pre-sented in Figure 7.1. However, on flashy

streams, mean daily values can underes-timate the influence of the high flows,and, therefore, it might be necessary toreduce the discharge averaging periodfrom 24 hours (mean daily) to 1 hour,or perhaps 15 minutes.

A sediment rating curve must be devel-oped to determine the effective dis-charge. (See the Sediment Yield andDelivery section in Chapter 8 for moredetails.) The bed material load shouldbe used in the calculation of the effec-tive discharge. This sediment load canbe determined from measured data orcomputed using an appropriate sedi-ment transport equation. If measuredsuspended sediment data are used, thewash load should be subtracted andonly the suspended bed material por-tion of the suspended load used. If thebed load is a significant portion of theload, it should be calculated using anappropriate sediment transport func-tion and added to the suspended bedmaterial load to provide an estimate ofthe total bed material load. If bed loadmeasurements are available, these datacan be used.

Although a channel-forming or domi-nant discharge is important for design,it is often not sufficient for channelrestoration initiatives. An assessmentof a wider range of discharges mightbe necessary to ensure that the func-tional objectives of the project are metFor example, a restoration initiativetargeting low-flow habitat conditionsmust consider the physical conditionsin the channel during low flows.


Determination of effective dischargeusing flow and sediment data is furtherdiscussed by Wolman and Miller(1960) and Carling (1988).

Determining Channel-FormingDischarge from Other WatershedVariables

When neither time nor resources permitfield determination of bankfull dis-charge or data are unavailable to calcu-late the effective discharge, indirectmethods based on regional hydrologicanalysis may be used (Ponce 1989). Inits simplest form, regional analysis en-tails regression techniques to developempirical relationships applicable tohomogeneous hydrologic regions. Forexample, some workers have used wa-tershed areas as surrogates for discharge(Brookes 1987, Madej 1982, Newburyand Gaboury 1993). Regional relation-ships of drainage area with bankfulldischarge can provide good startingpoints for selecting the channel-formingdischarge.

Within hydrologically homogeneous re-gions where runoff varies with con-tributing area, runoff is proportional towatershed drainage area. Dunne andLeopold (1978) and Leopold (1994) de-veloped average curves relating bankfulldischarge to drainage area for widelyseparated regions of the United States.For example, relationships betweenbankfull discharge and drainage area forBrandywine Creek in Pennsylvania andthe upper Green River basin inWyoming are shown in the Figure 7.6.

Two important points are immediatelyapparent from Figure 7.6. First, humidregions that have sustained, widely dis-tributed storms yield higher bankfulldischarges per unit of drainage areathan semiarid regions where storms ofhigh intensity are usually localized. Sec-ond, bankfull discharge is correlatedwith drainage area, and the general rela-

tionship can be represented by func-tions of the form:

Qbf

= aAb

where Qbf

is the bankfull discharge incfs, A is the drainage area in squaremiles, and a and b are regression coeffi-cients and exponents given in Table 7.1.

Establishing similar parametric relation-ships for other rivers of interest is usefulbecause the upstream area draining intoa stream corridor can be easily deter-mined from either maps or digital ter-rain analysis tools. Once the area isdetermined, an estimate of the expectedbankfull discharge for the corridor canbe made from the above equation.

Mean Annual Flow

Another frequently used surrogate forchannel-forming discharge in empiricalregression equations is the mean annualflow. The mean annual flow, Q

m, is

equivalent to the constant dischargethat would yield the same volume ofwater in a water year as the sum of allcontinuously measured discharges. Justas in the case of bankfull discharge, Q

m

varies proportionally with drainage areawithin hydrologically homogeneous

Because the mean annual flow for each stream gaugeoperated by the USGS is readily available, it is useful toestablish regional relationships between bankfull andmean annual discharges so that one can be estimatedwhenever the other is available. This information can becompared to the bankfull discharge estimated for anygiven ungauged site within a U.S. region. The user iscautioned, however, that regional curve valueshave a high degree of error and can vary signifi-cantly for specific sites or reaches to be restored.


basins. Given that both Qbf

and Qm

ex-hibit a similar functional dependenceon A, a consistent proportionality is tobe expected between these dischargemeasures within the same region. Infact, Leopold (1994) gives the followingaverage values of the ratio Q

bf/Q

mfor

three widely separated regions of theUnited States: 29.4 for 21 stations inthe Coast Range of California, 7.1 for20 stations in the Front Range of Col-orado, and 8.3 for 13 stations in theEastern United States.

Dis

char

ge

(cfs

)

Drainage Area (square miles)

1010

100

1000

10,000

1000100

mean annual flow

bankfull flow

Upper East Branch Brandywine Creek, PA

Upper East Branch Brandywine Creek, PAUpper Green River Basin, WY

Upper Green River Basin, WY

River Basin

Southeastern PA

a

61

b

0.82

Upper Salmon River, ID 36 0.68

Upper Green River, WY 28 0.69

San Francisco Bay Region, CA

Qbf = aAb

53 0.93

Table 7.1: Functional parameters used inregional estimates of bankfull discharge.In column a are regression coefficientsand in column b are exponents that canbe used in the bankfull discharge equation. Source: Dunne and Leopold 1978.

Figure 7.6: Regional relationships for bankfull and mean annual discharge as a function ofdrainage area. The mean annual flow is normally less than the bankfull flow.Source: Dunne and Leopold 1978.


Stage vs. DischargeRelationships

Surveys of stream channel cross sectionsare useful for analyzing channel form,function, and processes. Use of surveydata to construct relationships amongstreamflow, channel geometry, and vari-ous hydraulic characteristics providesinformation that serves a variety of ap-plications. Although stage-dischargecurves often can be computed fromsuch cross section data, users should becautioned to verify their computationswith direct discharge measurementswhenever possible.

Information on stream channel geome-try and hydraulic characteristics is use-ful for channel design, riparian arearestoration, and instream structureplacement. Ideally, once a channel-forming discharge is defined, the chan-nel is designed to contain that flow andhigher flows are allowed to spread overthe floodplain. Such periodic floodingis extremely important for the forma-tion of channel macrofeatures, such aspoint bars and meander bends, and forestablishing certain kinds of riparianvegetation. A cross section analysis alsomay help in optimal design and place-ment of items such as culverts and fishhabitat structures.

Additionally, knowledge of the relation-ships between discharge and channelgeometry and hydraulics is useful for re-constructing the conditions associatedwith a particular flow rate. For example,in many channel stability analyses, it iscustomary to relate movement of bedmaterials to some measure of streampower or average bed shear stress. If therelationships between discharge andcertain hydraulic variables (e.g., meandepth and water surface slope) areknown, it is possible to estimate streampower and average bed shear as a func-tion of discharge. A cross section analy-sis therefore makes it possible to

estimate conditions of substrate move-ment at various levels of streamflow.

Continuity Equation

Discharge at a cross section is com-puted using the simplified form of thecontinuity equation:

Q = AV

where:

Q = discharge

A = cross sectional area of the flow

V = average velocity in the down-stream direction

Computing the cross-sectional area is ageometry problem. The area of interestis bounded by the channel cross sectionand the water surface elevation (stage)(Figure 7.7). In addition to cross-sectional area, the top width, wettedperimeter, mean depth, and hydraulicradius are computed for selected stages(Figure 7.7).

Uniform flow equations may be usedfor estimating mean velocity as a function of cross section hydraulic parameters.

Manning’s Equation

Manning’s equation was developed forconditions of uniform flow in whichthe water surface profile and energygrade line are parallel to the streambed,and the area, hydraulic radius, and aver-age depth remain constant throughoutthe reach. The energy grade line is atheoretical line whose elevation abovethe streambed is the sum of the watersurface elevation and a term that repre-sents the kinetic energy of the flow(Chow 1959). The slope of the energygrade line represents the rate at whichenergy is dissipated through turbulenceand boundary friction. When the watersurface slope and the energy grade line


parallel the streambed, the slope of theenergy grade line is assumed to equalthe water surface slope. When the slopeof the energy grade line is known, vari-ous resistance formulas allow comput-ing mean cross-sectional velocity.

The importance of Manning’s equationin stream restoration is that it providesthe basis for computing differences inflow velocities and elevations due todifferences in hydraulic roughness.Note that the flow characteristics can bealtered to meet the goals of the restora-tion either by direct intervention or bychanging the vegetation and roughnessof the stream. Manning’s equation isalso useful in determining bankfull dis-charge for bankfull stage.

Manning’s equation is also used to cal-culate energy losses in natural channelswith gradually varied flow. In this case,calculations proceed from one cross sec-tion to the next, and unique hydraulicparameters are calculated at each crosssection. Computer models, such asHEC-2, perform these calculations andare widely used analytical tools.

Manning’s equation for mean velocity,V (in feet per second or meters per sec-ond), is given as:

V =k__n

R2/3 S1/2

where:

k = 1.486 for English units (1 for metricunits)

n = Manning’s roughness coefficient

R = hydraulic radius (feet or meters)

S = energy slope (water surface slope).

Manning’s roughness coefficient may bethought of as an index of the features ofchannel roughness that contribute tothe dissipation of stream energy. Table7.2 shows a range of n values for vari-ous boundary materials and conditions.

Two methods are presented for estimat-ing Manning’s roughness coefficient fornatural channels:

■ Direct solution of Manning’s equa-tion for n.

■ Comparison with computed n valuesfor other channels.

Each method has its own limitationsand advantages.

Direct Solution for DeterminingManning’s n

Even slightly nonuniform flow can bedifficult to find in natural channels. Themethod of direct solution for Man-ning’s n does not require perfectly uni-form flow. Manning n values arecomputed for a reach in which multiplecross sections, water surface elevations,and at least one discharge have beenmeasured. A series of water surface pro-files are then computed with different nvalues, and the computed profile thatmatches the measured profile isdeemed to have an n value that mostnearly represents the roughness of thatstream reach at the specific discharge.

wetted perimeter depth

(stage)

mean depth = top width

hydraulic radius = wetted perimeter

area

area

topwidth

area

water surface slope

channel bed slo

pe

Figure 7.7: Hydraulic parameters. Streams havespecific cross-sectional and longitudinal profilecharacteristics.


Using Manning’s n Measured atOther Channels

The second method for estimating nvalues involves comparing the reach toa similar reach for which Manning’s nhas already been computed. This proce-dure is probably the quickest and mostcommonly used for estimating Man-ning’s n. It usually involves using valuesfrom a table or comparing the studyreach with photographs of naturalchannels. Tables of Manning’s n valuesfor a variety of natural and artificialchannels are common in the literatureon hydrology (Chow 1959, Van Hav-eren 1986) (Table 7.2). Photographsof stream reaches with computed nvalues have been compiled by Chow(1959) and Barnes (1967). Estimatesshould be made for several stages, andthe relationship between n and stageshould be defined for the range of flowsof interest.

When the roughness coefficient is esti-mated from table values, the chosen nvalue (n

b) is considered a base value

that may need to be adjusted for addi-tional resistance features. Several publi-cations provide procedures for adjustingbase values of n to account for channelirregularities, vegetation, obstructions,and sinuosity (Chow 1959, Benson andDalrymple 1967, Arcement and Schnei-der 1984, Parsons and Hudson 1985).

The most common procedure uses thefollowing formula, proposed by Cowan(1959) to estimate the value of n:

n = (nb

+ n1

+ n2

+ n3

+ n4) m

where

nb

= base value of n for a straight,uniform, smooth channel innatural materials

n1

= correction for the effect of sur-face irregularities

Boundary

Smooth concrete

Ordinary concrete lining

Shot concrete, untroweled, and earth channels in best condition

Straight unlined earth canals in good condition

Rivers and earth canals in fair condition—some growth

Winding natural streams and canals in poor condition—considerablemoss growth

Manning Roughness, n Coefficient

0.012

0.013

Vitrified clay 0.015

0.017

0.020

0.025

0.035

Mountain streams with rocky beds and rivers with variable sections andsome vegetation along banks

0.040-0.050

Alluvial channels, sand bed, no vegetation

1. Lower regime

2. Washed-out dunes or transition 0.014-0.024

3. Upper regime

Plane bed 0.011-0.015

Standing waves 0.012-0.016

Antidunes 0.012-0.020

Ripples 0.017-0.028

Dunes 0.018-0.035

Table 7.2: Manning roughness coefficients for various boundaries.Source: Ven te Chow 1964.


Figure 7.8:Streamflowpaths for chan-nels with con-strictions orobstructions.(a) Riffle or bar,Nisqually,Washington.Source: J. McShane.

(b) Stream widthrestriction.(c) Sweeper log.(d) Stream linesthrough a reach.

riffle

or bar

width constriction

sweeper log

wake

wakerock

(a)

(b)

(c)

(d)

Under conditions of constant width, depth, area,and velocity, the water surface slope and energygrade line approach the slope of the streambed,producing a condition known as “uniform flow.”One feature of uniform flow is that the streamlinesare parallel and straight (Roberson and Crowe1996). Perfectly uniform flow is rarely realized innatural channels, but the condition is approachedin some reaches where the geometry of the chan-nel cross section is relatively constant throughoutthe reach.

Conditions that tend to disrupt uniform flow includebends in the stream course; changes in cross-section-al geometry; obstructions to flow caused by large

roughness elements, such as channel bars, largeboulders, and woody debris; or other features thatcause convergence, divergence, acceleration, ordeceleration of flow (Figure 7.8). Resistance equa-tions may also be used to evaluate these nonuniformflow conditions (gradually varied flow); however,energy-transition considerations (backwater calcula-tions) must then be factored into the analysis. Thisrequires the use of multiple-transect models (e.g.,HEC-2 and WSP2; HEC-2 is a water surface profilecomputer program developed by the U.S. ArmyCorps of Engineers, Hydrologic Engineering Center, inDavis, California; WSP2 is a similar program devel-oped by the USDA Natural Resources ConservationService.)


n2

= correction for variations in crosssection size and shape

n3

= correction for obstructions

n4

= correction for vegetation andflow conditions

m = correction for degree of channelmeandering

Table 7.3 is taken from Aldridge andGarrett (1973) and may be used to esti-mate each of the above correction fac-tors to produce a final estimated n.

Energy Equation

The energy equation is used to calculatechanges in water-surface elevation be-tween two relatively similar cross sec-tions. A simplified version of thisequation is:

z1

+ d1

+ V1

2/2g = z2

+ d2

+ V2

2/2g + he

where:

z = minimum elevation ofstreambed

d = maximum depth of flow

V = average velocity

g = acceleration of gravity

he= energy loss between the two sec-

tions

Subscript 1 indicates that the variable isat the upstream cross section, and sub-script 2 indicates that the variable is atthe downstream cross section.

This simplified equation is applicablewhen hydraulic conditions between thetwo cross sections are relatively similar(gradually varied flow) and the channelslope is small (less than 0.18).

Energy losses between the two cross sec-tions occur due to channel boundaryroughness and other factors describedabove. These roughnesses may be repre-sented by a Manning’s roughness coeffi-cient, n, and then energy losses can becomputed using the Manning equation.

he= L [Qn/kAR2/3]2

where:

L = distance between cross sections

Q = discharge

n = Manning’s roughness coefficient

A = channel cross-sectional area

R = hydraulic radius (Area/wettedperimeter)

k = 1 (SI units)

k = 1.486 (ft-lb-sec units)

Computer models (such as HEC-2 andothers) are available to perform thesecalculations for more complex cross-sectional shapes, including floodplains,and for cases where roughness varieslaterally across the cross section(USACE 1991).

Just as Manning’s n may vary significantly with changesin stage (water level), channel irregularities, obstructions,vegetation, sinuosity, and bed-material size distribution,n may also vary with bedforms in the channel. Thehydraulics of sand and mobile-bed channels producechanges in bedforms as the velocity, stream power, andFroude number increase with discharge. The Froudenumber is a dimensionless number that represents theratio of inertial forces to gravitational force. As velocityand stream power increase, bedforms evolve from rip-ples to dunes, to washed-out dunes, to plane bed, toantidunes, to chutes and pools. A stationary plane bed,ripples, and dunes occur when the Froude number (longwave equation) is less than 1 (subcritical flow); washed-out dunes occur at a Froude number equal to 1 (criticalflow); and a plane bed in motion, antidunes, and chutesand pools occur at a Froude number greater than 1(supercritical flow). Manning’s n attains maximum valueswhen dune bedforms are present, and minimum valueswhen ripples and plane bedforms are present (Parsonsand Hudson 1985).


Degree of meandering1

(adjustment valuesapply to flow confined in the channel and do not apply where downvalley flow crosses meanders) (m)

ChannelConditions

n Value Adjustment1/

Smooth 0.000

Example

Compares to the smoothest channel attainable in a given bed material.

Minor 0.001-0.005 Compares to carefully dredged channels in good condition but havingslightly eroded or scoured side slopes.

Moderate 0.006-0.010 Compares to dredged channels having moderate to considerable bedroughness and moderately sloughed or eroded side slopes.

Severe 0.011-0.020 Badly sloughed or scalloped banks of natural streams; badly eroded orsloughed sides of canals or drainage channels; unshaped, jagged, andirregular surfaces of channels in rock.

Degree of irregularity (n1)

Gradual 0.000 Size and shape of channel cross sections change gradually.

Alternatingoccasionally

0.001-0.005 Large and small cross sections alternate occasionally, or the main flowoccasionally shifts from side to side owing to changes in cross-sectional shape.

Alternatingfrequently

0.010-0.015 Large and small cross sections alternate frequently, or the main flowfrequently shifts from side to side owing to changes in cross-sectional shape.

Negligible 0.000-0.004 A few scattered obstructions, which include debris deposits, stumps,exposed roots, logs, piers, or isolated boulders, that occupy less than5 percent of the cross-sectional area.

Minor 0.005-0.015

Variation inchannel crosssection (n2)

Effect ofobstruction (n3)

Small 0.002-0.010Amount ofvegetation (n4)

Obstructions occupy less than 15 percent of the cross-sectional area and the spacing between obstructions is such that the sphere of influence around one obstruction does not extend to the sphere of influence around another obstruction. Smaller adjustments are used for curved smooth-surfaced objects than are used for sharp-edged angular objects.

Appreciable 0.020-0.030 Obstructions occupy from 15 to 20 percent of the cross-sectional area or the space between obstructions is small enough to cause the effects of several obstructions to be additive, thereby blocking an equivalent part of a cross section.

Severe 0.040-0.050 Obstructions occupy more than 50 percent of the cross-sectional area or the space between obstructions is small enough to cause turbulence across most of the cross section.

Dense growths of flexible turf grass, such as Bermuda, or weeds growing where the average depth of flow is at least two times the height of the vegetation; supple tree seedlings such as willow, cottonwood, arrowweed, or saltcedar growing where the average depth of flow is at least three times the height of the vegetation.

Minor 1.00 Ratio of the channel length to valley length is 1.0 to 1.2.

Appreciable 1.15 Ratio of the channel length to valley length is 1.2 to 1.5.

Severe 1.30 Ratio of the channel length to valley length is greater than 1.5.

Medium 0.010-0.025 Turf grass growing where the average depth of flow is from one to two times the height of the vegetation; moderately dense stemmy grass, weeds, or tree seedlings growing where the average depth of the flow is from two to three times the height of the vegetation; brushy, moderately dense vegetation, similar to 1- to 2-year-old willow trees in the dormant season, growing along the banks and no significant vegetation along the channel bottoms where the hydraulic radius exceeds 2 feet.

Large 0.025-0.050 Turf grass growing where the average depth of flow is about equal to the height of vegetation; 8- to 10-year-old willow or cottonwood trees intergrown with some weeds and brush (none of the vegetation in foliage) where the hydraulic radius exceeds 2 feet; bushy willows about 1 year old intergrown with some weeds along side slopes (all vegetation in full foliage) and no significant vegetation along channel bottoms where the hydraulic radius is greater than 2 feet.

Very Large 0.050-0.100 Turf grass growing where the average depth of flow is less than half the height of the vegetation; bushy willow trees about 1 year old intergrown with weeds along side slopes (all vegetation in full foliage) or dense cattails growing along channel bottom; trees intergrown with weeds and brush (all vegetation in full foliage).

1 Adjustments for degree of irregularity, variations in cross section, effect of obstructions, and vegetation are added to the base n value before multiplying by the adjustment for meander.

Table 7.3: “n” value adjustments.Source: Aldridge and Garrett (1973).

Analyzing Composite andCompound Cross Sections

Natural channel cross sections are rarelyperfectly uniform, and it may be neces-sary to analyze hydraulics for very irreg-ular cross sections (compoundchannel). Streams frequently have over-flow channels on one or both sides thatcarry water only during unusually highflows. Overflow channels and overbankareas, which may also carry out-of-bankflows at various flood stages, usuallyhave hydraulic properties significantlydifferent from those of the main chan-nel. These areas are usually treated asseparate subchannels, and the dischargecomputed for each of these subsectionsis added to the main channel to com-pute total discharge. This procedure ig-nores lateral momentum losses, whichcould cause n values to be underesti-mated.

A composite cross section has rough-ness that varies laterally across the sec-tion, but the mean velocity can still becomputed by a uniform flow equationwithout subdividing the section. For ex-ample, a stream may have heavily vege-tated banks, a coarse cobble bed at itslowest elevations, and a sand bar vege-tated with small annual willow sprouts.

A standard hydraulics text or reference(such as Chow 1959, Henderson 1986,USACE 1991, etc.) should be consultedfor methods of computing a compositen value for varying conditions across asection and for varying depths of flow.

Reach Selection

The intended use of the cross sectionanalysis plays a large role in locatingthe reach and cross sections. Cross sec-tions can be located in either a shortcritical reach where hydraulic character-


Straight channel reaches with perfectly uniform flow are rare in nature and, in mostcases, may only be approached to varying degrees. If a reach with constant cross-sec-tional area and shape is not available, a slightly contracting reach is acceptable, provid-ed there is no significant backwater effect from the constriction. Backwater occurswhere the stage vs. discharge relationship is controlled by the geometry downstream ofthe area of interest (e.g., a high riffle controls conditions in the upstream pool at lowflow). Manning’s equation assumes uniform flow conditions. Manning’s equation usedwith a single cross section, therefore, will not produce an accurate stage vs. dischargerelationship in backwater areas. In addition, expanding reaches also should be avoidedsince there are additional energy losses associated with channel expansions. When nochannel reaches are available that meet or approach the condition of uniform flow, itmight be necessary to use multitransect models (e.g., HEC-2) to analyze cross sectionhydraulics. If there are elevation restrictions corresponding to given flows (e.g., floodcontrol requirements), the water surface profile for the entire reach is needed and useof a multitransect (backwater) model is required.


istics change or in a reach that is con-sidered representative of some largerarea. The reach most sensitive to changeor most likely to meet (or fail to meet)some important condition may be con-sidered a critical reach. A representativereach typifies a definable extent of thechannel system and is used to describethat portion of the system (Parsons andHudson 1985).

Once a reach has been selected, thechannel cross sections should be mea-sured at locations considered most suitable for meeting the uniform flowrequirements of Manning’s equation.The uniform flow requirement is ap-proached by siting cross sections wherechannel width, depth, and cross-sectional flow area remain relativelyconstant within the reach, and thewater surface slope and energy gradeline approach the slope of the stream-bed. For this reason, marked changes inchannel geometry and discontinuitiesin the flow (steps, falls, and hydraulicjumps) should be avoided. Generally,sections should be located where it appears the streamlines are parallel tothe bank and each other within the se-lected reach. If uniform flow conditionscannot be met and backwater computa-tions are required, defining cross sec-tions located at changes in channelgeometry is essential.

Field Procedures

The basic information to be collectedin the reach selected for analysis is asurvey of the channel cross sectionsand water surface slope, a measure-ment of bed-material particle sizedistribution, and a discharge measure-ment. The U.S. Forest Service has pro-duced an illustrated guide to fieldtechniques for stream channel refer-ence sites (Harrelson et al. 1994) thatis a good reference for conducting fieldsurveys.

Survey of Cross Section andWater Surface Slope

The cross section is established perpen-dicular to the flow line, and the pointsacross the section are surveyed relativeto a known or arbitrarily establishedbenchmark elevation. The distance/ele-vation paired data associated with eachpoint on the section may be obtainedby sag tape, rod-and-level survey, hydro-graphic surveys, or other methods.

Water surface slope is also required fora cross section analysis. The survey ofwater surface slope is somewhat morecomplicated than the cross section sur-vey in that the slope of the water sur-face at the location of the section (e.g.,pool, run, or riffle) must be distin-guished from the more constant slopeof the entire reach. (See Grant et al.1990 for a detailed discussion on recog-nition and characteristics of channel

Many computer programs (e.g., HEC-2)are available to compute water surfaceprofiles. The standard step method ofChow (1959, p. 265) can be used todetermine the water surface elevation(depth) at the upstream end of the reachby iterative approximations. This methoduses trial water surface elevations todetermine the elevation that satisfies theenergy and Manning equations writtenfor the end sections of the reach. Inusing this method, cross sections shouldbe selected so that velocities increase ordecrease continuously throughout thereach (USACE 1991).


units.) Water surface slope in individualchannel reaches may vary significantlywith changes in stage and discharge.

For this reason, when water surfaceslopes are surveyed in the field, thelow-water slope may be approximatedby the change in elevation over the in-dividual channel unit where the crosssection is located, approximately 1 to 5channel widths in length, while thehigh-water slope is obtained by mea-suring the change in elevation over amuch longer reach of channel, usuallyat least 15 to 20 channel widths inlength.

Bed Material Particle Size Distribution

Computing mean velocity with resis-tance equations based on relativeroughness, such as the ones suggestedby Thorne and Zevenbergen (1985), re-quires an evaluation of the particle sizedistribution of the bed material of thestream. For streams with no significantchannel armor and bed material finerthan medium gravel, bed material sam-plers developed by the Federal Inter-agency Sedimentation Project (FISP1986) may be used to obtain a repre-sentative sample of the streambed,which is then passed through a set ofstandard sieves to determine percentby weight of particles of various sizes.The cumulative percent of materialfiner than a given size may then bedetermined.

Particle size data are usually reportedin terms of d

i, where i represents some

nominal percentile of the distributionand d

irepresents the particle size, usu-

ally expressed in millimeters, at whichi percent of the total sample by weightis finer. For example, 84 percent of thetotal sample would be finer than thed

84particle size. For additional guidance

on bed material sampling in sand-bedstreams, refer to Ashmore et al. (1988).

For estimating velocity in steep moun-tain rivers with substrate much coarserthan the medium-gravel limitation ofFISP samplers, a pebble count, in whichat least 100 bed material particles aremanually collected from the streambedand measured, is used to measure sur-face particle size (Wolman 1954). Ateach sample point along a cross section,a particle is retrieved from the bed, andthe intermediate axis (not the longestor shortest axis) is measured. The mea-surements are tabulated as to numberof particles occurring within predeter-mined size intervals, and the percentageof the total number in each interval isthen determined. Again, the percentagein each interval is accumulated to give aparticle size distribution, and the parti-cle size data are reported as describedabove. Additional guidance for bed ma-terial sampling in coarse-bed streams isprovided in Yuzyk (1986). If an armorlayer or pavement is present, standardtechniques may be employed to charac-terize bed sediments, as described byHey and Thorne (1986).

Discharge Measurement

If several discharge measurements canbe made over a wide range of flows,relationships among stage, discharge,and other hydraulic parameters may bedeveloped directly. If only one dis-charge measurement is obtained, itlikely will occur during low water andwill be useful for defining the lowerend of the rating table. If two measure-ments can be made, it is desirable tohave a low-water measurement and ahigh-water measurement to define bothends of the rating table and to establishthe relationship between Manning’s nand stage. If high water cannot be mea-sured directly, it may be necessary to es-timate the high-water n (see thediscussion earlier in the chapter).


In planning a project along a river orstream, awareness of the fundamentalsof fluvial geomorphology and channelprocesses allows the investigator to seethe relationship between form andprocess in the landscape. The detailedstudy of the fluvial geomorphicprocesses in a channel system is oftenreferred to as a geomorphic assessment.The geomorphic assessment providesthe process-based framework to definepast and present watershed dynamics,develop integrated solutions, and assessthe consequences of restoration activi-ties. A geomorphic assessment generallyincludes data collection, field investiga-tions, and channel stability assessments.It forms the foundation for analysis anddesign and is therefore an essential firststep in the design process, whetherplanning the treatment of a single reachor attempting to develop a comprehen-sive plan for an entire watershed.

Stream Classification

The use of any stream classification sys-tem is an attempt to simplify what arecomplex relationships between streamsand their watersheds.

Although classification can be used as acommunications tool and as part of theoverall restoration planning process, theuse of a classification system is not re-quired to assess, analyze, and designstream restoration initiatives. The de-sign of a restoration does, however, re-quire site-specific engineering analysesand biological criteria, which are cov-ered in more detail in Chapter 8.

Restoration designs range from simpleto complex, depending on whether “noaction,” only management techniques,direct manipulation, or combinationsof these approaches are used. Completestream corridor restoration designs re-quire an interdisciplinary approach as

The Bureau of Reclamation Water Mea-surement Manual (USDI-BOR 1997) isan excellent source of information formeasuring channel and stream dis-charge (Figure 7.9). Buchanan andSomers (1969) and Rantz et al. (1982)also provide in-depth discussions ofdischarge measurement techniques.When equipment is functioning prop-erly and standard procedures are fol-lowed correctly, it is possible tomeasure streamflow to within 5 percentof the true value. The USGS considersa “good” measurement of discharge toaccount for plus or minus 5 percentand an “excellent” discharge measure-ment to be within plus or minus 3 per-cent of the true value.

Figure 7.9: Stationmeasuring discharge.Permanent stationsprovide measure-ments for a widerange of flow, butthe necessary mea-surements can bemade in other ways.Source: C. Zabawa.

7.B Geomorphic Processes

Geomorphic Processes 7–27

discussed in Chapter 4. A poorly de-signed restoration might be difficult torepair and can lead to more extensiveproblems.

More recent attempts to develop a com-prehensive stream classification systemhave focused on morphological formsand processes of channels and valleybottoms, and drainage networks. Classi-fication systems might be categorized assystems based on sediment transportprocesses and systems based on channelresponse to perturbation.

Stream classification methods are re-lated to fundamental variables andprocesses that form streams. Streams areclassified as either alluvial or non-alluvial. An alluvial stream is free toadjust its dimensions, such as width,depth, and slope, in response to changesin watershed sediment discharge. Thebed and banks of an alluvial stream arecomposed of material transported bythe river under present flow conditions.Conversely, a non-alluvial river, like abedrock-controlled channel, is not freeto adjust. Other conditions, such as ahigh mountain stream flowing in verycoarse glacially deposited materials orstreams which are significantly con-trolled by fallen timber, would suggesta non-alluvial system.

Streams may also be classified as eitherperennial, intermittent, or ephemeral,as discussed in Chapter 1. A perennialstream is one that has flow at all times.An intermittent stream has the potentialfor continued flow, but at times the en-tire flow is absorbed by the bed mater-ial. This may be seasonal in nature.An ephemeral stream has flow only fol-lowing a rainfall event. When carryingflow, intermittent and ephemeralstreams both have characteristics verysimilar to those of perennial streams.

Advantages of StreamClassification Systems

The following are some advantages ofstream classification systems:

■ Classification systems promote com-munication among persons trainedin different resource disciplines.

■ They also enable extrapolation ofinventory data collected on a fewchannels of each stream class to amuch larger number of channels overa broader geographical area.

■ Classification helps the restorationpractitioner consider the landscapecontext and determine the expectedrange of variability for parametersrelated to channel size, shape, andpattern and composition of bed andbank materials.

■ Stream classification also enables thepractitioner to interpret the channel-forming or dominant processes activeat the site, providing a base on whichto begin the process of designingrestoration.

■ Classified reference reaches can beused as the stable or desired form ofthe restoration.

■ A classification system is also veryuseful in providing an importantcross-check to verify if the selecteddesign values for width/depth ratio,sinuosity, etc., are within a reason-able range for the stream type beingrestored.

Limitations of StreamClassification Systems

All stream classification systems havelimitations that are inherent to their ap-proaches, data requirements, and rangeof applicabilities. They should be usedcautiously and only for establishingsome of the baseline conditions on


which to base initial restoration plan-ning. Standard design techniquesshould never be replaced by streamclassification alone.

Some limitations of classification sys-tems are as follows:

■ Determination of bankfull or channel-forming flow depth may be difficultor inaccurate. Field indicators areoften subtle or missing and are notvalid if the stream is not stable andalluvial.

■ The dynamic condition of the streamis not indicated in most classificationsystems. The knowledge of whetherthe stream is stable, aggrading, ordegrading or is approaching a criticalgeomorphic threshold is importantfor a successful restoration initiative.

■ River response to a perturbation orrestoration action is normally notdetermined from the classificationsystem alone.

■ Biological health of a stream is usual-ly not directly determined through astream classification system.

■ A classification system alone shouldnot be used for determining the type,location, and purpose of restorationactivities. These are determinedthrough the planning steps in Part IIand the design process in Chapter 8.

When the results of stream classifica-tion will be used for planning or de-sign, the field data collection should beperformed or directed by persons withexperience and training in hydrology,hydraulics, terrestrial and aquatic ecol-ogy, sediment transport, and river me-chanics. Field data collected bypersonnel with only limited formaltraining may not be reliable, particu-larly in the field determination of bank-full indicators and the assessment ofchannel instability trends.

Stream Classification Systems

Stream Order

Designation of stream order, using theStrahler (1957) method, described inChapter 1, is dependent on the scale ofmaps used to identify first-orderstreams. It is difficult to make directcomparisons of the morphologicalcharacteristics of two river basins ob-tained from topographic maps of differ-ent scales. However, the basicmorphological relationships defined byHorton (1945) and Yang (1971) arevalid for a given river basin regardlessof maps used, as shown in the casestudy of the Rogue River Basin (Yangand Stall 1971, 1973).

Horton (1945) developed some basicempirical stream morphology relations,i.e., Horton’s law of stream order,stream slope, and stream length. Theseshow that the relationships betweenstream order, average stream length,and slope are straight lines on semilogpaper.

Yang (1971) derived his theory of aver-age stream fall based on an analogywith thermodynamic principles. Thetheory states that the ratio of average fall(change in bed elevation) between anytwo stream orders in a given river basinis unity. These theoretical results weresupported by data from 14 river basinsin the United States with an average fallratio of 0.995. The Rogue River basindata were used by Yang and Stall(1973) to demonstrate the relationshipsbetween average stream length, slope,fall, and number of streams.

Stream order is used in the River Contin-uum Concept (Vannote et al. 1980),described in Chapter 1, to distinguishdifferent levels of biological activity.However, stream order is of little helpto planners and designers looking forclues to restore hydrologic and geomor-phic functions to stream corridors.


Schumm

Other classification schemes combinemorphological criteria with dominantmodes of sediment transport. Schumm(1977) identified straight, meandering,and braided channels and related bothchannel pattern and stability to modesof sediment transport (Figure 7.10).

Schumm recognized relatively stablestraight and meandering channels, withpredominantly suspended sedimentload and cohesive bank materials. Onthe other end of the spectrum are rela-tively unstable braided streams charac-terized by predominantly bedloadsediment transport and wide, sandychannels with noncohesive bank mate-rials. The intermediate condition is gen-erally represented by meanderingmixed-load channels.

Montgomery and Buffington

Schumm’s classification system primar-ily applies to alluvial channels; Mont-gomery and Buffington (1993) haveproposed a similar classification systemfor alluvial, colluvial, and bedrockstreams in the Pacific Northwest thataddresses channel response to sedimentinputs throughout the drainage net-work. Montgomery and Buffington rec-ognize six classes of alluvialchannels — cascade, step-pool, plane-bed, riffle-pool, regime, and braided(Figure 7.11).

The stream types are differentiated onthe basis of channel response to sedi-ment inputs, with steeper channels(cascade and step-pool) maintainingtheir morphology while transmittingincreased sediment loads, and low-gradient channels (regime and pool-riffle) responding to increased sedimentthrough morphological adjustments. Ingeneral, steep channels act as sediment-delivery conduits connecting zones ofsediment production with low-gradientresponse channels.

Rosgen Stream Classification System

One comprehensive stream classifica-tion system in common use is based onmorphological characteristics describedby Rosgen (1996) (Figure 7.12). TheRosgen system uses six morphologicalmeasurements for classifying a streamreach— entrenchment, width/depthratio, sinuosity, number of channels,slope, and bedmaterial particle size.These criteria are used to define eightmajor stream classes with about 100individual stream types.

Rosgen uses the bankfull dischargeto represent the stream-forming dis-charge or channel-forming flow. Bank-full discharge is needed to use thisclassification system because all of themorphological relationships are relatedto this flow condition: width and depthof flow are measured at the bankfullelevation, for example.

Except for entrenchment andwidth/depth ratio (both of which de-pend on a determination of bankfulldepth), the parameters used are rela-tively straightforward measurements.The problems in determining bankfulldepth were discussed earlier in Chapter1. The width/depth ratio is taken atbankfull stage and is the ratio of topwidth to mean depth for the bankfullchannel. Sinuosity is the ratio of streamlength to valley length or, alternatively,valley slope to stream slope. The bedmaterial particle size used in the classi-fication is the dominant bed surfaceparticle size, determined in the field bya pebble-count procedure (Wolman1954) or as modified for sand andsmaller sizes. Stream slope is measuredover a channel reach of at least 20widths in length.

Entrenchment describes the relation-ship between a stream and its valleyand is defined as the vertical contain-ment of the stream and the degree to


which it is incised in the valley floor. Itis, therefore, a measure of how accessi-ble a floodplain is to the stream. Theentrenchment ratio used in the Rosgenclassification system is the flood-pronewidth of the valley divided by the bank-full width of the channel. Flood-pronewidth is determined by doubling themaximum depth in the bankfull chan-nel and measuring the width of the val-ley at that elevation. If the flood-pronewidth is greater than 2.2 times thebankfull width, the stream is consid-ered to be slightly entrenched or con-fined and the stream has ready access toits floodplain. A stream is classified as

entrenched if its flood-prone width isless than 1.4 times the bankfull width.

A sample worksheet for collecting dataand classifying a stream using the Ros-gen system is shown in Figure 7.13. Afield book for collecting reference reachinformation is available (Leopold et al.1997).

Channel Evolution Models

Conceptual models of channel evolutiondescribe the sequence of changes astream undergoes after certain kindsof disturbances. The changes can in-clude increases or decreases in thewidth/depth ratio of the channel andalso involve alterations in the flood-plain. The sequence of changes is some-what predictable, so it is important thatthe current stage of evolution be identi-fied so appropriate actions can beplanned.

Mea

nd

erin

gSt

raig

ht

Bra

ided

Rel

ativ

e St

abili

tyH

igh

Low

Ch

an

nel

Patt

ern

wid

th/d

epth

rat

iog

rad

ien

th

igh

hig

hlo

wlo

w

High Relative Stability

bed load/total load ratiolow(3%>) high (>11%)small sediment size

sediment loadsmallflow velocitylowstream powerlow

largelargehighhigh

Low

Suspended Load

channel boundary

flow

bars

Mixed Load

Channel Type

Bed Load

Figure 7.10: Classification of alluvial channels.Schumm’s classification system relates channelstability to kind of sediment load and channeltype. Source: Schumm, The Fluvial System. © 1977.Reprinted by permission of John Wiley and Sons, Inc.


Typical Bed Material

Bedform Pattern

Reach Type

Dominant Roughness Elements

Dominant Sediment Sources

Sediment Storage Elements

Typical Slope (m/m)

Typical Confinement

Pool Spacing(Channel Widths)

Regime

Response

Multi-layered

Sand

Sinuosity, bedforms (dunes, ripples, bars) banks

Fluvial,bank failure,inactive channel

Overbank, bedforms, inactive channel

S < 0.001

Unconfined

5 to 7

Braided

Response

Laterally oscillary

Variable

Bedforms (bars, pools)

Fluvial,bank failure,debris flow

Overbank,bedforms

S < 0.03

Unconfined

Variable

Pool-Riffle

Response

Laterally oscillary

Gravel

Bedforms (bars, pools), grains, LWD, sinuosity, banks

Fluvial,bank failure,inactive channel, debris flows

Overbank, bedforms, inactive channel

0.001 < Sand S < 0.02

Unconfined

5 to 7

Plane-Bed

Response

None

Gravel,cobble

Grains,banks

Fluvial,bank failure,debris flow

Overbank, inactive channel

0.01 < S andS < 0.03

Variable

none

Step-Pool

Transport

Vertically oscillary

Cobble, boulder

Bedforms (steps, pools), grains, LWD, banks

Fluvial, hillslope, debris flow

Bedforms

0.03 < S andS < 0.08

Confined

1 to 4

Cascade

Transport

None

Boulder

Grains, banks


Lee & stoss sides of flow obstructions

0.08 < S andS < 0.30

Confined

< 1

Bedrock

Transport

•

N/A

Boundaries (bed & banks)


•

Variable

Confined

Variable

Colluvial

Source

Variable

Variable

Grains, LWD

Hillslope, debris flow

Bed

S > 0.20

Confined

Variable

Pool-Riffle Plane-Bed Step-Pool Cascade BedrockRegimeBraidedColluvial

AlluvialColluvial Bedrock

Transport Limited Supply Limited

Figure 7.11: Suggested stream classificationsystem for Pacific Northwest. Included areclassifications for nonalluvial streams.Source: Montgomery and Buffington 1993.


Schumm et al. (1984), Harvey and Wat-son (1986), and Simon (1989) haveproposed similar channel evolutionmodels due to bank collapse based on a“space-for-time” substitution, wherebydownstream conditions are interpretedas preceding (in time) the immediatelocation of interest and upstream con-ditions are interpreted as following (intime) the immediate location of inter-est. Thus, a reach in the middle of thewatershed that previously looked likethe channel upstream will evolve tolook like the channel downstream.

Downs (1995) reviews a number ofclassification schemes for interpretingchannel processes of lateral and verticaladjustment (i.e., aggradation, degrada-tion, bend migration, and bar forma-tion). When these adjustment processesare placed in a specific order of occur-rence, a channel evolution model(CEM) is developed. Although a num-ber of CEMs have been suggested, twomodels (Schumm et al. 1984 and

Simon 1989, 1995) have gained wideacceptance as being generally applicablefor channels with cohesive banks.

Both models begin with a pre-disturbance condition, in which thechannel is well vegetated and hasfrequent interaction with its flood-plain. Following a perturbation in thesystem (e.g., channelization or changein land use), degradation occurs, usu-ally as a result of excess stream powerin the disturbed reach. Channel degra-dation eventually leads to oversteep-ening of the banks, and when criticalbank heights are exceeded, bank fail-ures and mass wasting (the episodic

Sinuosity

StreamType

Slope

Bedrock

Boulders

Cobble

Gravel

Sand

Silt/Clay

ChannelMaterial

EntrenchmentRatio

Single-Threaded Channels Multiple Channels

Moderate Sinuosity(>1.2)



Very High Sinuosity(>1.5)

High Sinuosity(>1.2)

Low Sinuosity(<1.2)

Low-Hi Sinuosity(1.2-1.5 )

LowSinuosity(<1.2)

slope range slope range slope range slope range slope range slope range slopeslope range

0.02-0.039

0.02-0.039

>0.10 <0.02 <0.02 0.02-0.039

.04-0.099

0.02-0.039

<0.02 .001-0.02

.02-0.039

<.001 .001-0.02

.02-0.039

<.001 <.005<0.020.04-0.099

G6 F6bA6a+ G6c F6 B6B6a E6bB6c C6C6b C6c- D6D6b D6c- DA6E6A6



G3 F3bA3a+ G3c F3 B3B3a E3bB3c C3C3b C3c- D3D3bE3A3

G2 F2bA2a+ G2c F2 B2B2a B2c C2C2b C2c-A2

G1 F1bA1a+ G1c F1 B1B1a B1c C1C1b C1c-A1

moderate to High w/d(>12)

Moderate width/depth ratio(>12)

Very Low width/depth(<12)

moderate to High width/depth(>12)

very Highwidth/depth(>40)

Loww/d(<40)

Lowwidth/depth ratio(<12)

ModeratelyEntrenched (1.4-2.2) Slightly Entrenched (> 2.2)Entrenched (Ratio: < 1.4)

Width/DepthRatio

A G F B E C D DA

Figure 7.12: Rosgen’s stream channel classifica-tion system (Level II). This classification systemincludes a recognition of specific characteristicsof channel morphology and the relationshipbetween the stream and its floodplain.Source: Rosgen 1996. Published by permission ofWildland Hydrology.


Silt/Clay<.062

PEBBLE COUNT Site__________________________________

Metric(mm)

English(inches)

Particle Count Tot#

%Tot

%Cum

Count Tot#

%Tot

%Cum

Count Tot#

%Tot

%Cum

<.002

Fine Sand.062-0.25 .002-.01

Med Sand0.25-.5 .01-.02

Coarse Sand.5-1.0 .02-.04

Vy Coarse Sand1.0-2.0 .04-.08

Fine Gravel2-8 .08-.32

Med Gravel8-16 .32-.63

Coarse Gravel16-32 .63-1.26

Vy Coarse Gravel32-64 1.26-2.51

Small Cobbles64-128 2.51-5.0

Large Cobbles128-256 5.0-10.1

Sm Boulders256-512 10.1-20.2

Med Boulders512-1024 20.2-40.3

Lg Boulders1024-2048 40.3-80.6

Vy Lg Boulders2048-4096 80.6-161

STREAM CLASSIFICATION WORKSHEET

Party:______________________ Date:_______________________State:______________________ County:_____________________Stream: ___________________________________________________

Bankfull Measurements: Lat/Long __________________________Width _______________ Depth ______________ W/D ______________

Sinuosity (Stream Length/Valley Length) or (Valley Slope/Channel Slope):Strm. Length ___________________ Valley Slope __________________________Valley Length ___________________ Channel Slope ________________________ SL VSSinuosity VL _____________________ Sinuosity CS __________________________

Entrenchment Ratio (Floodprone Width/Bankfull Width):Floodprone width is water level at 2x maximum depth in bankfull cross-section, or width of intermediate floodplain (10-50 yr. event)Bankfull Width __________________ Floodprone Width ____________________Entrenchment Ratio ____________________ Slight = 2.2+ Moderate + 1.41-2.2 Entrenched = 1.0-1.4

Dominant Channel Soils:Bed Material _____________ Left Bank ______________ Right Bank ______________Description of Soil Profiles (from base of bank to top)Left: __________________________________________________________________________Right: _________________________________________________________________________

Riparian Vegetation:Left Bank: _______________________ Right Bank __________________________% Total Area (Mass) L _________________ R __________________% Total Ht w/Roots L __________________ R __________________Ratio of Actual Bank Height to Bankfull Height _____________________________

Bank Slope (Horizontal to Vertical ): L ____________________ R _____________________

STREAM TYPE ________________ Remarks ________________________________________

Figure 7.13: Example of stream classification worksheet used with Rosgen methods.Source: NRCS 1994 (worksheet) and Rosgen 1996 (pebble count). Published by permission of Wildland Hydrology.


downslope movement of soil and rock)lead to channel widening. As channelwidening and mass wasting proceed up-stream, an aggradation phase follows inwhich a new low-flow channel beginsto form in the sediment deposits.Upper banks may continue to be unsta-ble at this time. The final stage of evolu-tion is the development of a channelwithin the deposited alluvium with di-mensions and capacity similar to thoseof the predisturbance channel (Downs1995). The new channel is usuallylower than the predisturbance channel,and the old floodplain now functionsprimarily as a terrace.

Once streambanks become high, eitherby downcutting or by sediment deposi-tion on the floodplain, they begin tofail due to a combination of erosion atthe base of the banks and mass wasting.The channel continues to widen untilflow depths do not reach the depths re-quired to move the sloughed bank ma-terials. Sloughed materials at the baseof the banks may begin to be colonizedby vegetation. This added roughnesshelps increase deposition at the base ofthe banks, and a new small-capacitychannel begins to form between the sta-bilized sediment deposits. The finalstage of channel evolution results in anew bankfull channel and active flood-plain at a new lower elevation. Theoriginal floodplain has been aban-doned due to channel incision or exces-sive sediment deposition and is nowtermed a terrace.

Schumm et al. (1984) applied the basicconcepts of channel evolution to theproblem of unstable channelizedstreams in Mississippi. Simon (1989)built on Schumm’s work in a study ofchannelized streams in Tennessee.Simon’s CEM consisted of six stages(Figure 7.14). Both models use thecross section, longitudinal profile, andgeomorphic processes to distinguish

stages of evolution. Both models weredeveloped for landscapes dominated bystreams with cohesive banks. However,the same physical processes of evolu-tion can occur in streams with nonco-hesive banks but not necessarily in thesame well-defined stages.

Table 7.4 and Figure 7.15 show theprocesses at work in each of Simon’sstages.

Advantages of ChannelEvolution Models

CEMs are useful in stream corridorrestoration in the following ways(Note: Stages are from Simon’s 1989six-stage CEM):

■ CEMs help to establish the directionof current trends in disturbed or con-structed channels. For example, if areach of stream is classified as beingin Stage IV of evolution (Figure7.14), more stable reaches shouldoccur downstream and unstablereaches should occur upstream.Once downcutting or incision occursin a stream (Stage III), the headcutwill advance upstream until it reach-es a resistant soil layer, the drainagearea becomes too small to generateerosive runoff, or the slope flattens tothe point that the stream cannotgenerate enough energy to downcut.Stages IV to VI will follow the head-cut upstream.

■ CEMs can help to prioritize restora-tion activities if modification isplanned. By stabilizing a reach ofstream in early Stage III with gradecontrol measures, the potentialdegradation of that reach andupstream reaches can be prevented.It also takes less intensive efforts tosuccessfully restore stream reaches inStages V and VI than to restore thosein Stages III and IV.


aggradation zone

Class VI. Quasi Equilibriumh<hc

Class IV. Degradation and Wideningh>hc

Class II. Channelizedh<hc

Class IClass III

Class IVClass V

Class VI

Class III. Degradationh<hc

Class V. Aggradation and Widening h>hc

terrace

terracefloodplain

bank

bankfullh

terrace

aggraded material

primarynickpoint

secondarynickpoint

plunge pool

precursornickpoint

oversteepened reach

aggraded material

slumped material

slumped material

h

hc = critical bank height

= direction of bank or bed movement

hh

Class I. Sinuous, Premodifiedh<hc

h

h

aggraded material

top bank

direction of flow

Figure 7.14: Channel evolution model. A disturbed orunstable stream is in varying stages of disequilibriumalong its length or profile. A channel evolution model theoretically may help predict future upstream ordownstream changes in habitat and stream morphology.Source: Simon 1989, USACE 1990.


■ CEMs can help match solutions to theproblems. Downcutting in Stage IIIoccurs due to the greater capacity ofthe stream created by construction, orearlier incision, in Stage II. The down-cutting in Stage III requires treat-ments such as grade control aimed atmodifying the factors causing the bot-tom instability. Bank stability prob-lems are dominant in Stages IV and V,so the approaches to stabilizationrequired are different from those forStage III. Stages I and VI typicallyrequire only maintenance activities.

■ CEMs can help provide goals ormodels for restoration. Reaches ofstreams in Stages I and VI are gradedstreams, and their profile, form, andpattern can be used as models forrestoring unstable reaches.

Limitations of Channel EvolutionModels

The chief limitations in using CEMs forstream restoration are as follows:

■ Future changes in base level eleva-tions and watershed water and sedi-ment yield are not considered whenpredicting channel response.

■ Multiple adjustments by the streamsimultaneously are difficult to pre-dict.

Class

No.

I

Name

Premodified

Fluvial

Dominant Processes

Hillslope

Characteristic Forms GeobotanicalEvidence

Sediment transport - mild aggradation; basal erosion on outside bends; deposition on inside bends.

Vegetated banks to flow line.

Stable, alternate channel bars; convex top-bank shape; flow line high relative to top bank; channel straight or meandering.

II Constructed Removal of vegetation.Trapezoidal cross section; linear bank surfaces; flow line lower relative to top bank.

III Degradation Degradation; basal erosion on banks.

Pop-out failures.

Riparian vegetation high relative to flow line and may lean toward channel.

Heightening and steepening of banks; alternate bars eroded; flow line lower relative to top bank.

IV Threshold Degradation; basal erosion on banks.

Slab, rotational and pop-out failures.

Riparian vegetation high relative to flow line and may lean toward channel.

Large scallops and bank retreat; vertical face and upper-bank surfaces; failure blocks on upper bank; some reduction in bank angles; flow line very low relative to top bank.

V Aggradation Aggradation; development of meandering thalweg; initial deposition of alternate bars; reworking of failed material on lower banks.

Slab, rotational and pop-out failures; low-angle slides of previously failed material.

Tilted and fallen riparian vegetation; reestablishing vegetation on slough line; deposition of material above root collars of slough line vegetation.

Large scallops and bank retreat; vertical face, upper bank, and slough line; flattening of bank angles; flow line low relative to top bank; development of new floodplain.

VI Restabilization Aggradation; further development of meandering thalweg; further deposition of alternate bars; reworking of failed material; some basal erosion on outside bends deposition of floodplain and bank surfaces.

Low-angle slides; some pop-out failures near flow line.

Reestablishing vegetation extends up slough line and upper bank; deposition of material above root collars of slough-line and upper-bank vegetation; some vegetation establishing on bars.

Stable, alternate channel bars; convex-short vertical face on top bank; flattening of bank angles; development of new floodplain; flow line high relative to top bank.

Table 7.4: Dominant hillslope and instreamprocesses, characteristic cross section shapeand bedforms, and condition of vegetation inthe various stages of channel evolution.Source: Simon 1989.


Applications of GeomorphicAnalysis

Stream classification systems and chan-nel evolution models may be used to-gether in resource inventories andanalysis to characterize and groupstreams. Although many classificationsystems are based on morphological pa-rameters, and channel evolution modelsare based on adjustment processes, thetwo approaches to stream characteriza-tion complement each other. Both indi-cate the present condition of a stream

reach under investigation, but character-ization of additional reaches upstreamand downstream of the investigationarea can provide an understanding ofthe overall trend of the stream.

Stream classification systems and chan-nel evolution models also provide in-

verticalface 70-90̊

slab androtationalfailures

pop-outfailures

upper bank25-40̊

slough line 20-25̊

Class I. Premodified Class II. Constructed Class III. Degradation

Class IV. Threshold

Class V. Aggradation

Class IVa. Threshold

fluvialdeposition

degradedchannelbottom

previousprofile

substantial bed-levelrecovery

convexshapeupper bank

25-35̊

non-dispersivematerials

fluvial deposition

moderately dispersivematerials

slough line 20̊

verticalface 70-90̊

Class VI. Restabilization Class VIa. Restabilization

fluvial deposition

previous profile substantial

bed-levelrecovery

previous profile

previousprofile

vertical face 70-90̊

upper bank25-50̊

previousprofile

Class IIIa. Degradation

under-cutting

Figure 7.15: Simon’s channel evolution stagesrelated to streambank shape. The cross-sectional shape of the streambank may be agood indicator of its evolutionary stage.Source: Simon 1989. Published by permission of theAmerican Water Resources Association.


sights as to the type of stability prob-lems occurring within the stream corri-dor and potential opportunities forrestoration. Gullied stream channels aredowncutting, so grade stabilization is re-quired before time and money are spenton bank stabilization or floodplainrestoration. Similarly, incised channelswith lateral instabilities are in the initialstages of widening, a process that oftenmust be accommodated before equilib-rium conditions can be attained. Al-though most argue that channelwidening must be accommodated to re-store incised channels, in some casesnot allowing the stream to widen mightbe preferred, depending on the valueand priority placed on adjacent landuse and structures within the corridor.

On the other hand, incised streams thathave widened enough for a new innerchannel and floodplain to begin form-ing are excellent candidates for vegeta-tion management since these streams

are already tending toward renewed sta-bility and establishing riparian vegeta-tion can accelerate the process.

Both the stream classification and thestage of channel evolution inventoriescan serve as the foundation for assess-ing systemwide stability. Channelwidth/depth ratio (F) at mean annualdischarge and the percent of silt andclay in the channel boundary (M) areuseful diagnostics for determining sys-temwide adjustments. These variablescan be plotted on Schumm’s (1960)curve of width/depth ratio versus per-cent silt-clay (F = 255M–1.08) to assessstability (Figure 7.16). Schumm’swidth/depth ratio is the top width ofthe bankfull channel and the deepestdepth in the bankfull channel crosssection. The term “M” is defined by therelationship

M = [(ScW) + (S

b2D)] / (W + 2D)

where

Sc= percentage of silt and clay in the

bed material

Sb

= percentage of silt and clay in thebank material

W = channel width

D = channel depth

Data from aggrading streams generallyplot above the line of best fit, whereasdata for degrading streams plot belowthe line. Schumm’s graph could also beused as a guide in selecting an appro-priate width/depth ratio for an incisedor recently disturbed channel.

Finally, classification systems and evolu-tion models can help guide the selec-tion of restoration treatments. Asmentioned above, there is little oppor-tunity for successfully establishingstreambank vegetation in streams withvertical and horizontal instability. Thebanks of such streams are subject todeep-seated slope failures that are not

Wid

th/D

epth

Rat

io (

F)

Weighted Mean Percent Silt-Clay (M)

0.5 1.01.0

10.0

100.0

400.0

10.0

F = 255M-1.08

100.0

Figure 7.16: Schumm’s F versus M relationship. Data for aggradingstreams generally plot above or to the right of the line. Degradingor incising streams plot below the line.Source: Schumm 1960.


usually prevented even by maturewoody vegetation. Conversely, estab-lishing and managing perennial grassesand woody vegetation is critical to pro-tecting streams that are already func-tioning properly.

Proper FunctioningCondition (PFC)

The Bureau of Land Management (BLM)has developed guidelines and proce-dures to rapidly assess whether a streamriparian area is functioning properly interms of its hydrology, landform/soils,channel characteristics, and vegetation(Prichard et al. 1993, rev. 1995). Thisassessment, commonly called PFC, isuseful as a baseline analysis of streamcondition and physical function, and itcan also be useful in watershed analysis.

It is essential to do a thorough analysisof the stream corridor and watershedconditions prior to development ofrestoration plans and selection ofrestoration approaches to be used.There are many cases where selectionof the wrong approach has led tocomplete failure of stream restorationefforts and the waste of costs of restora-tion. In many cases, particularly inwildland situations, restoration throughnatural processes and control of landuses is the preferred and most cost-ef-fective method. If hydrologic conditionsare rapidly changing in a drainage, norestoration might be the wisest courseuntil equilibrium is restored.

Identifying streams and drainageswhere riparian areas along streams arenot in proper functioning condition,and those at risk of losing function, isan important first step in restorationanalysis. Physical conditions in riparianzones are excellent indicators of what ishappening in a stream or the drainageabove.

With the results of PFC analysis, it ispossible to begin to determine streamcorridor and watershed restorationneeds and priorities. PFC results mayalso be used to identify where gatheringmore detailed information is neededand where additional data are notneeded.

PFC is a methodology for assessing thephysical functioning of a riparian-wetland area. It provides informationcritical to determining the “health” of ariparian ecosystem. PFC considers bothabiotic and biotic components as theyrelate to the physical functioning of ri-parian areas, but it does not considerthe biotic component as it relates tohabitat requirements. For habitat analy-sis, other techniques must be employed.

The PFC procedure is currently a stan-dard baseline assessment for stream/ri-parian surveys for the BLM, and PFC isbeginning to be used by the U.S. ForestService in the West. This technique isnot a substitute for inventory or moni-toring protocols designed to yield de-tailed information on the habitat orpopulations of plants or animals depen-dent on the riparian-stream ecosystem.

PFC is a useful tool for watershedanalysis. Although the assessment isconducted on a stream reach basis, theratings can be aggregated and analyzedat the watershed scale. PFC, along withother watershed and habitat conditioninformation, provides a good picture ofwatershed “health” and causal factorsaffecting watershed “health.” Use ofPFC will help to identify watershed-scale problems and suggest manage-ment remedies.

The following are definitions of properfunction as set forth in TR 1737-9:

■ Proper Functioning Condition—Riparian-wetland areas are function-ing properly when adequate vegeta-


tion, landform, or large woodydebris is present to:

1. Dissipate stream energy associatedwith high waterflows, therebyreducing erosion and improvingwater quality.

2. Filter sediment, capture bedload,and aid floodplain development.

3. Improve floodwater retention andground water storage.

4. Develop root masses that stabilizestreambanks against cuttingaction.

5. Develop diverse ponding andchannel characteristics to providethe habitat and the water depth,duration, and temperature neces-sary for fish production, waterfowlbreeding, and other uses.

6. Support greater biodiversity.

■ Functional-at Risk — Riparian-wetlandareas that are in functional condi-tion, but an existing soil, water, orvegetation attribute makes them sus-ceptible to degradation.

■ Nonfunctional — Riparian-wetlandareas that clearly are not providingadequate vegetation, landform, orlarge debris to dissipate stream ener-gy associated with high flow andthus are not reducing erosion, im-proving water quality, or performingother functions as listed above underthe definition of proper function.The absence of certain physicalattributes, such as absence of a floodplain where one should be, is an indicator of nonfunctioningconditions.

Assessing functionality with the PFCtechnique involves procedures for deter-mining a riparian-wetland area’s capa-

bility and potential, and comparingthat potential with current conditions.

Although the PFC procedure definesstreams without floodplains (when afloodplain would normally be present)as nonfunctional, many streams thatlose their floodplains through incisionor encroachment still retain ecologicalfunctions. The importance of a flood-plain needs to be assessed in view ofthe site-specific aquatic and ripariancommunity.

When using the PFC technique, it isimportant not to equate “proper func-tion” with “desired condition.” Properfunction is intended to describe thestate in which the stream channel andassociated riparian areas are in a rela-tively stable and self-sustaining condi-tion. Properly functioning streams canbe expected to withstand intermediateflood events (e.g., 25- to 30-year floodevents) without substantial damage toexisting values. However, proper func-tioning condition will often developwell before riparian succession providesshrub habitat for nesting birds. Put an-other way, proper functioning conditionis a prerequisite to a variety of desiredconditions.

Although based on sound science, thePFC field technique is not quantitative.An advantage of this approach is that it is less time-consuming than othertechniques because measurements arenot required. The procedure is per-formed by an interdisciplinary teamand involves completing a checklistevaluating 17 factors dealing with hy-drology, vegetation, and erosional/depositional characteristics. Training inthe technique is required, but the tech-nique is not difficult to learn. Withtraining, the functional determinationsresulting from surveys are reproducibleto a high degree.


Other advantages of the PFC techniqueare that it provides an easy-to-under-stand “language” for discussing streamconditions with a variety of agenciesand publics, PFC training is readilyavailable, and there is growing intera-gency acceptance of the technique.

Hydraulic Geometry: Streamsin Cross Section

Stream corridor restoration initiativesfrequently involve partial or total recon-struction of channels that have been se-verely degraded. Channelreconstruction design requires criteriafor channel size and alignment. The fol-lowing material presents an overview ofhydraulic geometry theory and providessome sample hydraulic geometry rela-tionships for relating bankfull dimen-sions to bankfull discharge.Correlations between certain planformdimensions (e.g., meander characteris-tics) of stable alluvial stream channelsto bankfull discharge and channelwidth also are discussed.

Hydraulic geometry theory is based onthe concept that a river system tends todevelop in a way that produces an ap-proximate equilibrium between thechannel and the in-flowing water andsediment (Leopold and Maddock1953). The theory typically relates anindependent or driving variable, such asdrainage area or discharge, to depen-dent variables such as width, depth,slope, and velocity. Hydraulic geometryrelations are sometimes stratified ac-cording to bed material size or otherfactors. These relationships are empiri-cally derived, and their development re-quires a relatively large amount of data.

Figure 7.17 presents hydraulic geome-try relations based on the mean annualdischarge rather than the bankfull dis-charge. Similar hydraulic geometry rela-tionships can be determined for awatershed of interest by measuring

channel parameters at numerous crosssections and plotting them against adischarge. Such plots can be used withcare for planning and preliminary de-sign. The use of hydraulic geometry re-lationships alone for final design is notrecommended.

Careful attention to defining stablechannel conditions, channel-formingdischarge, and streambed and bankcharacteristics are required in the datacollection effort. The primary role ofdischarge in determining channel crosssections has been clearly demonstrated,but there is a lack of consensus aboutwhich secondary factors such as sedi-ment loads, bank materials, and vegeta-tion are significant, particularly withrespect to width. Hydraulic geometry re-lationships that do not explicitly con-sider sediment transport are applicablemainly to channels with relatively lowbed-material loads (USACE 1994).

Hydraulic geometry relations can be de-veloped for a specific river, watershed,or for streams with similar physio-graphic characteristics. Data scatter isexpected about the developed curveseven in the same river reach. The moredissimilar the stream and watershedcharacteristics are, the greater the ex-pected data scatter is. It is important torecognize that this scatter represents avalid range of stable channel configura-tions due to variables such as geology,vegetation, land use, sediment load andgradation, and runoff characteristics.

Figures 7.18 and 7.19 show hydraulicgeometry curves developed for theupper Salmon River watershed in Idaho(Emmett 1975). The scatter of data forstable reaches in the watershed indicatesthat for a drainage area of 10 squaremiles, the bankfull discharge could rea-sonably range from 100 to 250 cfs andthe bankfull width could reasonablyrange from 10 to 35 feet. These relations


were developed for a relatively homoge-neous watershed, yet there is still quite abit of natural variation in the data. Thisillustrates the importance of viewingthe data used to develop any curve (notjust the curve itself), along with statisti-cal parameters such as R2 values andconfidence limits. (Refer to a text onstatistics for additional information.)

Given the natural variation related tostream and watershed characteristics,

the preferred source of data for a hy-draulic geometry relationship would bethe restoration initiative reach. Thischoice may be untenable due to channelinstability. The second preferred choiceis the project watershed, although caremust be taken to ensure that data areacquired for portions of the watershedwith physiographic conditions similarto those of the project reach.

Statistically, channel-forming dischargeis a more reliable independent variablefor hydraulic geometry relations thandrainage area. This is because the mag-nitude of the channel forming dischargeis the driving force that creates the ob-served channel geometry, and drainage

Dep

th (

feet

)W

idth

(fe

et)

Wid

th (

feet

per

sec

on

d)

Mean Annual Discharge (cfs)

101

2

5

2

5

1.0

0.5

10

20

50

100

200

500

10,000100

11

11

11

10

10

10

12

12

12

15

15

15

16

16

16

13

13

13

6

6

6

14

14

14

7

7

7

8

8

8

9

9

9

2

2

2

3

3

3

4

4

4

5

5

5

1

1

1

19

19

19

18

18

18

17

17

17

1000

Figure 7.17: Channel morphology related toaverage annual discharge. Width, depth, andvelocity in relation to mean annual dischargeas discharge increases downstream on 19 riversin Wyoming and Montana.Source: Leopold and Maddock 1953.


Ban

kfu

ll D

isch

arg

e (Q

B) (

cfs)

Drainage Area (DA) (square miles)

1 1010

QB = 28.3DA0.69

1000

100

10,000

10,0001000100

Ban

kfu

ll Su

rfac

e W

idth

(W

B)

(fee

t)

Drainage Area (DA) (square miles)

Road Creek

1 101

100

10

1000

10,0001000100

WB = 8.1DA0.38

Figure 7.18: Bankfull discharge versus drainagearea—Upper Salmon River area. Curves basedon measured data such as this can be valuabletools for designing restorations (Emmett 1975).

Figure 7.19: Bankfull surface width versusdrainage area—Upper Salmon River area.Local variations in bankfull width may besignificant. Road Creek widths are narrowerbecause of lower precipitation.


area is merely a surrogate for discharge.Typically, channel-forming dischargecorrelates best with channel width. Cor-relations with depth are somewhat lessreliable. Correlations with slope and ve-locity are the least reliable.

Hydraulic Geometry and StabilityAssessment

The use of hydraulic geometry relationsto assess the stability of a given channelreach requires two things. First, the wa-tershed and stream channel characteris-tics of the reach in question must bethe same as (or similar to) the data setused to develop the hydraulic geometryrelations. Second, the reasonable scatterof the data in the hydraulic geometryrelations must be known. If the data fora specific reach fall outside the reason-able scatter of data for stable reaches ina similar watershed, there is reason tobelieve that the reach in question maybe unstable. This is only an indicator,since variability in other factors (geol-ogy, land use, vegetation, etc.) maycause a given reach to plot high or lowon a curve. For instance, in Figure 7.17,the data points from the Road Creeksubbasin plot well below the line (nar-rower bankfull surface width) becausethe precipitation in this subbasin islower. These reaches are not unstable;they have developed smaller channelwidths in response to lower discharges(as one would expect).

In summary, the use of hydraulic geom-etry relations requires that the actualdata be plotted and the statistical coeffi-cients known. Hydraulic geometry rela-tions can be used as a preliminaryguide to indicate stability or instabilityin stream reaches, but these indicationsshould be checked using other tech-niques due to the wide natural variabil-ity of the data (see Chapter 8 for moreinformation on assessment of channelstability).

Regional Curves

Dunne and Leopold (1978) looked atsimilar relationships from numerouswatersheds and published regionalcurves relating bankfull channel dimen-

Regime theory was developed about a century ago byBritish engineers working on irrigation canals in what isnow India and Pakistan. Canals that required little main-tenance were said to be “ in regime,” meaning that theyconveyed the imposed water and sediment loads in astate of dynamic equilibrium, with width, depth, andslope varying about some long-term average. Theseengineers developed empirical formulas linking low-maintenance canal geometry and design discharge byfitting data from relatively straight canals carrying near-constant discharges (Blench 1957, 1969; Simons andAlbertson 1963). Since few streams will be restored tolook and act as canals, the regime relationships are notpresented here.

About 50 years later, hydraulic geometry formulas similarto regime relationships were developed by geomorphol-ogists studying stable, natural rivers. These rivers, ofcourse, were not straight and had varying discharges. Asample of these hydraulic geometry relationship is pre-sented in the table on the following page. In general,these formulas take the form:

w = k1

Qk2 D50

k3

D = k4

Qk5 D50

k6

S = k7

Qk8 D50

k9

where w and D are reach average width and depth infeet, S is the reach average slope, D

50is the median bed

sediment size in millimeters, and Q is the bankfull dis-charge in cubic feet per second. These formulas aremost reliable for width, less reliable for depth, and leastreliable for slope.


sions to drainage area (Figure 7.20).Using these curves, the width anddepth of the bankfull channel can beapproximated once the drainage area ofa watershed within one of these regionsis known. Obviously, more curves suchas these are needed for regions that ex-perience different topographic, geo-

logic, and hydrologic regimes; there-fore, additional regional relationshipsshould be developed for specific areasof interest. Several hydraulic geometryformulas are presented in Table 7.5.

Regional curves should be used only asindicators to help identify the channelgeometry at a restoration initiative site

Wid

th (

feet

)C

ross

-sec

tio

nal

Are

a (s

qu

are

feet

)B

an

kfu

ll D

imen

sio

ns

Dep

th (

feet

)

Drainage Area in Square Miles

.10.5

1

10

5

1

50

5

10

50

100

500

.05 1 5 10 50 100 500

Upper Green River, Wyoming

San Francisco Bay region at 30" annual precipitationEastern United States

Upper Salmon River, Idaho

Figure 7.20: Regional curves for bankfull channel dimensions versus drainage area. Curvesshowing channel dimensions relating to drainage area for a region of the country can be usefulin determining departure from “ normal” conditions. The use of such curves must be temperedwith an understanding of the limitations of the specific data that produced the curves.Source: Dunne and Leopold 1978.


because of the large degree of naturalvariation in most data sets. Publishedhydraulic geometry relationships usu-ally are based on stable, single-threadalluvial channels. Channel geometry-discharge relationships are more com-plex for multithread channels.

Exponents and coefficients for hydraulicgeometry formulas are usually deter-mined from data sets for a specificstream or watershed. The relativelysmall range of variation of the expo-nents k

2, k

5, and k

8is impressive, con-

sidering the wide range of situationsrepresented. Extremes for the data setsused to generate the hydraulic geometryformulas are given in Tables 7.6 and7.7. Because formula coefficients vary,applying a given set of hydraulic geom-etry relationships should be limited tochannels similar to the calibration sites.This principle severely limits applying

the Lacey, Blench, and Simons and Al-bertson formulas in channel restorationwork since these curves were developedusing canal data. Additionally, hydraulicgeometry relationships developed forpristine or largely undeveloped water-sheds should not be applied to urbanwatersheds.

As shown in Table 7.5, hydraulic geom-etry relationships for gravel-bed riversare far more numerous than those forsand-bed rivers. Gravel-bed relation-ships have been adjusted for bank soilcharacteristics and vegetation, whereassand-bed formulas have been modifiedto include bank silt-clay content(Schumm 1977). Parker (1982) arguesin favor of regime-type relationshipsbased on dimensionless variables. Ac-cordingly, the original form of theParker formula was based on dimen-sionless variables.

Lacey 1958

1 Blench (1969) provides adjustment factors for sediment concentrations between 30 and 100 ppm.

Indian canals 0.1 to 0.4 Cohesive to slightly cohesive

100 to 10,000

< 500

Reference Data Source Median BedMaterial Size(mm)

Banks Discharge(ft3/s)

SedimentConcentration(ppm)

Slope Bedforms

Blench 1969 Indian canals 0.1 to 0.6 Cohesive 1 to 100,000 < 301 Not specified

Ripples to dunes

Nixon 1959 U.K. rivers gravel 700 to 18,050

Not measured

Kellerhals 1967 U.S., Canadian, and Swiss rivers of low sinuousity, and lab

7 to 265 Noncohesive 1.1 to 70,600

Negligible .00017 to .0131

Plane

Bray 1982 Sinuous Canadian rivers

1.9 to 145 194 to 138,400

“ Mobile” bed .00022 to .015

Parker 1982 Single channel Canadian rivers

Little cohesion

353 to 211,900

Hey and Thorne 1986

Meandering U.K. rivers

14 to 176 138 to 14,970

Qs computed to range up to 114

.0011 to

.021

Simons and Albertson 1963

U.S. and Indian canals

0.318 to 0.465

Sand 100 to 400 < 500 .000135 to .000388

Ripples to dunes

0.06 to 0.46 Cohesive 5 to 88,300 < 500 .000059 to .00034

Ripples to dunes

Cohesive,0.029 to 0.36

Cohesive 137 to 510 < 500 .000063 to .000114

Plane

Table 7.5: Limits of data sets used to derive regime formulas.Source: Hey 1988, 1990.


Planform and MeanderGeometry: Stream ChannelPatterns

Meander geometry variables are shownin Figure 7.21. Channel planformparameters may be measured in thefield or from aerial photographs andmay be compared with published rela-tionships, such as those identified inthe box. Developing regional relation-

ships or coefficients specific to the siteof interest is, however, preferable tousing published relationships that mayspan wide ranges in value. Figure 7.22shows some planform geometry rela-tions by Leopold (1994). Meandergeometries that do not fall within therange of predicted relationships mayindicate stream instability and deserveattention in restoration design.

Author

Nixon

Year

1959

Data Domain

Gravel-bed rivers

U.K. rivers

k1 k2

0.5

k3 k4

0.545

k5

0.33

k6 k7

1.258n2b

k8

-0.11

k9

Leopoldet al.

1964 Midwestern U.S.

1.65 0.5 0.4 -0.49

Ephemeral streams in semiarid U.S.

0.5 0.3 -0.95

Kellerhals 1967 Gravel-bed rivers with paved beds and small bed material concentration

Field (U.S., Canada, and Switzerland) and laboratory

1.8 0.5 0.33 0.4 -0.12a 0.00062 -0.4 0.92a

Schumm 1977 Sand-bed rivers with properties shown in Table 6

U.S. (Great Plains) and Australia (Riverine Plains of New South Wales)

37k1* 0.38 0.6k4* 0.29 -0.12a 0.01136k7* -0.32

Bray 1982 Gravel-bed rivers

Canadian rivers

3.1 0.53 -0.07 0.304 0.33 -0.03 0.00033 -0.33 0.59

Parker 1982 Gravel-bed rivers, banks with little cohesion

Single-channel Alberta rivers

6.06 0.444 -0.11 0.161 0.401 -0.0025 0.00127 -0.394 0.985

Hay andThorne

1986 Gravel-bed rivers with:U.K. rivers

Grassy banks with no trees or shrubs

2.39 0.5 0.41 0.37 -0.11 0.00296k7** -0.43 -0.09

1-5% tree/shrub cover

1.84 0.5 0.41 0.37 -0.11 0.00296k7** -0.43 -0.09

Greater than 5-50% tree/shrub cover

1.51 0.5 0.41 0.37 -0.11 0.00296k7** -0.43 -0.09

Greater than 50% shrub cover or incised flood plain

1.29 0.5 0.41 0.37 -0.11 0.00296k7** -0.43 -0.09

a Bed material size in Kellerhals ’ equation is D90.bn = Manning n.

k1* = M-0.39, where M is the percent of bank materials finer than 0.074 mm. The discharge used in this equation is mean annual rather than bankfull.

k4* = M0.432, where M is the percent of bank materials finer than 0.074 mm. The discharge used in this equation is mean annual rather than bankfull.

k7* = M-0.36, where M is the percent of bank materials finer than 0.074 mm. The discharge used in this equation is mean annual rather than bankfull.

k7** = D540.84 Qx

0.10, where Qx = bed material transport rate in kg s-1 at water discharge Q, and D54 refers to bed material and is in mm.

Table 7.6: Coefficients for selected hydraulic geometry formulas.


Stream System Dynamics

Stream management and restorationrequire knowledge of the complex inter-actions between watershed and streamprocesses, boundary sediments, andbank and floodplain vegetation. Identi-fying the causes of channel instabilityor potential instability and havingknowledge of the magnitude and distri-bution of channel adjustment processesare important for the following:

■ Estimating future channel changes.

■ Developing appropriate mitigationmeasures.

■ Protecting the stream corridor.

Derived empirical equations for river-meander and channel-size features. A = bankfull cross-sectional area.W = bankfull width.D = bankfull mean depth.Lm = meander wavelength.Lb = along-channel bend length.B = meander belt width.Rc = loop radius of curvature.K = channel sinuosity.

Equation Number

Interrelations between meander features

Equation Applicable Range Equation Number

Equation Applicable Range

2 Lm = 1.25Lb 18.0 ≤ Lb ≤ 43,600 ft

3 Lm = 1.63B 12.1 ≤ B ≤ 44,900 ft

4 Lm = 4.53Rc 8.5 ≤ Rc ≤ 11,800 ft

5 Lb = 0.8Lm 26 ≤ Lm ≤ 54,100 ft

6 Lb = 1.29B 12.1 ≤ B ≤ 32,800 ft

7 Lb = 3.77Rc 8.5 ≤ Rc ≤ 11,800 ft

8 B = 0.61Lm 26 ≤ Lm ≤ 76,100 ft

9 B = 0.78Lb 18.0 ≤ Lb ≤ 43,600 ft

10 B = 2.88Rc 8.5 ≤ Rc ≤ 11,800 ft

11 Rc = 0.22Lm 33 ≤ Lm ≤ 54,100 ft

12 Rc = 0.26Lb 22.3 ≤ Lb ≤ 43,600 ft

13 Rc = 0.35B 16 ≤ B ≤ 32,800 ft

Relations of channel size to meander features

14 A = 0.0094Lm1.53 33 ≤ Lm ≤ 76,100 ft

15 A = 0.0149Lb1.53 20 ≤ Lb ≤ 43,600 ft

16 A = 0.021B1.53 16 ≤ B ≤ 38,100 ft

17 A = 0.117Rc1.53 7 ≤ Rc ≤ 11,800 ft

18 W = 0.019Lm0.89 26 ≤ Lm ≤ 76,100 ft

19 W = 0.026Lb0.89 16 ≤ Lb ≤ 43,600 ft

20 W = 0.031B0.89 10 ≤ B ≤ 44,900 ft

21 W = 0.81Rc0.89 8.5 ≤ Rc ≤ 11,800 ft

22 D = 0.040Lm0.66 33 ≤ Lm ≤ 76,100 ft

23 D = 0.054Lb0.66 23 ≤ Lb ≤ 43,600 ft

24 D = 0.055B0.66 16 ≤ B ≤ 38,100 ft

25 D = 0.127Rc0.66 8.5 ≤ Rc ≤ 11,800 ft

Relations of meander features to channel size

26 Lm = 21A0.65 0.43 ≤ A ≤ 225,000 ft

27 Lb = 15A0.65 0.43 ≤ A ≤ 225,000 ft

28 B = 13A0.65 0.43 ≤ A ≤ 225,000 ft

29 Rc = 4.1A0.65 0.43 ≤ A ≤ 225,000 ft

30 Lm = 6.5W1.12 4.9 ≤ W ≤ 13,000 ft

31 Lb = 4.4W1.12 4.9 ≤ W ≤ 7,000 ft

32 B = 3.7W1.12 4.9 ≤ W ≤ 13,000 ft

33 Rc = 1.3W1.12 4.9 ≤ W ≤ 7,000 ft

34 Lm = 129D1.52 0.10 ≤ D ≤ 59 ft

35 Lb = 86D1.52 0.10 ≤ D ≤ 57.7 ft

36 B = 80D1.52 0.10 ≤ D ≤ 59 ft

37 Rc = 23D1.52 0.10 ≤ D ≤ 57.7 ft

38 W = 12.5D1.45 0.10 ≤ D ≤ 59 ft

39 D = 0.17W0.89 4.92 ≤ W ≤ 13,000 ft

40 W = 73D1.23K-2.35 0.10 ≤ D ≤ 59 ft and 1.20 ≤ K ≤ 2.60

41 D = 0.15w0.50K1.48 4.9 ≤ W ≤ 13,000 ft and 1.20 ≤ K ≤ 2.60

Relations between channel width, channel depth, and channel sinuosity

L

L meander wavelengthML meander arc lengthw average width at bankfull dischargeMA meander amplituderc radius of curvature

arc angle

w

rc MA

ML

Figure 7.21: Meander geometry variables.Adapted from Williams 1986.

Table 7.7: Meander geometry equations.Source: Williams 1986.


Adjustment processes that affect entirefluvial systems often include channelincision (lowering of the channel bedwith time), aggradation (raising of thechannel bed with time), planformgeometry changes, channel widening ornarrowing, and changes in the magni-tude and type of sediment loads. Theseprocesses differ from localizedprocesses, such as scour and fill, whichcan be limited in magnitude and extent.

In contrast, the processes of channelincision and aggradation can affect longreaches of a stream or whole streamsystems. Long-term adjustmentprocesses, such as incision, aggradation,and channel widening, can exacerbatelocal scour problems. Whetherstreambed erosion occurs due to local

scour or channel incision, sufficient bedlevel lowering can lead to bank instabil-ity and to changes in channel planform.

It is often difficult to differentiate be-tween local and systemwide processeswithout extending the investigation up-stream and downstream of the site inquestion. This is because channels mi-grate over time and space and so mayaffect previously undisturbed reaches.For example, erosion at a logjam ini-tially may be attributed to the deflec-tion of flows caused by the woodydebris blocking the channel. However,the appearance of large amounts ofwoody debris may indicate upstreamchannel degradation related to instabil-ity of larger scope.

Reviews of meander geometry formulas are provided by Nunnally and Shields (1985,Table 3) and Chitale (1973). Ackers and Charlton (1970) developed a typical formulathat relates meander wavelength and bankfull discharge, Q (cfs), using laboratory dataand checking against field data from a wide range of stream sizes:

L = 38Q0.467

There is considerable scatter about this regression line; examination of the plotted datais recommended. Other formulas, such as this one by Schumm (1977), also incorporatebed sediment size or the fraction of silt-clay in the channel perimeter:

L = 1890Qm

0.34/ M0.74

where Qmis average discharge (cfs) and M is the percentage of silt-clay in the perimeter

of the channel. These types of relationships are most powerful when developed fromregional data sets with conditions that are typical of the area being restored. Radius ofcurvature, r

c, is generally between 1.5 and 4.5 times the channel width, w, and more

commonly between 2w and 3w, while meander amplitude is 0.5 to 1.5 times themeander wavelength, L (USACE 1994). Empirical (Apmann 1972, Nanson and Hickin1983 ) and analytical (Begin 1981) results indicate that lateral migration rates aregreatest for bends with radii of curvature between 2w and 4w.


Determining StreamInstability: Is It Local orSystemwide?

Stage of channel evolution is the pri-mary diagnostic variable for differentiat-ing between local and systemwidechannel stability problems in a dis-turbed stream or constructed channel.During basinwide adjustments, stage ofchannel evolution usually varies system-atically with distance upstream. Down-stream sites might be characterized byaggradation and the waning stages ofwidening, whereas upstream sites mightbe characterized (in progressive up-stream order) by widening and milddegradation, then degradation, and ifthe investigation is extended far enoughupstream, the stable, predisturbed con-dition (Figure 7.23). This sequence ofstages can be used to reveal systemwideinstabilities. Stream classification can beapplied in a similar manner to naturalstreams. The sequence of stream typescan reveal systemwide instabilities.

Restoration measures often fail, not asthe result of inadequate structural de-

Figure 7.23: Bank instability. Determining ifinstability is localized or systemwide is impera-tive to establish a correct path of action.

Mea

nd

er L

eng

th (

feet

)

Channel Width (feet) Mean Radius of Curvature (feet)

1 10

L = 10.9w1.01

meanders of rivers and in flumesmeanders of Gulf Streammeanders on glaciers

L = 4.7rm0.98

10

100

10,000

1000

100,000

1,000,000

100,000

100,000

10,000

10,000

100 1000 100105 1000

Figure 7.22: Planform geometry relationships.Meander geometries that do not plot close tothe predicted relationship may indicate streaminstability.Source: Leopold 1994.


sign, but rather because of the failure ofthe designers to incorporate the existingand future channel morphology intothe design. For this reason, it is impor-tant for the designer to have some gen-eral understanding of stream processesto ensure that the selected restorationmeasures will work in harmony withthe existing and future river conditions.This will allow the designer to assesswhether the conditions at a particularsite are due to local instability processesor are the result of some systemwide in-stability that may be affecting the entirewatershed.

Systemwide Instability

The equilibrium of a stream system canbe disrupted by various factors. Oncethis occurs, the stream will attempt toregain equilibrium by making adjust-ments in the dependent variables. Theseadjustments in the context of physicalprocesses are generally reflected inaggradation, degradation, or changes inplanform characteristics (meanderwavelength, sinuosity, etc.). Dependingon the magnitude of the change andthe basin characteristics (bed and bankmaterials, hydrology, geologic or man-made controls, sediment sources, etc.),these adjustments can propagatethroughout the entire watershed andeven into neighboring systems. For thisreason, this type of disruption of theequilibrium condition is referred to assystem instability. If system instability isoccurring or expected to occur, it is im-perative that the restoration initiativeaddress these problems before any bankstabilization or instream habitat devel-opment is considered.

Local Instability

Local instability refers to erosion anddeposition processes that are not symp-tomatic of a disequilibrium conditionin the watershed (i.e., system instabil-

ity). Perhaps the most common form oflocal instability is bank erosion alongthe concave bank in a meander bendthat is occurring as part of the naturalmeander process. Local instability canalso occur in isolated locations as theresult of channel constriction, flow ob-structions (ice, debris, structures, etc.),or geotechnical instability. Local insta-bility problems are amenable to localbank protection. Local instability canalso exist in channels where severe sys-tem instability exists. In these situa-tions, the local instability problems willprobably be accelerated due to the sys-tem instability, and a more comprehen-sive treatment plan will be necessary.

Caution must be exercised if only localtreatments on one site are implemen-ted. If the upstream reach is stable and the downstream reach is unstable, a systemwide problem may again be indicated. The instability may continuemoving upstream unless the root causeof the instability at the watershed levelis removed or channel stabilization atand downstream of the site is imple-mented.

Local channel instabilities often can beattributed to redirection of flow causedby debris, structures, or the approachangle from upstream. During moderateand high flows, obstructions often re-sult in vortices and secondary-flow cellsthat accelerate impacts on channelboundaries, causing local bed scour,erosion of bank toes, and ultimatelybank failures. A general constriction ofthe channel cross section from debrisaccumulation or a bridge causes a back-water condition upstream, with acceler-ation of the flow and scour through theconstriction.

Bed Stability

In unstable channels, the relationshipbetween bed elevation and time (years)


can be described by nonlinear func-tions, where change in response to adisturbance occurs rapidly at first andthen slows and becomes asymptoticwith time (Figure 7.24). Plotting bedelevations against time permits evaluat-ing bed-level adjustment and indicateswhether a major phase of channel inci-sion has passed or is ongoing. Variousmathematical forms of this functionhave been used to characterize bed-leveladjustment at a site and to predict fu-ture bed elevations. This method alsocan provide valuable information ontrends of channel stability at gaugedlocations where abundant data fromdischarge measurements are available.

Specific Gauge Analysis

Perhaps one of the most useful toolsavailable to the river engineer or geo-morphologist for assessing the histori-cal stability of a river system is thespecific gauge record. A specific gaugerecord is a graph of stage for a specificdischarge at a particular stream gauginglocation plotted against time (Blench1969). A channel is considered to be inequilibrium if the specific gauge recordshows no consistent increasing or de-creasing trends over time, while an in-creasing or decreasing trend isindicative of an aggradational or degra-dational condition, respectively. An ex-ample of a specific gauge record isshown in Figure 7.25.

The first step in a specific gauge analysisis to establish the stage vs. discharge re-lationship at the gauge for the period ofrecord being analyzed. A rating curve isdeveloped for each year in the period ofrecord. A regression curve is then fittedto the data and plotted on the scatterplot. Once the rating curves have beendeveloped, the discharges to be used inthe specific gauge record must be se-lected. This selection depends largelyon the objectives of the study. It is usu-ally advisable to select discharges thatencompass the entire range of observedflows. A plot is then developed showingthe stage for the given flow plottedagainst time.

Specific gauge records are an excellenttool for assessing the historical stabilityat a specific location. However, specificgauge records indicate only the condi-tions in the vicinity of the particulargauging station and do not necessarilyreflect river response farther upstreamor downstream of the gauge. Therefore,even though the specific gauge record isone of the most valuable tools used byriver engineers, it should be coupledwith other assessment techniques to

Elev

atio

n o

f C

han

nel

Bed

(fe

et a

bo

ve s

ea le

vel)

1962 1966Year

270

275

280

285

290

295

1970

South Fork Obion Riverriver mile = 5.9

1974 1978 1982 1986

1958 1962276

280

284

288

292

1986

Rutherford Fork Obion Riverriver mile = 4.9

1970 1974 1978 1982 1986

1964 1968255

260

265

270South Fork Forked Deer River

river mile = 7.9

1972 1976 1980 1984

Figure 7.24:Changes in bedelevations overtime. Plotting riverbed elevations ata point along theriver over time canindicate whethera major phase ofchannel incision isongoing or haspassed.


assess reach conditions or to make pre-dictions about the ultimate response ona river.

Comparative Surveys and Mapping

One of the best methods for directly as-sessing channel changes is to comparechannel surveys (thalweg and cross section).

Thalweg surveys are taken along thechannel at the lowest point in the crosssection. Comparison of several thalwegsurveys taken at different points in timeallows the engineer or geomorphologistto chart the change in the bed elevationthrough time (Figure 7.26).

Certain limitations should be consid-ered when comparing surveys on ariver system. When comparing thalwegprofiles, it is often difficult, especiallyon larger streams, to determine anydistinct trends of aggradation or degra-

dation if there are large scour holes,particularly in bendways. The existenceof very deep local scour holes maycompletely obscure temporal variationsin the thalweg. This problem can some-times be overcome by eliminating thepool sections and focusing only on thecrossing locations, thereby allowingaggradational or degradational trendsto be more easily observed.

Although thalweg profiles are a usefultool, it must be recognized that they re-flect only the behavior of the channelbed and do not provide informationabout the channel as a whole. For thisreason it is usually advisable to studychanges in the cross-sectional geometry.Cross-sectional geometry refers towidth, depth, area, wetted perimeter,hydraulic radius, and channel con-veyance at a specific cross section.

If channel cross sections are surveyedat permanent monumented rangelocations, the cross-sectional geometryat different times can be compared

Gag

e H

eig

ht

(msl

)

Year

19401930250

255

260

265

270

1950 1960

60,000 cfs

40,000 cfs

26,000 cfs

11,000 cfs

6000 cfs

3800 cfs

1970 1980 1990

Figure 7.25: Specific gauge plot for Red Riverat Index, Arkansas. Select discharges from thegauge data that represent the range of flows.Source: Biedenharn et al. 1997.


directly. The cross section plots for eachrange at the various times can be over-laid and compared. It is seldom thecase, however, that the cross sections arelocated in the exact same place yearafter year. Because of these problems, itis often advisable to compare reach-average values of the cross-sectionalgeometry parameters. This requires thestudy area to be divided into distinctreaches based on geomorphic character-istics. Next, the cross-sectional parame-ters are calculated at each cross sectionand then averaged for the entire reach.Then the reach-average values can becompared for each survey. Cross-sectional variability between bends(pools) and crossings (riffles) can ob-scure temporal trends, so it is oftenpreferable to use only cross sectionsfrom crossing reaches when analyzinglong-term trends of channel change.

Comparison of time-sequential mapscan provide insight into the planforminstability of the channel. Rates andmagnitude of channel migration (bankcaving), locations of natural and man-made cutoffs, and spatial and temporalchanges in channel width and planformgeometry can be determined frommaps. With these types of data, channelresponse to imposed conditions can bedocumented and used to substantiatepredictions of future channel responseto a proposed alteration. Planform datacan be obtained from aerial photos,maps, or field investigations.

Regression Functions for Degradation

Two mathematical functions have beenused to describe bed level adjustmentswith time. Both may be used to predictchannel response to a disturbance, sub-ject to the caution statements below.The first is a power function (Simon1989a):

E = a tb

where E = elevation of the channel bed,in feet; a = coefficient, determined by

Elev

atio

n (

feet

)

Stationing (100 feet)

260+00

county road bridge

1977 thalweg1985 thalweg

260

280

300

320

340

280+00 300+00 320+00 340+00 360+00 380+00

Figure 7.26: Comparative thalweg profiles.Changes in bed elevation over the length of a stream can indicate areas of transition andreaches where more information is needed.Source: Biedenharn et al., USACE 1997.


regression, representing the premodi-fied elevation of the channel bed, infeet; t = time since beginning of adjust-ment process, in years, where t

0= 1.0

(year prior to onset of the adjustmentprocess); and b = dimensionless expo-nent, determined by regression and in-dicative of the nonlinear rate of channelbed change (negative for degradationand positive for aggradation).

The second function is a dimensionlessform of an exponential equation(Simon 1992):

z / z0

= a + b e (– k t)

where

z = the elevation of the channel bed(at time t)

z0

= the elevation of the channel bedat t

0

a = the dimensionless coefficient,determined by regression andequal to the dimensionless ele-vation (z/z

0) when the equation

becomes asymptotic, a>1 =aggradation, a<1 = degradation

b = the dimensionless coefficient,determined by regression andequal to the total change in thedimensionless elevation (z/z

0)

when the equation becomes as-ymptotic

k = the coefficient determined by re-gression, indicative of the rate ofchange on the channel bed perunit time

t = the time since the year prior tothe onset of the adjustmentprocess, in years (t

0=0)

Future elevations of the channel bedcan, therefore, be estimated by fittingthe equations to bed elevations and bysolving for the period of interest. Eitherequation provides acceptable results,depending on the statistical significanceof the fitted relation. Statistical signifi-

cance of the fitted curves improves withadditional data. Degradation and aggra-dation curves for the same site are fitseparately. For degrading sites, the equa-tions will provide projected minimumchannel elevations when the value of tbecomes large and, by subtracting thisresult from the floodplain elevation,projected maximum bank heights. Arange of bed adjustment trends can beestimated by using different startingdates in the equations when the initialtiming of bed level change is unknown.Use of the equations, however, may belimited in some areas because of a lackof survey data.

Regression Functions for Aggradation

Once the minimum bed elevation hasbeen obtained, that elevation can beused as the starting elevation at a newt

0for the secondary aggradation phase

that occurs during channel widening(see discussion of channel evolutionabove). Secondary aggradation occurs ata site after degradation reduces channelgradient and stream power to such anextent that sediment loads deliveredfrom degrading reaches upstream canno longer be transported (Simon1989a). Coefficient values for Simon’spower function for estimating sec-ondary aggradation can be obtained ei-ther from interpolating existing data orfrom estimating their values as about60 percent less than the correspondingvalue obtained for the degradationphase.

The variation of the regression coeffi-cients a and b with longitudinal dis-tance along the channel can be used asan empirical model of bed level adjust-ment providing there are data fromenough sites. Examples using bothequations are provided for the ObionRiver system, West Tennessee (Figure7.27). Estimates of bed-level changewith time for unsurveyed sites can be


obtained using interpolated coefficienta values and t

0. For channels down-

stream from dams without significanttributary sediment inputs, the shape ofthe a-value curve would be similar butinverted; maximum amounts of degra-dation (minimum a values) occur im-mediately downstream of the dam andattenuate nonlinearly with distance far-ther downstream.

Caution: If one of the above mathemati-cal functions is used to predict futurebed elevations, the assumption is madethat no new disturbances have occurredto trigger a new phase of channelchange. Downstream channelization,construction of a reservoir, formation ofa large woody debris jam that blocksthe channel, or even a major flood areexamples of disturbances that can trig-ger a new period of rapid change.

a-V

alu

e

Distance Upstream from Mouth (miles)

.96

.94

.98

1.00

1.02

1.04

1.06

20 30 40 50 60

1st adjustment2nd adjustment

area of maximumdisturbance andmouth of ObionRiver forks

a-values from equationz/zØ = a + be(-kt)

70 80 90 100

b-V

alu

e

Distance Upstream from Mouth (miles)

-0.025

-0.020

-0.030

-0.015

-0.010

-0.005

-0.000

-0.005

-0.010

20 30 40 50 60

1st adjustment2nd adjustment

70 80 90 100

area of maximumdisturbance andmouth of ObionRiver forks

b-values from equation E = atb

Figure 7.27:Coefficient a and bvalues for regressionfunctions for esti-mating bed leveladjustment versusongitudinal distancealong stream. Futurebed elevations canbe estimated byusing empirical equations.Source: Simon 1989,992.


The investigator is cautioned that theuse of regression functions to computeaggradation and degradation is an em-pirical approach that might be appro-priate for providing insight into thedegradational and aggradationalprocesses during the initial planningphases of a project. However, this pro-cedure does not consider the balancebetween supply and transport of waterand sediment and, therefore, is not ac-ceptable for the detailed design ofrestoration features.

Sediment Transport Processes

This document does not provide com-prehensive coverage of sedimentationprocesses and analyses critical to streamrestoration. These processes includeerosion, entrainment, transport, deposi-tion, and compaction. Refer to standardtexts and reference on sediment, includ-ing Vanoni (1975), Simons and Senturk(1977), Chang (1988), Richards(1982), and USACE (1989a).

Numerical Analyses and Modelsto Predict Aggradation andDegradation

Numerical analyses and models such asHEC-6 are used to predict aggradationand degradation (incision) in streamchannels, as discussed in Chapter 8.

Bank Stability

Streambanks can be eroded by movingwater removing soil particles or by col-lapse. Collapse or mass failure occurswhen the strength of bank materials istoo low to resist gravity forces. Banksthat are collapsing or about to collapseare referred to as being geotechnicallyunstable (Figure 7.28). The physicalproperties of bank materials should bedescribed to aid characterization of po-tential stability problems and identifica-

tion of dominant mechanisms of bankinstability.

The level of intensity of geotechnicalinvestigations varies in planning anddesign. During planning, enough infor-mation must be collected to determinethe feasibility of alternatives being con-sidered. For example, qualitative de-scriptions of bank stratigraphyobtained during planning may be allthat is required for identifying domi-nant modes of failure in a study reach.Thorne (1992) describes stream recon-naissance procedures particularly forrecording streambank data.

Qualitative Assessment of BankStability

Natural streambanks frequently arecomposed of distinct layers reflectingthe depositional history of the bankmaterials. Each individual sedimentlayer can have physical properties quitedifferent from those of other layers. Thebank profile therefore will respond ac-cording to the physical properties ofeach layer. Since the stability of stream-

Figure 7.28: Bank erosion by undercutting.Removal of toe slope support leads to instabilityrequiring geotechnical solutions.


banks with respect to failures due togravity depends on the geometry of thebank profile and the physical propertiesof the bank materials, dominant failuremechanisms tend to be closely associ-ated with characteristic stratigraphy orsuccession of layers (Figure 7.29).

A steep bank consisting of uniform lay-ers of cohesive or cemented soils gener-ally develops tension cracks at the topof the bank parallel to the bank align-ment. Slab failures occur when theweight of the soil exceeds the strengthof the grain-to-grain contacts within thesoil. As clay content or cementing agentdecreases, the slope of the bank de-creases; vertical failure planes becomemore flat and planar failure surfaces de-velop. Rotational failures occur whenthe bank soils are predominantly cohe-sive. Block-type failures occur when aweak soil layer is eroded away and thelayers above the weak layer lose struc-tural support.

The gravity failure processes describedin Figure 7.29 usually occur after thebanks have been saturated due to pre-cipitation or high stream stages. Thewater adds weight to the soil and re-duces grain-to-grain contacts and cohe-sion forces while increasing the porepressure. Pore pressure occurs when soilwater in the pore spaces is under pres-sure from overlying soil and water. Porepressure therefore is internal to the soilmass. When a stream is full, the flowing

Figure 7.29: Relationship of dominant bankfailure mechanisms and associated stratigraph-ics. (a) Uniform bank undergoing planar typefailure (b) Uniform shallow bank undergoingrotational type failure (c) Cohesive upper bank,noncohesive lower bank leads to cantilevertype failure mechanism (d) Complex bankstratigraphy may lead to piping or sappingtype failures.Source: Hagerty 1991. In Journal of HydraulicEngineering. Vol. 117 Number 8. Reproduced bypermission of ASCE.

3. Failed Blocks Topple or Slide

outflowcontinues

fine-grainedsoil layer

sandy pervioussoil layer

outflowcontinues

D.

outflow ofsand andwater

fine-grained soil layer

fine-grained soil layer

fine-grained soil layers




2. Undermined Upper Layer Falls, Blocks Detached


1. Seepage Outflow Initiates Soil Loss

C.

B.

A.steep bankprofile

overhanggenerated onupper bank

preferentialretreat oferodiblebasal layer

fine-grained cohesiveupper bank

arcuate orrotationalfailure surface

shallow bankprofile

incipient failure plane

coarse non-cohesivelower bank

planarfailure surface


water provides some support to thestreambanks. When the stream leveldrops, the internal pore pressure pushesout from within and increases the po-tential for bank failure.

The last situation described in Figure7.29 involves ground water sapping orpiping. Sapping or piping is the erosionof soil particles beneath the surface byflowing ground water. Dirty or sediment-laden seepage from a streambank indi-cates ground water sapping or piping isoccurring. Soil layers above the areas ofground water piping eventually will col-lapse after enough soil particles havebeen removed from the support layer.

Quantitative Assessment of BankStability

When restoration design requires morequantitative information on soil prop-erties, additional detailed data need tobe collected (Figure 7.30). Values of co-hesion, friction angle, and unit weightof the bank material need to be quanti-fied. Because of spatial variability, care-ful sampling and testing programs arerequired to minimize the amount ofdata required to correctly characterizethe average physical properties of indi-vidual layers or to determine a bulk av-erage statistic for an entire bank.

Care must be taken to characterize soilproperties not only at the time of mea-surement but also for the “worst case”conditions at which failure is expected(Thorne et al. 1981). Unit weight, cohe-sion, and friction angle vary as a func-tion of moisture content. It usually isnot possible to directly measure bankmaterials under worst-case conditions,due to the hazardous nature of unstablesites under such conditions. A qualifiedgeotechnical or soil mechanics engineershould estimate these operationalstrength parameters.

Quantitative analysis of bank instabili-ties is considered in terms of force and

resistance. The shear strength of thebank material represents the resistanceof the boundary to erosion by gravity.Shear strength is composed of cohesivestrength and frictional strength. For thecase of a planar failure of unit length,the Coulomb equation is applicable

Sr= c + (N – µ) tan φ

where Sr= shear strength, in pounds per

square foot; c = cohesion, in poundsper square foot; N = normal stress, inpounds per square foot; µ = pore pres-sure, in pounds per square foot; and φ =friction angle, in degrees.Also:

N = W cos θ

where W = weight of the failure block,in pounds per square foot; and θ =angle of the failure plane, in degrees.

H = bank heightL = failure plane lengthc = cohesion

= friction angle= bulk unit weight

W = weight of failure blockI = bank angleSa = Wsin (driving force)Sr = cL + Ntan (resisting force)N = Wcos

= (0.5I = 0.5 ) (failure plane angle)

Explanation

bank

stream bed

W

N

L

I

Sa

SrH

for the critical case Sa = Sr and:

soilproperties

Hc = (1 - cos [I - ])4c sin I cos

Figure 7.30: Forces acting on a channel bankassuming there is zero pore-water pressure.Bank stability analyses relate strength of bankmaterials to bank height and angles, and tomoisture conditions.


The gravitational force acting on thebank is:

Sa

= W sin θ

Factors that decrease the erosional resis-tance (S

r), such as excess pore pressure

from saturation and the developmentof vertical tension cracks, favor bank in-stabilities. Similarly, increases in bankheight (due to channel incision) andbank angle (due to undercutting) favorbank failure by increasing the gravita-tional force component. In contrast,vegetated banks generally are drier andprovide improved bank drainage, whichenhances bank stability. Plant rootsprovide tensile strength to the soil re-sulting in reinforced earth that resistsmass failure, at least to the depth ofroots (Yang 1996).

Bank Instability and ChannelWidening

Channel widening is often caused byincreases in bank height beyond thecritical conditions of the bank material.Simon and Hupp (1992) show thatthere is a positive correlation betweenthe amount of bed level lowering bydegradation and amounts of channelwidening. The adjustment of channelwidth by mass-wasting processes repre-sents an important mechanism of chan-nel adjustment and energy dissipation inalluvial streams, occurring at rates cover-ing several orders of magnitude, up tohundreds of feet per year (Simon 1994).

Present and future bank stability may beanalyzed using the following procedure:

■ Measure the current channel geome-try and shear strength of the channelbanks.

■ Estimate the future channel geome-tries and model worst-case pore pres-sure conditions and average shearstrength characteristics.

For fine-grained soils, cohesion andfriction angle data can be obtainedfrom standard laboratory testing (triax-ial shear or unconfined compressiontests) or by in situ testing with a bore-hole shear test device (Handy and Fox1967, Luttenegger and Hallberg 1981,Thorne et al. 1981, Simon and Hupp1992). For coarse-grained, cohesionlesssoils, estimates of friction angles can beobtained from reference manuals. Bycombining these data with estimates offuture bed elevations, relative bank sta-bility can be assessed using bank stabil-ity charts.

Bank Stability Charts

To produce bank stability charts such asthe one following, a stability number(N

s) representing a simplification of the

bank (slope) stability equations is used.The stability number is a function ofthe bank-material friction angle (φ) andthe bank angle (i) and is obtained froma stability chart developed by Chen(1975) (Figure 7.31) or from Lohnesand Handy (1968):

Ns= (4 sin i cos φ) / [1 – cos (i – φ)]

The critical bank height Hc, where dri-

ving force Sa

= resisting force Srfor a

given shear strength and bank geometryis then calculated (Carson and Kirkby1972):

Hc= N

s (c / γ)

where c =cohesion, in pounds persquare foot, and γ = bulk unit weight ofsoil in pounds per cubic foot.

Equations are solved for a range ofbank angles using average or ambientsoil moisture conditions to produce theupper line “Ambient field conditions,unsaturated.” Critical bank height forworst-case conditions (saturated banksand rapid decline in river stage) are ob-tained by solving the equations, assum-ing that φ and the frictional componentof shear strength goes to 0.0 (Lutton

Geomorphic Processes7–61

1974) and by using a saturated bulk-unit weight. These results are repre-sented by the lower line, “saturatedconditions.”

The frequency of bank failure for thethree stability classes (unstable, at-risk,and stable) is subjective and is basedprimarily on empirical field data (Fig-ure 7.32). An unstable channel bankcan be expected to fail at least annuallyand possibly after each major storm-flow in which the channel banks aresaturated, assuming that there is at leastone major stormflow in a given year.At-risk conditions translate to a bankfailure every 2 to 5 years, again assum-ing that there is a major flow event tosaturate the banks and to erode toe ma-terial. Stable banks by definition do notfail by mass wasting processes. How-ever, channel banks on the outside ofmeander bends may experience erosionof the bank toe, leading to oversteepen-ing of the bank profile and eventuallyto bank caving episodes.

Generalizations about critical bankheights (H

c) and angles can be made

with knowledge of the variability in co-hesive strengths. Five categories ofmean cohesive strength of channelbanks are identified in Figure 7.33.Critical bank heights above the meanlow-water level and saturated condi-tions were used to construct the figurebecause bank failures typically occurduring or after the recession of peakflows. The result is a nomograph givingcritical bank heights for a range of bankangles and cohesive strengths that canbe used to estimate stable bank config-urations for worst-case conditions, suchas saturation during rapid decline inriver stage. For example, a saturatedbank at an angle of 55 degrees and a

90

Slope Angle I (degrees)

I

10080

60

40

20

10

5

275 60 45 30 15

Stab

ility

Nu

mb

er N

s =

Hc

c

= 40

º

= 0

20º25 º

30 º

35º

5º

10º

15 º

Figure 7.31: Stabilitynumber (N

S) as a

function of bankangle (i) for a failuresurface passingthrough the banktoe. Critical bankheight for worst-casecondition can becomputed.Source: Chen 1975.

Cri

tica

l Ban

k H

eig

ht

(Hc)

(fe

et)

90Bank Angle (degrees)

1678

10

20

30

40

80 70 60 50 40

unstable

at risk

ambient fieldconditions,unsaturated

saturatedconditionsstable

Figure 7.32:Example of a bankstability chart forestimating criticalbank height (H

c).

Existing bankstability can beassessed, as wellas potential stabledesign heightsand slopes.


cohesive strength of 1.75 pounds persquare inch would be unstable whenbank heights exceed about 10 feet.

Predictions of Bank Stability andChannel Width

Bank stability charts can be used todetermine the following:

■ The timing of the initiation of gener-al bank instabilities (in the case ofdegradation and increasing bankheights).

■ The timing of renewed bank stability(in the case of aggradation anddecreasing bank heights).

■ The bank height and angle neededfor a stable bank configuration undera range of moisture conditions.

Estimates of future channel wideningalso can be made using measuredchannel-width data over a period ofyears and then fitting a nonlinear func-tion to the data (Figure 7.34). Williamsand Wolman (1984) used a dimension-less hyperbolic function of the follow-ing form to estimate channel wideningdownstream from dams:

(Wi/ W

t) = j

1+ j

2(1 / t)

where:

Wi = initial channel width, in feet

Wt= channel width at t years after

W1, in feet

t = time, in years

j1

= intercept

j2

= slope of the fitted straight line on aplot of W

i / W

t versus 1/t

Wilson and Turnipseed (1994) used apower function to describe wideningafter channelization and to estimate fu-ture channel widening in the loess areaof northern Mississippi:

W = x td

where:

W = channel width, in feet

x = coefficient, determined by regres-sion, indicative of the initial channelwidth

t = time, in years

d = coefficient, determined by regres-sion, indicative of the rate of channelwidening.

Elev

atio

n B

elo

w A

ssu

med

Dat

um

(fe

et)

100

future channel widening by mass-wasting process

projection of slough-line angle

channel centerline

original floodplain surface

Distance from Centerline of Channel (feet)

0

5

10

15

20

25

30

35

40

80 60 40 20 0

Figure 7.34: Method to estimate future channel widening(10–20 years) for one side of the channel. The ultimate bankwidth can be predicted so that the future stream morphologycan be visualized.

c = 1.51 - 2.00

c = 1.01 - 1.50

c = 0.51 - 1.00

c = 0 - 0.50

Cri

tica

l Ban

k H

eig

ht

(Hc)

(fe

et)

90100

c = cohesion, in pounds per square inch

Bank Angle (degrees)

1

10

100

80 70 60 50 40 30

c = 2.01 - 3.70

Figure 7.33: Critical bank-slope configurationsfor various ranges of cohesive strengths undersaturated conditions. Specific data on thecohesive strength of bank materials can becollected to determine stable configurations.

Chemical Characteristics 7–63

Assessing water chemistry in a streamrestoration initiative can be one of theways to determine if the restoration wassuccessful. A fundamental understand-ing of the chemistry of a given system iscritical for developing appropriate datacollection and analysis methods. Al-though data collection and analysis areinterdependent, each has individualcomponents. It is also critical to have abasic understanding of the hydrologicand water quality processes of interestbefore data collection and analysisbegin. Averett and Schroder (1993) dis-cuss some fundamental concepts usedwhen determining a data collection andanalysis program.

Data Collection

Constituent Selection

Hundreds of chemical compounds canbe used to describe water quality. It istypically too expensive and too time-consuming to analyze every possiblechemical of interest in a given system.In addition to selecting a particularconstituent to sample, the analyticaltechniques used to determine the con-stituent also must be considered. An-other consideration is the chemistry ofthe constituent; for example, whetherthe chemical is typically in the dis-solved state or sorbed onto sedimentmakes a profound difference in themethods used for sampling and analy-sis, as well as the associated costs.

Often it is effective to use parametersthat integrate or serve as indicators for anumber of other variables. For instance,dissolved oxygen and temperature mea-surements integrate the net impact ofmany physical and chemical processeson a stream system, while soluble reac-tive phosphorus concentration is often

taken as a readily available indicator ofthe potential for growth of attachedalgae. Averett and Schroder (1993)discuss additional factors involved inselecting constituents to sample.

Sampling Frequency

The needed frequency of sampling de-pends on both the constituent of inter-est and management objectives. Forinstance, a management goal of reduc-ing average instream nutrient concentra-tions may require monitoring at regularintervals, whereas a goal of maintainingadequate dissolved oxygen (DO) duringsummer low flow and high temperatureperiods may require only targeted mon-itoring during critical conditions. Ingeneral, water quality constituents thatare highly variable in space or time re-quire more frequent monitoring to beadequately characterized.

In many cases, the concentration of aconstituent depends on the flow condi-tion. For example, concentrations of ahydrophobic pesticide, which sorbsstrongly to particulate matter, are likelyto be highest during scouring flows orerosion washoff events, whereas con-centrations of a dissolved chemical thatis loaded to the stream at relativelysteady rates will exhibit highest concen-trations in extremely low flows.

In fact, field sampling and water qualityanalyses are time-consuming and ex-pensive, and schedule and budget con-straints often determine the frequencyof data collection. Such constraintsmake it even more important to designdata collection efforts that maximizethe value of the information obtained.

Statistical tools often are used to helpdetermine the sampling frequency. Sta-tistical techniques, such as simple ran-

7.C Chemical Characteristics


dom sampling, stratified random sam-pling, two-stage sampling, and system-atic sampling, are described in Gilbert(1987) and Averett and Schroder(1993). Sanders et al. (1983) also de-scribe methods of determining sam-pling frequency.

Site Selection

The selection of sampling sites is thethird critical part of a sampling design.Most samples represent a point in spaceand provide direct information only onwhat is happening at that point. A keyobjective of site selection is to choose asite that gives information that is repre-sentative of conditions throughout aparticular reach of stream. Because mosthydrologic systems are very complex, itis essential to have a fundamental un-derstanding of the area of interest tomake this determination.

External inputs, such as tributaries or ir-rigation return flow, as well as output,such as ground water recharge, can dras-tically change the water quality alongthe length of a stream. It is because ofthese processes that the hydrologic sys-tem must be understood to interpretthe data from a particular site. For ex-ample, downstream from a significantlateral source of a load, the dissolvedconstituent(s) might be distributed uni-formly in the stream channel. Particu-late matter, however, typically isstratified. Therefore, the distribution ofa constituent sorbed onto particulatematter is not evenly distributed. Averettand Schroder (1993) discuss differentapproaches to selecting sites to sampleboth surface water and ground water.Sanders et al. (1983) and Stednick(1991) also discuss site selection.

Finally, practical considerations are animportant part of sample collection.Sites first must be accessible, preferablyunder a full range of potential flow and

weather conditions. For this reason,sampling is often conducted at bridgecrossings, taking into consideration thedegree to which artificial channels atbridge crossings may influence sampleresults. Finally, where constituent loadsand concentrations are of interest, it isimportant to align water quality samplesites with locations at which flow canbe accurately gauged.

Sampling Techniques

This section provides a brief overview ofwater quality sampling and data collec-tion techniques for stream restorationefforts. Many important issues can betreated only cursorily within the contextof this document, but a number of ref-erences are available to provide thereader with more detailed guidance.

Key documents describing methods ofwater sample collection for chemicalanalysis are the U.S. Geological Survey(USGS) protocol for collecting and pro-cessing surface water samples for deter-mining inorganic constituents infiltered water (Horwitz et al. 1994), thefield guide for collecting and processingstream water samples for the NationalWater Quality Assessment program(Shelton 1994), and the field guide forcollecting and processing samples ofstreambed sediment for analyzing traceelements and organic contaminants forthe National Water Quality Assessmentprogram (Shelton and Capel 1994). Astandard reference document describingmethods of sediment collection is theUSGS Techniques for Water-Resource In-vestigations, Field Methods for Measure-ment of Fluvial Sediment (Guy andNorman 1982). The USGS is preparinga national field manual that describestechniques for collecting and processingwater quality samples (Franceska Wilde,personal communication, 1997).


Sampling Protocols forWater and Sediment

Stream restoration monitoring may in-volve sampling both water and sedi-ment quality. These samples may becollected by hand (manual samples), byusing an automated sampler (automaticsamples), as individual point-in-timesamples (grab or discrete samples), orcombined with other samples (compos-ite samples). Samples collected andmixed in relation to the measured vol-ume within or flow through a systemare commonly termed volume- or flow-weighted composite samples, whereasequal-volume samples collected at regu-lar vertical intervals through a portionor all of the water column may bemixed to provide a water column com-posite sample.

Manual Sampling and Grab Sampling

Samples collected by hand using vari-ous types of containers or devices tocollect water or sediment from a receiv-ing water or discharge often are termedgrab samples. These samples can re-quire little equipment and allow record-ing miscellaneous additional fieldobservations during each sampling visit.

Manual sampling has several advan-tages. These approaches are generallyuncomplicated and often inexpensive(particularly when labor is alreadyavailable). Manual sampling is requiredfor sampling some pollutants. For ex-ample, according to Standard Methods(APHA 1995), oil and grease, volatilecompounds, and bacteria must be ana-lyzed from samples collected usingmanual methods. (Oil, grease, and bac-teria can adhere to hoses and jars usedin automated sampling equipment,causing inaccurate results; volatile com-pounds can vaporize during automatedsampling procedures or can be lostfrom poorly sealed sample containers;and bacteria populations can grow and

community compositions change dur-ing sample storage.)

Disadvantages of grab sampling includethe potential for personnel to be avail-able around the clock to sample duringstorms and the potential for personnelto be exposed to hazardous conditionsduring sampling. Long-term samplingprograms involving many sampling lo-cations can be expensive in terms oflabor costs.

Grab sampling is often used to collectdiscrete samples that are not combinedwith other samples. Grab samples canalso be used to collect volume- or flow-weighted composite samples, whereseveral discrete samples are combinedby proportion to measured volume orflow rates; however, this type of sam-pling is often more easily accomplishedusing automated samplers and flow me-ters. Several examples of manual meth-ods for flow weighting are presented inUSEPA (1992a). Grab sampling alsomay be used to composite vertical watercolumn or aerial composite samples ofwater or sediment from various kinds ofwater bodies.

Automatic Sampling

Automated samplers have been im-proved greatly in the last 10 years andnow have features that are useful formany sampling purposes. Generally,such sampling devices require largerinitial capital investments or the pay-ment of rental fees, but they can reduceoverall labor costs (especially for long-running sampling programs) and in-crease the reliability of flow-weightedcompositing.

Some automatic samplers include anupper part consisting of a microproces-sor-based controller, a pump assembly,and a filling mechanism, and a lowerpart containing a set of glass or plasticsample containers and a well that canbe filled with ice to cool the collected


samples. More expensive automaticsamplers can include refrigerationequipment in place of the ice well; suchdevices, however, require a 120-voltpower supply instead of a battery. Also,many automatic samplers can acceptinput signals from a flowmeter to acti-vate the sampler and to initiate a flow-weighting compositing program. Somesamplers can accept input from a raingauge to activate a sampling program.

Most automatic samplers allow collect-ing multiple discrete samples or singleor multiple composited samples. Also,samples can be split between samplebottles or can be composited into a sin-gle bottle. Samples can be collected ona predetermined time basis or in pro-portion to flow measurement signalssent to the sampler.

In spite of the obvious advantages ofautomated samplers, they have somedisadvantages and limitations. Somepollutants cannot be sampled by auto-mated equipment unless only qualita-tive results are desired. Although thecleaning sequence provided by mostsuch samplers provide reasonably sepa-rate samples, there is some cross-conta-mination of the samples since waterdroplets usually remain in the tubing.Debris in the sampled receiving watercan block the sampling line and pre-vent sample collection. If the samplingline is located in the vicinity of aflowmeter, debris caught on the sam-pling line can also lead to erroneousflow measurements.

While automatic samplers can reducemanpower needs during storm andrunoff events, these devices must bechecked for accuracy during theseevents and must be regularly tested andserviced. If no field checks are madeduring a storm event, data for the entireevent may be lost. Thus, automatic sam-plers do not eliminate the need for field

personnel, but they can reduce theseneeds and can produce flow-weightedcomposite samples that might be te-dious or impossible using manualmethods.

Discrete versus Composite Sampling

Flow rates, physical conditions, andchemical constituents in surface watersoften vary continuously and simultane-ously. This presents a difficulty whendetermining water volumes, pollutantconcentrations, and masses of pollu-tants or their loads in the waste dis-charge flows and in receiving waters.Using automatic or continuouslyrecording flowmeters allows obtainingreasonable and continuous flow ratemeasurements for these waters. Pollu-tant loads can then be computed bymultiplying these flow volumes over theperiod of concern by the average pollu-tant concentration determined from thediscrete or flow-composited samples.When manual (instantaneous) flowmeasurements are used, actual volumeflows over time can be estimated onlyfor loading calculations, adding addi-tional uncertainty to loading estimates.

Analyzing constituents of concern in asingle grab sample collection providesthe minimum information at the mini-mum cost. Such an approach, however,could be appropriate where conditionsare relatively stable; for example, duringperiods without rainfall or other poten-tial causes of significant runoff andwhen the stream is well-mixed. Mostoften, the usual method is to collect arandom or regular series of grab sam-ples at predefined intervals duringstorm or runoff events.

When samples are collected oftenenough, such that concentrationchanges between samples are mini-mized, a clear pattern or time series forthe pollutant’s concentration dynamicscan be obtained. When sampling inter-


vals are spaced too far apart in relationto changes in the pollutant concentra-tion, less clear understanding of theserelationships is obtained. Mixing sam-ples from adjacent sampling events orregions (compositing) requires fewersamples to be analyzed; for some as-sessments, this is a reasonable ap-proach. Sample compositing provides asavings, especially related to costs forwater quality analyses, but it also resultsin loss of information. For example, in-formation on maximum and minimumconcentrations during a runoff event isusually lost. But compositing manysamples collected through multiple pe-riods during the events can help ensurethat the samples analyzed do not in-clude only extreme conditions that arenot entirely representative of the event.

Even though analytical results fromcomposited samples rarely equal aver-age conditions for the event, they canstill be used, when a sufficient distribu-tion of samples is included, to providereasonably representative conditions forcomputing loading estimates. In someanalyses, however, considerable errorscan be made when using analytical re-sults from composited samples in com-pleting loading analyses. For example,when maximum pollutant concentra-tions accompany the maximum flowrates, yet concentrations in high andlow flows are treated equally, true load-ings can be underestimated.

Consequently, when relationships be-tween flow and pollutant concentra-tions are unknown, it is oftenpreferable initially to include in themonitoring plan at least three discreteor multiple composite sample collec-tions: during the initial period of in-creasing flow, during the period of thepeak or plateau flow, and during the pe-riod of declining flow.

The most useful method for samplecompositing is to combine samples inrelation to the flow volume occurringduring study period intervals. There aretwo variations for accomplishing flow-weighted compositing:

1. Collect samples at equal time inter-vals at a volume proportional to theflow rate (e.g., collect 100 mL of sam-ple for every 100 gallons of flow thatpassed during a 10-minute interval)or

2. Collect equal-volume samples atvarying times proportional to theflow (e.g., collect a 100-mL samplefor each 100 gallons of flow, irrespec-tive of time).

The second method is preferable for es-timating load accompanying wetweather flows, since it results in sam-ples being collected most often whenthe flow rate is highest.

Another compositing method is time-composited sampling, where equalsample volumes are collected at equallyspaced time intervals (e.g., collect 100mL of sample every 10 minutes duringthe monitored event). This approachprovides information on the averageconditions at the sampling point duringthe sampling period. It should be used,for example, to determine the averagetoxic concentrations to which residentaquatic biota are exposed during themonitored event.

Field Analyses of Water QualitySamples

Concentrations of various water qualityparameters may be monitored both inthe field and in samples submitted to alaboratory (Figure 7.35). Some parame-ters, such as water temperature, must beobtained in the field. Parameters suchas concentrations of specific syntheticorganic chemicals require laboratoryanalysis. Other parameters, such as nu-


trient concentrations, can be measuredby both field and laboratory analyticalmethods. For chemical constituents,field measurements generally should beconsidered as qualitative screening val-ues since rigorous quality control is notpossible. In addition, samples collectedfor compliance with Clean Water Act re-quirements must be analyzed by a labo-ratory certified by the appropriateauthority, either the state or the USEPA.The laboratories must use analytic tech-niques listed in the Code of Federal Regu-lations (CFR), Title 40, Part 136,“Guidelines Establishing Test Proce-dures for Analysis of Pollutants Underthe Clean Water Act.”

The balance of this subsection notesspecial considerations regarding thoseparameters typically sampled and ana-

lyzed in the field, including pH, tem-perature, and dissolved oxygen (DO).

pH

Levels of pH can change rapidly in sam-ples after collection. Consequently, pHoften is measured in the field using ahand-held pH electrode and meter.Electrodes are easily damaged and con-taminated and must be calibrated witha standard solution before each use.During calibrations and when site mea-surements are conducted, field instru-ments should be at thermal equilibriumwith the solutions being measured.

Temperature

Because water temperature changesrapidly after collection, it must be mea-sured either in the field (using in situprobes) or immediately after collectinga grab sample. EPA Method 170.1 de-scribes procedures for thermometric de-termination of water temperature.Smaller streams often experience widediurnal variations in temperature, aswell as pH and DO. Many streams alsoexperience vertical and longitudinalvariability in temperature from shadingand flow velocity. Because of the effectof temperature on other water qualityfactors, such as dissolved oxygen con-centration, temperatures always shouldbe recorded when other field measure-ments are made.

Dissolved Oxygen

When multiple DO readings are re-quired, a DO electrode and meter (EPAmethod 360.1) are typically used. Toobtain accurate measurements, the Win-kler titration method should be used tocalibrate the meter before and after eachday’s use. Often it is valuable to recheckthe calibration during days of intensiveuse, particularly when the measure-ments are of critical importance.

Oxygen electrodes are fragile and sub-ject to contamination, and they need

Figure 7.35: Field sampling. Sampling can alsobe automated.


frequent maintenance. Membranes cov-ering these probes must be replacedwhen bubbles form under the mem-brane, and the electrode should be keptfull of fresh electrolyte solution. If themeter has temperature and salinitycompensation controls, they should beused carefully, according to the manu-facturer’s instructions.

Water Quality SamplePreparation and Handling forLaboratory Analysis

Sample collection, preparation, preser-vation, and storage guidelines are de-signed to minimize altering sampleconstituents. Containers must be madeof materials that will not interact withpollutants in the sample, and theyshould be cleaned in such a way thatneither the container nor the cleaningagents interfere with sample analysis.Sometimes, sample constituents mustbe preserved before they degrade ortransform prior to analysis. Also, speci-fied holding times for the sample mustnot be exceeded. Standard proceduresfor collecting, preserving, and storingsamples are presented in APHA (1995)and at 40 CFR Part 136. Useful materialalso is contained in the USEPA NPDESStorm Water Sampling Guidance Docu-ment (1992a).

Most commercial laboratories provideproperly cleaned sampling containerswith appropriate preservatives. The lab-oratories also usually indicate the maxi-mum allowed holding periods for eachanalysis. Acceptable procedures forcleaning sample bottles, preservingtheir contents, and analyzing for appro-priate chemicals are detailed in variousmethods manuals, including APHA(1995) and USEPA (1979a). Water sam-plers, sampling hoses, and sample stor-age bottles always should be made ofmaterials compatible with the goals ofthe study. For example, when heavy

metals are the concern, bottles shouldnot have metal components that cancontaminate the collected water sam-ples. Similarly, when organic contami-nants are the concern, bottles and capsshould be made of materials not likelyto leach into the sample.

Sample Preservation, Handling,and Storage

Sample preservation techniques andmaximum holding times are presentedin APHA (1995) and 40 CFR Part 136.Cooling samples to a temperature of4 degrees Celsius (º C) is required formost water quality variables. To accom-plish this, samples are usually placed ina cooler containing ice or an ice substi-tute. Many automated samplers have awell next to the sample bottles to holdeither ice or ice substitutes. Some moreexpensive automated samplers have re-frigeration equipment requiring asource of electricity. Other preservationtechniques include pH adjustment andchemical fixation. When needed, pHadjustments are usually made usingstrong acids and bases, and extremecare should be exercised when handingthese substances.

Bacterial analysis may be warranted,particularly where there are concerns re-garding inputs of sewage and otherwastes or fecal contamination. Bacterialsamples have a short holding time andare not collected by automated sampler.Similarly, volatile compounds must becollected by grab sample, since they arelost through volatilization in automaticsampling equipment.

Sample Labeling

Samples should be labeled with water-proof labels. Enough informationshould be recorded to ensure that eachsample label is unique. The informationrecorded on sample container labelsalso should be recorded in a samplingnotebook kept by field personnel. The


label typically includes the followinginformation:

■ Name of project.

■ Location of monitoring.

■ Specific sample location.

■ Date and time of sample collection.

■ Name or initials of sampler.

■ Analysis to be performed.

■ Sample ID number.

■ Preservative used.

■ Type of sample (grab, composite).

Sample Packaging and Shipping

It is sometimes necessary to ship sam-ples to the laboratory. Holding timesshould be checked before shipment toensure that they will not be exceeded.Although wastewater samples are notusually considered hazardous, somesamples, such as those with extremepH, require special procedures. If thesample is shipped through a commoncarrier or the U.S. Postal Service, it mustcomply with Department of Transporta-tion Hazardous Material Regulations(49 CFR Parts 171-177). Air shipmentof samples defined as hazardous maybe covered by the requirements of theInternational Air Transport Association.

Samples should be sealed in leakproofbags and padded against breakage.Many samples must be packed with anice substitute to maintain a temperatureof 4 degrees C during shipment. Plasticor metal recreational coolers make idealshipping containers because they pro-tect and insulate the samples. Accompa-nying paperwork, such as thechain-of-custody documentation,should be sealed in a waterproof bag inthe shipping container.

Chain of Custody

Chain-of-custody forms document eachchange in possession of a sample, start-

ing at its collection and ending when itis analyzed. At each transfer of posses-sion, both the relinquisher and the re-ceiver of the samples are required tosign and date the form. The form andthe procedure document possession ofthe samples and help prevent tamper-ing. The container holding samples alsocan be sealed with a signed tape or sealto help ensure that samples are notcompromised.

Copies of the chain-of-custody formshould be retained by the sampler andby the laboratory. Contract laboratoriesoften supply chain-of-custody formswith sample containers. The form isalso useful for documenting whichanalyses will be performed on the sam-ples. These forms typically contain thefollowing information:

■ Name of project and sampling loca-tions.

■ Date and time that each sample iscollected.

■ Names of sampling personnel.

■ Sample identification names andnumbers.

■ Types of sample containers.

■ Analyses performed on each sample.

■ Additional comments on each sample.

■ Names of all those transporting thesamples.

Collecting and HandlingSediment Quality Samples

Sediments are sinks for a wide varietyof materials. Nonpoint source dis-charges typically include large quanti-ties of suspended material that settleout in sections of receiving waters hav-ing low water velocities. Nutrients,metals, and organic compounds canbind to suspended solids and settle tothe bottom of a water body when flow


velocity is insufficient to keep them insuspension. Contaminants bound tosediments may remain separated fromthe water column, or they may be resus-pended in the water column.

Flood scouring, bioturbation (mixingby biological organisms), desorption,and biological uptake all promote therelease of adsorbed pollutants. Organ-isms that live and feed in sediment areespecially vulnerable to contaminantsin sediments. Having entered the foodchain, contaminants can pass to feedersat higher food (trophic) levels and canaccumulate or concentrate in these or-ganisms. Humans can ingest these con-taminants by eating fish.

Sediment deposition also can physicallyalter benthic (bottom) habitats and af-fect habitat and reproductive potentialsfor many fish and invertebrates. Sedi-ment sampling should allow all theseimpact potentials to be assessed.

Collection Techniques

Sediment samples are collected usinghand- or winch-operated dredges. Al-though a wide variety of dredges areavailable, most operate in the followingsimilar fashion:

1. The device is lowered or pushedthrough the water column by handor winch.

2. The device is released to allow clo-sure, either by the attached line or bya weighted messenger that isdropped down the line.

3. The scoops or jaws of the deviceclose either by weight or springaction.

4. The device is retrieved to the surface.

Ideally, the device disturbs the bottomas little as possible and closes fully sothat fine particles are not lost. Com-mon benthic sampling devices includethe Ponar, Eckman, Peterson, Orange-

peel, and Van Veen dredges. When in-formation is needed about how chemi-cal depositions and accumulations havevaried through time, sediment cores canbe collected with a core sampling de-vice. Very low density or very coarsesediments can be sampled by freezecoring. A thorough description of sedi-ment samplers is included in Klemm etal. (1990).

Sediment sampling techniques are use-ful for two types of investigations re-lated to stream assessments:(1) chemical analysis of sediments and(2) investigation of benthic macroinver-tebrate communities. In either type ofinvestigation, sediments from referencestations should be sampled so that theycan be compared with sediments in theaffected receiving waters. Sedimentsused for chemical analyses should beremoved from the dredge or core sam-ples by scraping back the surface layersof the collected sediment and extractingsediments from the central mass of thecollected sample. This helps to avoidpossible contamination of the sampleby the sample device. Sediment samplesfor toxicological and chemical examina-tion should be collected followingmethod E 1391 detailed in ASTM(1991). Sediments for benthic popula-tion analyses may be returned in totalfor cleaning and analysis or may receivea preliminary cleaning in the field usinga No. 30 sieve.

Sediment Analyses

There are a variety of sediment analysistechniques, each designed with inherentassumptions about the behavior of sed-iments and sediment-bound contami-nants. An overview of developingtechniques is presented in Adams et al.(1992). EPA has evaluated 11 of themethods available for assessing sedi-ment quality (USEPA 1989b). Some ofthe techniques may help to demonstrate


attainment of narrative requirements ofsome water quality standards. Two ofthese common analyses are introducedbriefly in the following paragraphs.

Bulk sediment analyses analyze thetotal concentration of contaminantsthat are either bound to sediments orpresent in pore water. Results are re-ported in milligrams or micrograms perkilogram of sediment material. Thistype of testing often serves as a screen-ing analysis to classify dredged material.Results of bulk testing tend to overesti-mate the mass of contaminants thatwill be available for release or for bio-logical uptake because a portion of thecontaminants are not biologically avail-able or likely to dissolve.

Elutriate testing estimates the amount ofcontaminants likely to be released fromsediments when mixed with water. Inan elutriate test, sediment is mixed withwater and then agitated. The standardelutriate test for dredge material mixesfour parts water from the receivingwater body with one part sediment(USEPA 1990). After vigorous mixing,the sample is allowed to settle beforethe supernatant is filtered and analyzedfor contaminants. This test was de-signed to estimate the amount of mate-rial likely to enter the dissolved phaseduring dredging; however, it is also use-ful as a screening test for determiningwhether further testing should be per-formed and as a tool for comparingsediments upstream and downstream ofpotential pollutant sources.

Data Management

All monitoring data should be orga-nized and stored in a readily accessibleform. The potentially voluminous anddiverse nature of the data, and the vari-ety of individuals who can be involvedin collecting, recording, and enteringdata, can easily lead to the loss of data

or the recording of erroneous data. Lostor erroneous data can severely damagethe quality of monitoring programs. Asound and efficient data managementprogram for a monitoring programshould focus on preventing such prob-lems. This requires that data be man-aged directly and separately from theactivities that use them.

Data management systems include tech-nical and managerial components. Thetechnical components involve selectingappropriate computer equipment andsoftware and designing the database, in-cluding data definition, data standard-ization, and a data dictionary. Themanagerial components include dataentry, data validation and verification,data access, and methods for users toaccess the data.

To ensure the integrity of the database,it is imperative that data quality be con-trolled from the point of collection tothe time the information is entered intothe database. Field and laboratory per-sonnel must carefully enter data intoproper spaces on data sheets and avoidtransposing numbers. To avoid tran-scription errors, entries into a databaseshould be made from original datasheets or photocopies. As a preliminaryscreen for data quality, the database de-sign should include automatic parame-ter range checking. Values outside thedefined ranges should be flagged by theprogram and immediately corrected orincluded in a follow-up review of theentered data. For some parameters, itmight be appropriate to include auto-matic checks to disallow duplicate val-ues. Preliminary database files shouldbe printed and verified against the orig-inal data to identify errors.

Additional data validation can includeexpert review of the verified data toidentify possible suspicious values.Sometimes, consultation with the indi-


viduals responsible for collecting or en-tering original data is required to resolveproblems. After all data are verified andvalidated, they can be merged into themonitoring program’s master database.To prevent loss of data from computerfailure, at least one set of duplicate(backup) database files should bemaintained at a location other thanwhere the master database is kept.

Quality Assurance and QualityControl (QA/QC)

Quality assurance (QA) is the manage-ment process to ensure the quality ofdata. In the case of monitoring projects,it is managing environmental data col-lection to ensure the collection of high-quality data. QA focuses on systems,policies, procedures, program structures,and delegation of responsibility thatwill result in high-quality data. Qualitycontrol (QC) is a group of specific pro-cedures designed to meet defined dataquality objectives. For example, equip-ment calibration and split samples areQC procedures. QA/QC procedures areessential to ensure that data collected inenvironmental monitoring programs areuseful and reliable.

The following are specific QA plans re-quired of environmental monitoringprojects that receive funding from EPA:

■ State and local governments receivingEPA assistance for environmentalmonitoring projects must complete aquality assurance program planacceptable to the award official.Guidance for producing the programplan is contained in USEPA (1983d).

■ Environmental monitoring projectsthat receive EPA funding must file aquality assurance project plan, orQAPP, (40 CFR 30.503), the purposeof which is to ensure quality of a spe-cific project. The QAPP describesquality assurance practices designed

to produce data of quality sufficientto meet project objectives. Guidancefor producing the QAPP (formerlytermed the QAPjP) is contained inUSEPA (1983e). The plan mustaddress the following items:

■ Title of project and names ofprincipal investigators.

■ Table of contents.

■ Project description.

■ Project organization and QA/QCresponsibility.

■ Quality assurance objectives andcriteria for determining precision,accuracy, completeness, representa-tiveness, and comparability of data.

■ Sampling procedures.

■ Sample custody.

■ Calibration procedures.

■ Analytical procedures.

■ Data reduction, validation, andreporting.

■ Internal quality control checks.

■ Performance and system audits.

■ Preventive maintenance proce-dures.

■ Specific routine procedures toassess data precision, accuracy,representativeness, and compara-bility.

■ Corrective action.

■ Quality assurance reports.

Sample and Analytical Quality Control

The following quality control tech-niques are useful in assessing samplingand analytic performance (see alsoUSEPA 1979b, Horwitz et al. 1994):

■ Duplicate samples are independentsamples collected in such a mannerthat they are equally representative ofthe contaminants of interest. Dupli-


cate samples, when analyzed by thesame laboratory, provide precisioninformation for the entire measure-ment system, including samplecollection, homogeneity, handling,shipping, storage, preparation, andanalysis.

■ Split samples have been divided intotwo or more portions at some pointin the measurement process. Splitsamples that are divided in the fieldyield results relating precision tohandling, shipping, storage, prepara-tion, and analysis. The split samplesmay be sent to different laboratoriesand subjected to the same measure-ment process to assess interlaborato-ry variation. Split samples serve anoversight function in assessing theanalytical portion of the measure-ment system, whereas error due tosampling technique may be estimat-ed by analyzing duplicate versions ofthe same sample.

■ Spiked samples are those to which aknown quantity of a substance isadded. The results of spiking a sam-ple in the field are usually expressedas percent recovery of the addedmaterial. Spiked samples provide acheck of the accuracy of laboratoryand analytic procedures.

Sampling accuracy can be estimated byevaluating the results obtained fromblanks. The most suitable types ofblanks for this appraisal are equipment,field, and trip blanks.

■ Equipment blanks are samples obtainedby running analyte-free water throughsample collection equipment, such asa bailer, pump, or auger, after decon-tamination procedures are complet-ed. These samples are used to deter-mine whether variation is introducedby sampling equipment.

■ Field blanks are made by transferringdeionized water to a sample contain-

er at the sampling site. Field blankstest for contamination in the deion-ized water and contamination intro-duced through the sampling proce-dure. They differ from trip blanks,which remain unopened in the field.

■ Trip blanks test for cross-contamina-tion during transit of volatile con-stituents, such as many syntheticorganic compounds and mercury. Foreach shipment of sample containerssent to the analytical laboratory, onecontainer is filled with analyte-freewater at the laboratory and is sealed.The blanks are transported to the sitewith the balance of the sample con-tainers and remain unopened.Otherwise, they are handled in thesame manner as the other samples.The trip blanks are returned to thelaboratory with the samples and areanalyzed for the volatile constituents.

Field Quality Assurance

Errors or a lack of standardization infield procedures can significantly de-crease the reliability of environmentalmonitoring data. If required, a qualityassurance project plan should be fol-lowed for field measurement proce-dures and equipment. If the QAPP isnot formally required, a plan includingsimilar material should be developed toensure the quality of data collected.Standard operating procedures shouldbe followed when available and shouldbe developed when not.

It is important that quality proceduresbe followed and regularly examined.For example, field meters can provideerroneous values if they are not regu-larly calibrated and maintained.Reagent solutions and probe electrolytesolutions have expiration periods andshould be refreshed periodically.

Biological Characteristics 7–75

Nearly all analytical procedures for as-sessing the condition of biological re-sources can be used in stream corridorrestoration. Such procedures differ,however, in their scale and focus and inthe assumptions, knowledge, and effortrequired to apply them. These proce-dures can be grouped into two broadclasses — synthetic measures of systemcondition and analyses based on howwell the system satisfies the life historyrequirements of target species or speciesgroups.

The most important difference betweenthese classes is the logic of how they areapplied in managing or restoring astream corridor system. This chapter fo-cuses on metrics of biological condi-tions and does not describe, forexample, actual field methods forcounting organisms.

Synthetic Measures of SystemCondition

Synthetic measures of system conditionsummarize some aspect of the struc-tural or functional status of a system ata particular point in time. Completemeasurement of the state of a streamcorridor system, or even a completecensus of all of the species present, isnot feasible. Thus, good indicators ofsystem condition are efficient in thesense that they summarize the health ofthe overall system without having tomeasure everything about the system.

Use of indicators of system condition inmanagement or restoration dependscompletely on comparison to values ofthe indicator observed in other systemsor at other times. Thus, the currentvalue of an indicator for a degradedstream corridor can be compared to apreviously measured preimpact valuefor the corridor, a desired future value

for the corridor, a value observed at an“unimpacted” reference site, a range ofvalues observed in other systems, or anormative value for that class of streamcorridors in a stream classification sys-tem. However, the indicator itself andthe analysis that establishes the value of the indicator provide no direct infor-mation about what has caused the sys-tem to have a particular value for theindicator.

Deciding what to change in the systemto improve the value of the indicatordepends on a temporal analysis inwhich observed changes in the indica-tor in one system are correlated withvarious management actions or on aspatial analysis in which values of theindicator in different systems are corre-lated with different values of likely con-trolling variables. In both cases, nomore than a general empirical correla-tion between specific causal factors andthe indicator variable is attempted.Thus, management or restoration basedon synthetic measures of system condi-tion relies heavily on iterative monitor-ing of the indicator variable and trialand error, or adaptive management, ap-proaches. For example, an index ofspecies composition based on the pres-ence or absence of a set of sensitivespecies might be generally correlatedwith water quality, but the index itselfprovides no information on how waterquality should be improved. However,the success of management actions inimproving water quality could betracked and evaluated through iterativemeasurement of the index.

Synthetic measures of system conditionvary along a number of important di-mensions that determine their applica-bility. In certain situations, singlespecies might be good indicators of

7.D Biological Characteristics


some aspect of a stream corridor sys-tem; in others, community metrics,such as diversity, might be more suit-able. Some indicators incorporate phys-ical variables, and others do not.Measurements of processes and rates,such as primary productivity and chan-nel meandering rates, are incorporatedinto some and not into others. Each ofthese dimensions must be evaluated rel-ative to the objectives of the restorationeffort to determine which, if any, indi-cator is most appropriate.

Indicator Species

Landres et al. (1988) define an indicatorspecies as an organism whose character-istics (e.g., presence or absence, popula-tion density, dispersion, reproductivesuccess) are used as an index of attrib-utes too difficult, inconvenient, or ex-pensive to measure for other species orenvironmental conditions of interest.Ecologists and management agencieshave used aquatic and terrestrial indica-tor species for many years as assessmenttools, the late 1970s and early 1980sbeing a peak interest period. During thattime, Habitat Evaluation Procedures(HEP) were developed by the U.S. Fishand Wildlife Service, and the U.S. ForestService’s use of management indicatorspecies was mandated by law with pas-sage of the National Forest ManagementAct in 1976. Since that time, numerousauthors have expressed concern aboutthe ability of indicator species to meetthe expectations expressed in the abovedefinition. Most notably, Landres et al.(1988) critically evaluated the use ofvertebrate species as ecological indica-tors and suggested that rigorous justifi-cation and evaluation are needed beforethe concept is used. The discussion ofindicator species below is largely basedon their paper.

The Good and Bad of Indicator Species

Indicator species have been used to pre-dict environmental contamination,population trends, and habitat quality;however, their use in evaluating waterquality is not covered in this section.The assumptions implicit in using indi-cators are that if the habitat is suitablefor the indicator it is also suitable forother species (usually in a similar eco-logical guild) and that wildlife popula-tions reflect habitat conditions.However, because each species hasunique life requisites, the relationshipbetween the indicator and its guild may

This is another assessment tool that provides a basiclevel of stream health evaluation. It is intended to be thefirst level in a four-part hierarchy of assessment protocolsthat facilitate planning stream restorations. Scores areassigned by the planners for the following:

■ Channel condition■ Hydrologic alteration■ Riparian zone width■ Bank stability■ Canopy cover■ Water appearance■ Nutrient enrichment■ Manure presence■ Salinity■ Barriers to fish movement■ Instream fish cover■ Pools■ Riffle quality■ Invertebrate habitat■ Macroinvertebrates observed

The planning assessment concludes with narratives ofthe suspected causes of observed problems, as well asrecommendations or further steps in the planningprocess (USDA-NRCS 1998).


not be completely reliable, although theliterature is inconsistent in this regard(see Riparian Response Guilds subsec-tion below). It is also difficult to in-clude all the factors that might limit apopulation when selecting a group ofspecies that an indicator is expected torepresent. For example, similarities inbreeding habitat between the indicatorand its associates might appear togroup species when in fact differencesin predation rates, disease, or winterhabitat actually limit populations.

Some management agencies use verte-brate indicators to track changes inhabitat condition or to assess the influ-ence of habitat alteration on selectedspecies. Habitat suitability indices andother habitat models are often used forthis purpose, though the metric chosento measure a species’ response to itshabitat can influence the outcome ofthe investigation. As Van Horne (1983)pointed out, density and other abun-dance metrics may be misleading indi-cators of habitat quality. Use ofdiversity and other indices to estimatehabitat quality also creates problemswhen the variation in measures yieldsan average value for an index thatmight not represent either extreme.

Selecting Indicators

Landres et al. (1988) suggest that if thedecision is made to use indicators, thenseveral factors are important to considerin the selection process:

■ Sensitivity of the species to the envi-ronmental attribute being evaluated.When possible, data that suggest acause-and-effect relationship are pre-ferred to correlates (to ensure theindicator reflects the variable of inter-est and not a correlate).

■ Indicator accurately and preciselyresponds to the measured effect.High variation statistically limits theability to detect effects. Generalist

species do not reflect change as wellas more sensitive endemics. However,because specialists usually have lowerpopulations, they might not be thebest for cost-effective sampling.When the goal of monitoring is toevaluate on-site conditions, usingindicators that occur only within thesite makes sense. However, althoughpermanent residents may betterreflect local conditions, the goal ofmany riparian restoration efforts is toprovide habitat for neotropicalmigratory birds. In this case, resi-dents such as cardinals or woodpeck-ers might not serve as good indica-tors for migrating warblers.

■ Size of the species home range. Ifpossible, the home range should belarger than that of other species inthe evaluation area. Managementagencies often are forced to use high-profile game or threatened andendangered species as indicators.Game species are often poor indica-tors simply because their populationsare highly influenced by huntingmortality, which can mask environ-mental effects. Species with low pop-ulations or restrictions on samplingmethods, such as threatened andendangered species, are also poorindicators because they are difficultto sample adequately, often due tobudget constraints. For example,Verner (1986) found that costs todetect a 10 percent change in a ran-domly sampled population of pileat-ed woodpeckers would exceed a mil-lion dollars per year.

■ Response of an indicator species toan environmental stressor cannot beexpected to be consistent across vary-ing geographic locations or habitatswithout corroborative research.


Riparian Response Guilds

Vertebrate response guilds as indicatorsof restoration success in riparian ecosys-tems may be a valuable monitoring toolbut should be used with the same cau-tions presented above. Croonquist andBrooks (1991) evaluated the effects ofanthropogenic disturbances on smallmammals and birds along Pennsylvaniawaterways. They evaluated species infive different response guilds, includingwetland dependency, trophic level,species status (endangered, recreational,native, exotic), habitat specificity, andseasonality (birds).

They found that community coefficientindices were better indicators thanspecies richness. The habitat specificityand seasonality response guilds for birdswere best able to distinguish thosespecies sensitive to disturbance fromthose which were not affected or werebenefited. Neotropical migrants andspecies with specific habitat require-ments were the best predictors of distur-bance. Edge and exotic species weregreater in abundance in the disturbedhabitats and might serve as good indica-tors there. Seasonality analysis showedmigrant breeders were more common inundisturbed areas, which, as suggestedby Verner (1984), indicates the ability ofguild analysis to distinguish local im-pacts. Mammalian response guilds didnot exhibit any significant sensitivity todisturbance and were considered unsuit-able as indicators.

In contrast, Mannan et al. (1984)found that in only one of the five avianguilds tested was the density of birdsconsistent across managed and undis-turbed forests. In other words, popula-tion response to restoration might notbe consistent across different indicatorguilds. Also, periodically monitoringrestoration initiatives is necessary to

document when, during the recoverystage, the more sensitive species out-compete generalists.

Aquatic Invertebrates

Aquatic invertebrates have been used asindicators of stream and riparian healthfor many years. Perhaps more thanother taxa, they are closely tied to bothaquatic and riparian habitat. Their lifecycles usually include periods in andout of the water, with ties to riparianvegetation for feeding, pupation, emer-gence, mating, and egg laying (Erman1991).

It is often important to look at the en-tire assemblage of aquatic invertebratesas an indicator group. Impacts to astream often decrease diversity butmight increase the abundance of somespecies, with the size of the first speciesto be affected often larger (Wallace andGurtz 1986). In summary, a good indi-cator species should be low on the foodchain to respond quickly, should have anarrow tolerance to change, and shouldbe a native species (Erman 1991).

Diversity and Related Indices

Biological diversity refers to the numberof species in an area or region and in-cludes a measure of the variety ofspecies in a community that takes intoaccount the relative abundance of eachspecies (Ricklefs 1990). When measur-ing diversity, it is important to clearlydefine the biological objectives, statingexactly what attributes of the system areof concern and why (Schroeder andKeller 1990). Different measures of di-versity can be applied at various levelsof complexity, to different taxonomicgroups, and at distinct spatial scales.Several factors should be considered in using diversity as a measure of sys-tem condition for stream corridorrestoration.


Levels of Complexity

Diversity can be measured at severallevels of complexity— genetic, popula-tion/species, community/ecosystem,and landscape (Noss 1994). There is nosingle correct level of complexity to usebecause different scientific or manage-ment issues are focused on differentlevels (Meffe et al. 1994). The level ofcomplexity chosen for a specific streamcorridor restoration initiative should bedetermined based on careful considera-tion of the biological objectives of theproject.

Subsets of Concern

Overall diversity within any given levelof complexity may be of less concernthan diversity of a particular subset ofspecies or habitats. Measures of overalldiversity include all of the elements ofconcern and do not provide informa-tion about the occurrence of specific el-ements. For example, measures ofoverall species diversity do not provideinformation about the presence of indi-vidual species or species groups of man-agement concern.

Any important subsets of diversityshould be described in the process ofsetting biological objectives. At thecommunity level, subsets of species ofinterest might include native, endemic,locally rare or threatened, specificguilds (e.g., cavity users), or taxonomicgroups (e.g., amphibians, breedingbirds, macroinvertebrates). At the terres-trial landscape level, subsets of diversitycould include forest types or seral stages(Noss 1994). Thus, for a specific streamcorridor project, measurement of diver-sity may be limited to a target group ofspecial concern. In this manner, com-parison of diversity levels becomesmore meaningful.

Spatial Scale

Diversity can be measured within thebounds of a single community, acrosscommunity boundaries, or in largeareas encompassing many communi-ties. Diversity within a relativelyhomogeneous community is knownas alpha diversity. Diversity betweencommunities, described as the amountof differentiation along habitat gradi-ents, is termed beta diversity. The totaldiversity across very large landscapesis gamma diversity. Noss and Harris(1986) note that management foralpha diversity may increase localspecies richness, while the regionallandscape (gamma diversity) may be-come more homogeneous and lessdiverse overall. They recommend agoal of maintaining the regional speciespool in an approximately natural rela-tive abundance pattern. The specificsize of the area of concern should bedefined when diversity objectives areestablished.

Measures of Diversity

Magurran (1988) describes three maincategories of diversity measures — rich-ness indices, abundance models, andindices based on proportional abun-dance. Richness indices are measuresof the number of species (or otherelement of diversity) in a specific sam-pling unit and are the most widely usedindices (Magurran 1988). Abundancemodels account for the evenness (equi-tability) of distribution of species andfit various distributions to known mod-els, such as the geometric series, log se-ries, lognormal, or broken stick. Indicesbased on the proportional abundanceof species combine both richness andevenness into a single index. A varietyof such indices exist, the most commonof which is the Shannon-Weaver diver-sity index (Krebs 1978):

H = –Σpilog

ep

i


where

H = index of species diversity

S = number of species

pi = proportion of total sample

belonging to the ith species

Results of most studies using diversityindices are relatively insensitive to theparticular index used (Ricklefs 1979).For example, bird species diversity in-dices from 267 breeding bird censuseswere highly correlated (r = 0.97) withsimple counts of bird species richness(Tramer 1969). At the species level, asimple measure of richness is mostoften used in conservation biologystudies because the many rare speciesthat characterize most systems are gen-erally of greater interest than the com-mon species that dominate in diversityindices and because accurate popula-tion density estimates are often notavailable (Meffe et al. 1994).

Simple measures of species richness,however, are not sensitive to the actualspecies composition of an area. Similarrichness values in two different areasmay represent very different sets ofspecies. The usefulness of these mea-sures can be increased by consideringspecific subsets of species of most con-cern, as mentioned above. Magurran(1988) recommends going beyond theuse of a single diversity measure and ex-amining the shape of the species abun-dance distribution as well. Breedingbird census data from an 18-hectare(ha) riparian deciduous forest habitatin Ohio (Tramer 1996) can be used toillustrate these different methods ofpresentation (Figure 7.36). Breedingbird species richness in this riparianhabitat was 38.

Pielou (1993) recommends the use ofthree indices to adequately assess diver-sity in terrestrial systems:

■ A measure of plant species diversity.

■ A measure of habitat diversity.

■ A measure of local rarity.

Other indices used to measure variousaspects of diversity include vegetationmeasures, such as foliage height diver-sity (MacArthur and MacArthur 1961),and landscape measures, such as fractaldimension, fragmentation indices, andjuxtaposition (Noss 1994).

Related Integrity Indices

Karr (1981) developed the Index of Bi-otic Integrity to assess the diversity andhealth of aquatic communities. Thisindex is designed to assess the presentstatus of the aquatic community usingfish community parameters related tospecies composition, species richness,and ecological factors. Species composi-tion and richness parameters may in-clude the presence of intolerant species,the richness and composition of spe-cific species groups (e.g., darters), or theproportion of specific groups (e.g., hy-brid individuals). Ecological parametersmay include the proportion of top car-nivores, number of individuals, or pro-portion with disease or otheranomalies. Key parameters are devel-oped for the stream system of interest,and each parameter is assigned a rating.The overall rating of a stream is used toevaluate the quality of the aquaticbiota.

Rapid Bioassessment

Rapid bioassessment techniques aremost appropriate when restorationgoals are nonspecific and broad, suchas improving the overall aquatic com-munity or establishing a more balancedand diverse community in the streamcorridor. Bioassessment often refers touse of biotic indices or compositeanalyses, such as those used by OhioEPA (1990), and rapid bioassessmentprotocols (RBP), such as those docu-mented by Plafkin et al. (1989). Ohio


EPA evaluates biotic integrity by usingan invertebrate community index (ICI)that emphasizes structural attributes ofinvertebrate communities and com-pares the sample community with a ref-erence or control community. The ICI isbased on 10 metrics that describe differ-ent taxonomic and pollution tolerancerelationships within the macroinverte-brate community. The RBP establishedby USEPA (Plafkin et al. 1989) were de-veloped to provide states with the tech-nical information necessary forconducting cost-effective biological as-sessments. The RBP are divided into fivesets of protocols (RBP I to V), three formacroinvertebrates and two for fish(Table 7.8).

Algae

Although not detailed by Plafkin et al.(1989), algal communities are usefulfor bioassessment. Algae generally haveshort life spans and rapid reproductionrates, making them useful for evaluatingshort-term impacts. Sampling impactsare minimal to resident biota, and col-lection requires little effort. Primaryproductivity of algae is affected by phys-ical and chemical impairments. Algalcommunities are sensitive to some pol-lutants that might not visibly affectother aquatic communities. Algal com-munities can be examined for indicatorspecies, diversity indices, taxa richness,community respiration, and coloniza-tion rates. A variety of nontaxonomicevaluations, such as biomass andchlorophyll, may be used and are sum-marized in Weitzel (1979). Rodgers etal. (1979) describe functional measure-ments of algal communities, such asprimary productivity and communityrespiration, to evaluate the effects ofnutrient enrichment.

Although collecting algae in streams re-quires little effort, identifying for met-rics, such as diversity indices and taxa

Ab

un

dan

ce

0 5 10 15 20 25 30 35 40Species Sequence

0

5

10

15

20

SpeciesSequence

1

Species

American robin

Abundancein 18-ha Plot

18.5

2 House wren 13

3 Gray catbird 10.5

4 Song sparrow 9.5

5 Northern cardinal 7.5

Baltimore oriole 76

Warbling vireo 67

Wood thrush 4.58

Common grackle 4.59

Eastern wood-pewee 410

Red-eyed vireo 411

Indigo bunting 412

Red-winged blackbird 413

Mourning dove 314

Northern flicker 315

Blue jay 316

Tufted titmouse 317

White-breasted nuthatch 318

American redstart 319

Rose-breasted grosbeak 320

Downy woodpecker 221

Great crested flycatcher 222

Black-capped chickadee 223

Carolina wren 224

European starling 225

Yellow warbler 226

Brown-headed cowbird 227

American goldfinch 228

Wood duck 129

Ruby-throated hummingbird 130

Red-bellied woodpecker 131

Hairy woodpecker 132

Tree swallow 133

Blue-gray gnatcatcher 134

35 Prothonotary warbler 1

36 Common yellowthroat 1

37 Eastern phoebe 1

38 N. rough-winged swallow 1

Figure 7.36: Breeding bird census data. Speciesabundance curve in a riparian deciduous foresthabitat.Source: Tramer 1996.


richness, may require considerable ef-fort. A great deal of effort may be ex-pended to document diurnal andseasonal variations in productivity.

Benthic Macroinvertebrates

The intent of the benthic rapidbioassessment is to evaluate overall bio-logical condition, optimizing the use ofthe benthic community’s capacity to re-flect integrated environmental effectsover time. Using benthic macroinverte-brates is advantageous for the followingreasons:

■ They are good indicators of localizedconditions.

■ They integrate the effects of short-term environmental variables.

■ Degraded conditions are easilydetected.

■ Sampling is relatively easy.

■ They provide food for many fish ofcommercial or recreational impor-tance.

■ Macroinvertebrates are generallyabundant.

■ Many states already have backgrounddata.

As indicated above, the RBP are dividedinto three sets of protocols (RBP I toIII) for macroinvertebrates. RBP I is a“screening” or reconnaissance-levelanalysis used to discriminate obviouslyimpaired and nonimpaired sites frompotentially affected areas requiring fur-ther investigation. RBP II and III use aset of metrics based on taxon toleranceand community structure similar to theICI used by the state of Ohio. Both aremore labor-intensive than RBP I and in-corporate field sampling. RBP II usesfamily-level taxonomy to determine thefollowing set of metrics used in describ-ing the biotic integrity of a stream:

■ Taxa richness.

■ Hilsenhoff biotic index (Hilsenhoff1988).

■ Ratio of scrapers to filtering collectors.

■ Ratio of Ephemeroptera/Plecoptera/Trichoptera (EPT) and chironomidabundances.

■ Percent contribution of dominanttaxa.

■ EPT index.

■ Community similarity index.

■ Ratio of shredders to total number ofindividuals.

Low; 1-2 hr per site (no standardized sampling)

Levelor Tier

OrganismGroup

I Benthicinvertebrates

Relative Level of Effort

Order, family/field

Level of Taxonomy/ Where Performed

One highly-trained biologist

Intermediate; 1.5-2.5 hr per site (all taxonomy performed in field)

II Benthicinvertebrates

Family/field One highly-trained biologist and one technician

Most rigorous; 3-5 hr per site (2-3 hr of total are for lab taxonomy)

III Benthicinvertebrates

Genus or species/laboratory

One highly-trained biologist and one technician

Low; 1-3 hr per site (no fieldwork involved)

IV Fish Not applicable One highly-trained biologist

Most rigorous; 2-7 hr per site (1-2 hr per site are for data analysis)

V Fish Species/field One highly-trained biologist and 1-2 technicians

Level of ExpertiseRequired

Table 7.8: Five tiers of the rapid bioassessment protocols. RBPs are used to conduct cost-effectivebiological assessments.Source: Plafkin et al. 1989.


RBP III further defines the level of bi-otic impairment and is essentially anintensified version of RBP II that usesspecies-level taxonomy. As with ICI, theRBP metrics for a site are compared tometrics from a control or reference site.

Fish

Hocutt (1981) states “perhaps the mostcompelling ecological factor is thatstructurally and functionally diverse fishcommunities both directly and indi-rectly provide evidence of water qualityin that they incorporate all the local en-vironmental perturbations into the sta-bility of the communities themselves.”

The advantages of using fish as bioindi-cators are as follows:

■ They are good indicators of long-term effects and broad habitat condi-tions.

■ Fish communities represent a varietyof trophic levels.

■ Fish are at the top of the aquaticfood chain and are consumed byhumans.

■ Fish are relatively easy to collect andidentify.

■ Water quality standards are oftencharacterized in terms of fisheries.

■ Nearly one-third of the endangeredvertebrate species and subspecies inthe United States are fish.

The disadvantages of using fish asbioindicators are as follows:

■ The cost.

■ Statistical validity may be hard toattain.

■ It is difficult to interpret findings.

Electrofishing is the most commonlyused field technique. Each collectingstation should be representative of thestudy reach and similar to other reachessampled; effort between reaches should

be equal. All fish species, not just gamespecies, should be collected for the fishcommunity assessment (Figure 7.37).Karr et al. (1986) used 12 biologicalmetrics to assess biotic integrity usingtaxonomic and trophic compositionand condition and abundance of fish.Although the Index of Biological In-tegrity (IBI) developed by Karr was de-signed for small midwestern streams, ithas been modified for many regions ofthe country and for use in large rivers(see Plafkin et al. 1989).

Establishing a Standard ofComparison

With stream restoration activities, it isimportant to select a desired end condi-tion for the proposed management ac-tion. A predetermined standard ofcomparison provides a benchmarkagainst which to measure progress. Forexample, if the chosen diversity mea-sure is native species richness, the stan-dard of comparison might be themaximum expected native species rich-ness for a defined geographic area andtime period.

Figure 7.37: Fish samples. Water qualitystandards are often characterized in termsof fisheries.


Historical conditions in the regionshould be considered when establishinga standard of comparison. If currentconditions in a stream corridor aredegraded, it may be best to establishthe standard at a period in the past thatrepresented more natural or desiredconditions. Knopf (1986) notes that forcertain western streams, historical diver-sity might have been less than currentdue to changes in hydrology and en-croachment of native and exotic ripar-ian vegetation in the floodplain. Thus,it is important to agree on what condi-tions are desired prior to establishingthe standard of comparison. In addi-tion, the geographic location and sizeof the area should be considered. Pat-terns of diversity vary with geographiclocation, and larger areas are typicallymore diverse than smaller areas.

The IBI is scaled to a standard of com-parison determined through either pro-fessional judgment or empirical data,and such indices have been developedfor a variety of streams (Leonard andOrth 1986, Bramblett and Fausch 1991,Lyons et al. 1996).

Evaluating the Chosen Index

For a hypothetical stream restorationinitiative, the following biological diver-sity objective might be developed. As-sume that a primary concern in the areais conserving native amphibian speciesand that 30 native species of amphib-ians have been known to occur histori-cally in the 386 m2 watershed. Theobjective could be to manage thestream corridor to provide and main-tain suitable habitat for the 30 nativeamphibian species.

Stream corridor restoration efforts mustbe directed toward those factors thatcan be managed to increase diversity tothe desired level. Those factors might bethe physical and structural features ofthe stream corridor or possibly the pres-

ence of an invasive species in the com-munity. Knowledge of the importantfactors can be obtained from existingliterature and from discussions withlocal and regional experts.

Diversity can be measured directly orpredicted from other information. Di-rect measurement requires an actual in-ventory of the element of diversity, suchas counting the amphibian species inthe study area. The IBI requires sam-pling fish populations to determine thenumber and composition of fishspecies. Measures of the richness of aparticular animal group require counts.Determining the number of species in acommunity is best accomplished with along-term effort because there can bemuch variation over short periods. Vari-ation can arise from observer differ-ences, sampling design, or temporalvariation in the presence of species.

Direct measures of diversity are mosthelpful when baseline information isavailable for comparing different sites.It is not possible, however, to directlymeasure certain attributes, such asspecies richness or the population levelof various species, for various futureconditions. For example, the IBI cannotbe directly computed for a predictedstream corridor condition, followingmanagement action.

Predictions of diversity for various fu-ture conditions, such as with restora-tion or management, require the use ofa predictive model. Assume the diver-sity objective for a stream corridorrestoration effort is to maximize nativeamphibian species richness. Based onknowledge of the life history of thespecies, including requirements forhabitat, water quality, or landscapeconfiguration, a plan can be developedto restore a stream corridor to meetthese needs. The plan could include aset of criteria or a model to describethe specific features that should be


included to maximize amphibian rich-ness. Examples of indirect methods toassess diversity include habitat models(Schroeder and Allen 1992, Adamus1993) and cumulative impact assess-ment methods (Gosselink et al. 1990,Brooks et al. 1991).

Predicting diversity with a model isgenerally more rapid than directly mea-suring diversity. In addition, predictivemethods provide a means to analyzealternative future conditions before im-plementing specific restoration plans.The reliability and accuracy of diversitymodels should be established beforetheir use.

Classification Systems

Classification is an important compo-nent of many of the scientific disci-plines relevant to streamcorridors— hydrology, geomorphology,limnology, plant and animal ecology.Table 7.9 lists some of the classificationsystems that might be useful in identify-ing and planning riverine restorationactivities. It is not the intent of this sec-tion to exhaustively review all classifica-

tion schemes or to present a single rec-ommended classification system. Rather,we focus on some of the principal dis-tinctions among classification systemsand factors to consider in the use ofclassification systems for restorationplanning, particularly in the use of aclassification system as a measure ofbiological condition. It is likely thatmultiple systems will be useful in mostactual riverine restoration programs.

The common goal of classification systems is to organize variation. Impor-tant dimensions in which riverine clas-sification systems differ include thefollowing:

■ Geographic domain. The range of sitesbeing classified varies from rivers ofthe world to local differences in thecomposition and characteristics ofpatches within one reach of a singleriver.

■ Variables considered. Some classifica-tions are restricted to abiotic vari-

Riparian vegetation of Yampa, San Miguel/Dolores River Basins

Classification System

Plant communities

Subject

Colorado

GeographicDomain

Kittel and Lederer (1993)

Riparian and scrubland communities of Arizona and New Mexico

Plant communities Arizona and New Mexico

Szaro (1989)

Classification of Montana riparian and wetland sites

Plant communities Montana Hansen et al. (1995)

Integrated riparian evaluation guide

Hydrology, geomorphology, soils, vegetation

Intermountain U.S. Forest Service (1992)

Streamflow cluster analysis Hydrology with correlations to fish and invertebrates

National Pott and Ward (1989)

River Continuum Hydrology, stream order, water chemistry, aquatic communities

International,national

Vannote et al. (1980)

World-wide stream classification

Hydrology, water chemistry, substrate, vegetation

International Pennak (1971)

Rosgen’s river classification Hydrology, geomorphology: stream and valley types

National Rosgen (1996)

Hydrogeomorphic wetland classification

Hydrology, geomorphology, vegetation

National Brinson (1993)

Recovery classes following channelization

Hydrology, geomorphology, vegetation

Tennessee Hupp (1992)

Citation

Table 7.9: Selected riverine and riparian classi-fication systems. Classification systems areuseful in characterizing biological conditions.


ables of hydrology, geomorphology,and aquatic chemistry. Other com-munity classifications are restrictedto biotic variables of species compo-sition and abundance of a limitednumber of taxa. Many classificationsinclude both abiotic and biotic vari-ables. Even purely abiotic classifica-tion systems are relevant to biologi-cal evaluations because of the impor-tant correlations (e.g., the whole con-cept of physical habitat) between abi-otic structure and community com-position.

■ Incorporation of temporal relations.Some classifications focus ondescribing correlations and similari-ties across sites at one, perhaps ideal-ized, point in time. Other classifica-tions identify explicit temporal tran-sitions among classes, for example,succession of biotic communities orevolution of geomorphic landforms.

■ Focus on structural variation or func-tional behavior. Some classificationsemphasize a parsimonious descrip-tion of observed variation in the clas-sification variables. Others use classi-fication variables to identify typeswith different behaviors. For exam-ple, a vegetation classification can bebased primarily on patterns ofspecies co-occurrence, or it can bebased on similarities in functionaleffect of vegetation on habitat value.

■ The extent to which management alter-natives or human actions are explicitlyconsidered as classification variables.To the extent that these variables arepart of the classification itself, theclassification system can directly predict the result of a managementaction. For example, a vegetationclassification based on grazing in-tensity would predict a change fromone class of vegetation to anotherclass based on a change in grazingmanagement.

Use of Classification Systems inRestoring Biological Conditions

Restoration efforts may apply severalnational and regional classification sys-tems to the riverine site or sites of inter-est because these are efficient ways tosummarize basic site description andinventory information and they can fa-cilitate the transference of existing in-formation from other similar systems.

Most classification systems are generallyweak at identifying causal mechanisms.To varying degrees, classification sys-tems identify variables that efficientlydescribe existing conditions. Rarely dothey provide unequivocal assuranceabout how variables actually cause theobserved conditions. Planning efficientand effective restoration actions gener-ally requires a much more mechanisticanalysis of how changes in controllablevariables will cause changes toward de-sired values of response variables. A sec-ond limitation is that application of aclassification system does not substitutefor goal setting or design. Comparisonof the degraded system to an actualunimpacted reference site, to the idealtype in a classification system, or to arange of similar systems can provide aframework for articulating the desiredstate of the degraded system. However,the desired state of the system is amanagement objective that ultimatelycomes from outside the classification ofsystem variability.

Analyses of SpeciesRequirements

Analyses of species requirements in-volve explicit statements of how vari-ables interact to determine habitat orhow well a system provides for the liferequisites of fish and wildlife species.Complete specification of relations be-tween all relevant variables and allspecies in a stream corridor system isnot possible. Thus, analyses based on


species requirements focus on one ormore target species or groups of species.In a simple case, this type of analysismay be based on an explicit statementof the physical factors that distinguishgood habitat for a species (places whereit is most likely to be found or where itbest reproduces) from poor habitat(places where it is unlikely to be foundor reproduces poorly). In more compli-cated cases, such approaches incorpo-rate variables beyond those of purelyphysical habitat, including other speciesthat provide food or biotic structure,other species as competitors or preda-tors, or spatial or temporal patterns ofresource availability.

Analyses based on species requirementsdiffer from synthetic measures of sys-tem condition in that they explicitly in-corporate relations between “causal”variables and desired biological attri-butes. Such analyses can be used di-rectly to decide what restoration actionswill achieve a desired result and to eval-uate the likely consequences of a pro-posed restoration action. For example,an analysis using the habitat evaluationprocedures might identify mast produc-tion (the accumulation of nuts from aproductive fruiting season which servesas a food source for animals) as a factorlimiting squirrel populations. If squir-rels are a species of concern, at leastsome parts of the stream restoration ef-fort should be directed toward increas-ing mast production. In practice, thislogical power is often compromised byincomplete knowledge of the specieshabitat requirements.

The complexity of these methods variesalong a number of important dimen-sions, including prediction of habitatsuitability versus population numbers,analysis for a single place and singletime versus a temporal sequence ofspatially complex requirements, andanalysis for a single target species versus

a set of target species involving trade-offs. Each of these dimensions must becarefully considered in selecting ananalysis procedure appropriate to theproblem at hand.

The Habitat EvaluationProcedures (HEP)

Habitat evaluation procedures (HEP)can be used for several different types ofhabitat studies, including impact assess-ment, mitigation, and habitat manage-ment. HEP provides information fortwo general types of habitat compar-isons— the relative value of differentareas at the same point in time and therelative value of the same area at differ-ent points in time. Potential changes inwildlife (both aquatic and terrestrial)habitat due to proposed projects arecharacterized by combining these twotypes of comparisons.

Basic Concepts

HEP is based on two fundamental eco-logical principles— habitat has a defin-able carrying capacity, or suitability, tosupport or produce wildlife popula-tions (Fretwell and Lucas 1970), andthe suitability of habitat for a givenwildlife species can be estimated usingmeasurements of vegetative, physical,and chemical traits of the habitat. Thesuitability of a habitat for a givenspecies is described by a habitat suit-ability index (HSI) constrained between0 (unsuitable habitat) and 1 (optimumhabitat). HSI models have been devel-oped and published by the U.S. Fishand Wildlife Service (Schamberger et al.1982; Terrell and Carpenter, in press),and USFWS (1981) provides guidelinesfor use in developing HSI models forspecific projects. HSI models can bedeveloped for many of the previouslydescribed metrics, including species,guilds, and communities (Schroederand Haire 1993).


The fundamental unit of measure inHEP is the Habitat Unit, computed asfollows:

HU = AREA x HSI

where HU is the number of habitatunits (units of area), AREA is the arealextent of the habitat being described(units of area), and HSI is the index ofsuitability of the habitat (unitless).Conceptually, an HU integrates thequantity and quality of habitat into asingle measure, and one HU is equiva-lent to one unit of optimal habitat.

Use of HEP to Assess Habitat Changes

HEP provides an assessment of the netchange in the number of HUs attribut-able to a proposed future action, suchas a stream restoration initiative. A HEPapplication is essentially a two-stepprocess— calculating future HUs for aparticular project alternative and calcu-lating the net change as compared to abase condition.

The steps involved in using and apply-ing HEP to a management project areoutlined in detail in USFWS (1980a).However, some early planning decisionsoften are given little attention althoughthey may be the most important part ofa HEP study. These initial decisions in-clude forming a study team, definingthe study boundaries, setting study ob-jectives, and selecting the evaluationspecies. The study team usually consistsof individuals representing differentagencies and viewpoints. One memberof the team is generally from the leadproject planning agency and othermembers are from resources agencieswith an interest in the resources thatwould be affected.

One of the first tasks for the team is todelineate the study area boundaries.The study area boundaries should bedrawn to include any areas of direct im-pact, such as a flood basin for a new

reservoir, and any areas of secondaryimpact, such as a downstream riverreach that might have an altered flow,increased turbidity, or warmer tempera-ture, or riparian or upland areas subjectto land use changes as a result of an in-creased demand on recreational lands.Areas such as an upstream spawningground that are not contiguous to theprimary impact site also might be af-fected and therefore should be includedin the study area.

The team also must establish projectobjectives, an often neglected aspect ofproject planning. Objectives shouldstate what is to be accomplished in theproject and specify an endpoint to theproject. An integral aspect of objectivesetting is selecting evaluation species,the specific wildlife resources of con-cern for which HUs will be computedin the HEP analysis. These are often in-dividual species, but they do not haveto be. Depending on project objectives,species’ life stages (e.g., juvenilesalmon), species’ life requisites (e.g.,spawning habitat), guilds (e.g., cavity-nesting birds), or communities (e.g.,avian richness in riparian forests) canbe used.

Instream Flow IncrementalMethodology

The Instream Flow IncrementalMethodology (IFIM) is an adaptive sys-tem composed of a library of modelsthat are linked to describe the spatialand temporal habitat features of a givenriver. IFIM is described in Chapter 5under Supporting Analysis for SelectingRestoration Alternatives.

Physical Habitat Simulation

The Physical Habitat Simulation(PHABSIM) model was designed by theU.S. Fish and Wildlife Service primarilyfor instream flow analysis (Bovee1982). It represents the habitat evalua-


tion component of a larger instreamflow incremental methodology for in-corporating fish habitat considerationinto flow management, presented inChapter 5. PHABSIM is a collection ofcomputer programs that allows evalua-tion of available habitat within a studyreach for various life stages of differentfish species. The two basic componentsof the model are hydraulic simulation(based on field-measured cross-sec-tional data) and several standard hy-draulic methods for predicting watersurface elevations and velocities at un-measured discharges (e.g., stage vs. dis-charge relations, Manning’s equation,step-backwater computations). Habitatsimulation integrates species and life-stage-specific habitat suitability curvesfor water depth, velocity, and substratewith the hydraulic data. Output is aplot of weighted usable area (WUA)against discharge for the species and lifestages of interest. (Figure 7.38)

The stream hydraulic component pre-dicts depths and water velocities at un-observed flows at specific locations on across section of a stream. Field measure-ments of depth, velocity, substrate ma-terial, and cover at specific samplingpoints on a cross section are taken atdifferent observable flows. Hydraulicmeasurements, such as water surface el-evations, also are collected during thefield inventory. These data are used tocalibrate the hydraulic simulation mod-els. The models then are used to predictdepths and velocities at flows differentfrom those measured.

The habitat component weights eachstream cell using indices that assign arelative value between 0 and 1 for eachhabitat attribute (depth, velocity, sub-strate material, cover), indicating howsuitable that attribute is for the lifestage under consideration. These at-tribute indices are usually termed habi-tat suitability indices and are developed

from direct observations of the attrib-utes used most often by a life stage,from expert opinion about what the liferequisites are, or a combination. Vari-ous approaches are taken to factor as-sorted biases out of these suitabilitydata, but they remain indices that areused as weights of suitability. In the laststep of the habitat component, hy-draulic estimates of depth and velocityat different flow levels are combinedwith the suitability values for those at-tributes to weight the area of each cellat the simulated flows. The weightedvalues for all cells are summed to pro-duce the WUA.

There are many variations on the basicapproach outlined above, with specificanalyses tailored for different watermanagement phenomena (such as hy-dropeaking and unique spawning habi-tat needs), or for special habitat needs(such as bottom velocity instead ofmean column velocity) (Milhous et al.1989). However, the fundamentals ofhydraulic and habitat modeling remainthe same, resulting in a WUA versus dis-charge function. This function shouldbe combined with the appropriate hy-drologic time series (water availability)to develop an idea of what life statesmight be affected by a loss or gain ofavailable habitat and at what time ofthe year. Time series analysis plays thisrole and also factors in any physicaland institutional constraints on watermanagement so that alternatives can beevaluated (Milhous et al. 1990).

Several things must be rememberedabout PHABSIM. First, it provides anindex to microhabitat availability; it isnot a measure of the habitat actuallyused by aquatic organisms. It can beused only if the species under consider-ation exhibit documented preferencesfor depth, velocity, substrate material,cover, or other predictable microhabitatattributes in a specific environment of

REVERSE

Review Chapter5’ s SupportingAnalysis forSelectingRestorationAlternatives


competition and predation. The typicalapplication of PHABSIM assumes rela-tively steady flow conditions such thatdepths and velocities are comparablystable within the chosen time step.PHABSIM does not predict the effects offlow on channel change. Finally, thefield data and computer analysis re-quirements can be relatively large.

Two-dimensional Flow Modeling

Concern about the simplicity of theone-dimensional hydraulic models usedin PHABSIM has led to current researchinterest in the use of more sophisticatedtwo-dimensional hydraulic models to

V1 V2 V3 . . . . . . . . . . velocityD1 D2 D3. . . . . . . . . . depthC1 C2 C3 . . . . . . . . . . coverA1 A2 A3. . . . . . . . . . area

cross section B

0

00

100,000

00 100

100

0.20.40.6

0.81.0

Velocity (ft/sec)

Sl

2 3 4

A. Site-Specific Microhabitat Data

C. Seasonal Relation Between Discharge andMicrohabitat for Each Life Stage

B. Habitat SuitabilityCriteria

cross section A

flow

10

0.20.40.6

0.81.0

Depth (ft)

Discharge

Sl

Wei

gh

ted

Un

stab

le A

rea

2 3 4 0 1Cover

2 3 4

Figure 7.38: Conceptualization of how PHAB-SIM calculates habitat values as a function ofdischarge. A. First, depth (Di), velocity (Vi),cover conditions (Ci), and area (Ai) are mea-sured or simulated for a given discharge. B.Suitability index (SI) criteria are used to weightthe area of each cell for the discharge. Thehabitat values for all cells in the study reachare summed to obtain a single habitat valuefor the discharge. C. The procedure is repeatedfor a range of discharges.Modified from Nestler et al. 1989.


simulate physical conditions of depthand velocity for use in fish habitatanalysis. A two-dimensional hydraulicmodel can be spatially adjusted to rep-resent the scale of aquatic habitat andthe variability of other field data. Forexample, the physical relationship be-tween different aquatic habitat types isoften a key parameter when consideringfish habitat use. The spatial nature oftwo-dimensional flow modeling allowsfor the analysis of these relationships.The model can also consider the dryingand wetting of intermittent streamchannels.

Leclerc et al. (1995) used two-dimen-sional flow modeling to study the effectof a water diversion on the habitat ofjuvenile Atlantic salmon (Salmo salar)in the Moisie River in Quebec, Canada.Average model error was reduced whencompared with traditional one-dimen-sional models. Output from the two-di-mensional modeling was combinedwith habitat suitability indexes with fi-nite element calculation techniques.Output from the analysis includedmaps displaying the spatial distributionof depth, velocity, and habitat suitabil-ity intervals.

Physical data collection for this model-ing tool is intensive. Channel contourand bed material mapping is requiredalong with discharge relationships andthe upstream and downstream bound-aries of each study reach. Velocity andwater-surface measurements for variousdischarges are required for model cali-bration. Two-dimensional modelingdoes not address all of the issues relatedto hydrodynamics and flow modeling.Mobile bed systems and variability inManning’s coefficient are still problem-atic using this tool (Leclerc et al. 1995).Moderate to large rivers with a stablebedform are most suited to thismethodology.

Riverine Community HabitatAssessment and RestorationConcept Model (RCHARC)

Another modeling approach to aquatichabitat restoration is the Riverine Com-munity Habitat Assessment andRestoration (RCHARC) concept. Thismodel is based on the assumption thataquatic habitat in a restored streamreach will best mimic natural condi-tions if the bivariate frequency distribu-tion of depth and velocity in the subjectchannel is similar to a reference reachwith good aquatic habitat. Study siteand reference site data can be measuredor calculated using a computer model.The similarity of the proposed designand reference reach is expressed withthree-dimensional graphs and statistics(Nestler et al. 1993, Abt 1995).RCHARC has been used as the primarytool for environmental analysis onstudies of flow management for theMissouri River and the Alabama-Coosa-Tallapoosa Apalachicola-Chatta-hoochee-Flint Basin.

Time Series Simulations

A relatively small number of applica-tions have been made of time seriessimulations of fish population or indi-vidual fish responses to riverine habitatchanges. Most of these have usedPHABSIM to accomplish hydraulicmodel development and validation andhydraulic simulation, but some havesubstituted time-series simulations ofindividual or population responses forhabitat suitability curve developmentand validation, and habitat suitabilitymodeling. PHABSIM quantifies the rela-tionship of hydraulic estimates (depthand velocity) and measurements (sub-strate and cover) with habitat suitabilityfor target fish and invertebrate lifestages or water-related recreation suit-ability. It is useful when relativelysteady flow is the major determinant


controlling riverine resources. Use ofPHABSIM is generally limited to riversystems in which dissolved oxygen, sus-pended sediment, nutrient loading,other chemical aspects of water quality,and interspecific competition do notplace the major limits on populationsof interest. These limitations to the useof PHABSIM can be abated or removedwith models that simulate response ofindividual fish or fish populations.

Individual-based Models

The Electric Power Research Institute(EPRI) program on compensatorymechanisms in fish populations(CompMech) has the objective of im-proving predictions of fish populationresponse to increased mortality, loss ofhabitat, and release of toxicants (EPRI1996). This technique has been appliedby utilities and resource managementagencies in assessments involving directmortality due to entrainment, impinge-ment, or fishing; instream flow; habitatalteration (e.g., thermal discharge,water-level fluctuations, water diver-sions, exotic species); and ecotoxicity.Compensation is defined as the capac-ity of a population to self-mitigate de-creased growth, reproduction, orsurvival of some individuals in the pop-ulation by increased growth, reproduc-tion, or survival of the remainingindividuals. The CompMech approachover the past decade has been to repre-sent in simulation models the processesunderlying daily growth, reproduction,and survival of individual fish (hencethe classification of individual-basedmodels) and then to aggregate over in-dividuals to the population level.

The models can be used to make short-term predictions of survival, growth,habitat utilization, and consumptionfor critical life stages. For the longerterm, the models can be used to projectpopulation abundance through time to

assess the risk that abundance will fallbelow some threshold requiring mitiga-tion. For stream situations, severalCompMech models have been devel-oped that couple the hydraulic simula-tion method of PHABSIM directly withan individual-based model of reproduc-tion and young-of-year dynamics,thereby eliminating reliance on thehabitat-based component of PHABSIM(Jager et al. 1993). The CompMechmodel of smallmouth bass is beingused to evaluate the effects of alterna-tive flow regimes on nest success,growth, mortality, and ultimately yearclass strength in a Virginia stream toidentify instream flows that protect fish-eries with minimum impact on hy-dropower production.

A model of coexisting populations ofrainbow and brown trout in Californiais being used to evaluate alternative in-stream flow and temperature scenarios(Van Winkle et al. 1996). Model predic-tions will be compared with long-termfield observations before and after ex-perimental flow increases; numerousscientific papers are expected from thisintensive study.

An individual-based model of smoltproduction by Chinook salmon, as partof an environmental impact statementfor the Tuolumne River in California,considered the minimum stream flowsnecessary to ensure continuation andmaintenance of the anadromous fishery(FERC 1996). That model, the OakRidge Chinook salmon model (ORCM),predicts annual production of salmonsmolts under specified reservoir mini-mum releases by evaluating critical fac-tors, including influences on upstreammigration of adults, spawning and incu-bation of eggs, rearing of young, andpredation and mortality losses duringthe downstream migration of smolts.Other physical habitat analyses wereused to supplement the population


model in evaluating benefits of alterna-tive flow patterns. These habitat evalua-tions are based on data from aninstream flow study; a stream tempera-ture model was used to estimate flowsneeded to maintain downstream tem-peratures within acceptable limits forsalmon.

SALMOD

The conceptual and mathematical mod-els for the Salmonid Population Model(SALMOD) were developed for Chi-nook salmon in concert with a 12-yearflow evaluation study in the TrinityRiver of California using experts on thelocal river system and fish species inworkshop settings (Williamson et al.1993, Bartholow et al. 1993). SALMODwas used to simulate young-of-year pro-duction, assuming that the flow sched-ules to be evaluated were released fromLewiston Reservoir in every year from1976 to 1992 (regardless of observedreservoir inflow, storage, and releaselimitations).

The structure of SALMOD is a middleground between a highly aggregatedclassical population model that trackscohorts/size groups for a generally largearea without spatial resolution, and anindividual-based model that tracks indi-viduals at a great level of detail for agenerally small area. The conceptualmodel states that fish growth, move-ment, and mortality are directly relatedto physical hydraulic habitat and watertemperature, which in turn relate to thetiming and amount of regulated stream-flow. Habitat capacity is characterized bythe hydraulic and thermal properties ofindividual mesohabitats, which are themodel’s spatial computational units.

Model processes include spawning(with redd superimposition), growth(including maturation), movement(freshet-induced, habitat-induced, and

seasonal), and mortality (base, move-ment-related, and temperature-related).The model is limited to freshwaterhabitat for the first 9 months of life; es-tuarine and ocean habitats are not in-cluded. Habitat area is computed fromflow/habitat area functions developedempirically. Habitat capacity for eachlife stage is a fixed maximum numberper unit of habitat available. Thus, amaximum number of individuals foreach computational unit is calculatedfor each time step based on streamflowand habitat type. Rearing habitat capac-ity is derived from empirical relationsbetween available habitat area andnumber of individual fish observed.

Partly due to drought conditions, mostof the flow alternatives to be evaluateddid not actually occur during the flowevaluation study. When there is insuffi-cient opportunity to directly observeand evaluate impacts of flow alterna-tives on fish populations, SALMOD canbe used to simulate young-of-the-yearproduction that may result from pro-posed flow schedules to be released orregulated by a control structure such asa reservoir or diversion.

Other physical habitat analyses can beused to supplement population modelsin evaluating benefits of alternative flowpatterns. In the Trinity River FlowStudy, a stream temperature model wasused to estimate flows needed to main-tain downstream temperatures withinacceptable limits for salmon. Both theORCM (FERC 1996) and SALMODmodels concentrated on development,growth, movement, and mortality ofyoung-of-year Chinook salmon butwith different mechanistic inputs, spa-tial resolution, and temporal precision.


Vegetation-HydroperiodModeling

In most cases, the dominant factor thatmakes the riparian zone distinct fromthe surrounding uplands, and the mostimportant gradient in structuring varia-tion within the riparian zone, is sitemoisture conditions, or hydroperiod(Figure 7.39). Hydroperiod is definedas the depth, duration, and frequencyof inundation and is a powerful deter-minant of what plants are likely to befound in various positions in the ripar-ian zone. Formalizing this relation as avegetation-hydroperiod model can pro-vide a powerful tool for analyzing exist-ing distributions of riparian vegetation,casting forward or backward in time toalternative distributions, and designingnew distributions. The suitability of siteconditions for various species of plantscan be described with the same concep-tual approach used to model habitatsuitability for animals. The basic logicof a vegetation-hydroperiod model isstraightforward. How wet a site is has alot to do with what plants typicallygrow on the site. It is possible to mea-sure how wet a site is and, more impor-tantly, to predict how wet a site will bebased on the relation of the site to astream. From this, it is possible to esti-mate what vegetation is likely to occuron the site.

Components of a Vegetation-hydroperiod Model

The two basic elements of the vegeta-tion-hydroperiod relation are the physi-cal conditions of site moisture atvarious locations and the suitability ofthose sites for various plant species. Inthe simplest case of describing existingpatterns, site moisture and vegetationcan be directly measured at a numberof locations. However, to use the vege-tation-hydroperiod model to predict ordesign new situations, it is necessary to

predict new site moisture conditions.The most useful vegetation-hydroperiodmodels have the following three com-ponents:

■ Characterization of the hydrology or pat-tern of streamflow. This can take theform of a specific sequence of flows,a summary of how often differentflows occur, such as a flow durationor flood frequency curve, or a repre-sentative flow value, such as bankfulldischarge or mean annual discharge.

■ A relation between streamflow and mois-ture conditions at sites in the riparianzone. This relation can be measuredas the water surface elevation at avariety of discharges and summarizedas a stage vs. discharge curve. It canalso be calculated by a number ofhydraulic models that relate watersurface elevations to discharge, takinginto account variables of channelgeometry and roughness or resistanceto flow. In some cases, differences insimple elevation above the channelbottom may serve as a reasonableapproximation of differences ininundating discharge.

■ A relation between site moisture condi-tions and the actual or potential vegeta-tion distribution. This relation express-es the suitability of a site for a plantspecies or cover type based on themoisture conditions at the site. It canbe determined by sampling the dis-tribution of vegetation at a variety ofsites with known moisture condi-tions and then deriving probabilitydistributions of the likelihood offinding a plant on a site given themoisture conditions at the site.General relations are also availablefrom the literature for many species.

The nature and complexity of thesecomponents can vary substantially andstill provide a useful model. However,the components must all be expressed

FASTFORWARD

Preview Chapter 8 ’ snformation onvegetation-hydroperiodmodel.


in consistent units and must have a do-main of application that is appropriateto the questions being asked of themodel (i.e., the model must be capableof changing the things that need to bechanged to answer the question). Inmany cases, it may be possible to for-mulate a vegetation-hydroperiod modelusing representations of stream hydrol-ogy and hydraulics that have been de-veloped for other analyses such aschannel stability, fish habitat suitability,or sediment dynamics.

Identifying Non-equilibriumConditions

In altered or degraded stream systems,current moisture conditions in the ri-parian zone may be dramatically un-suitable for the current, historical, ordesired riparian vegetation. Several con-ditions can be relatively easily identi-fied by comparing the distribution ofvegetation to the distribution of vegeta-tion suitabilities.

■ The hydrology of the stream hasbeen altered; for example, if stream-flow has diminished by diversion orflood attenuation, sites in the ripari-an zone may be drier and no longersuitable for the historic vegetation orfor current long-lived vegetation thatwas established under a previoushydrologic regime.

■ The inundating discharges of plots inthe riparian zone have been alteredso that streamflow no longer has thesame relation to site moisture condi-tions; for example, levees, channelmodifications, and bank treatmentsmay have either increased ordecreased the discharge required toinundate plots in the riparian zone.

■ The vegetation of the riparian zonehas been directly altered, for exam-ple, by clearing or planting so thatthe vegetation on plots no longer

corresponds to the natural vegetationfor which the plots are suitable.

In many degraded stream systems all ofthese things have happened. Under-standing how the moisture conditionsof plots correspond to the vegetation inthe current system, as well as how theywill correspond in the restored system,is an important element of formulatingreasonable restoration objectives anddesigning a restoration plan.

Vegetation Effects of SystemAlterations

In a vegetation-hydroperiod model,vegetation suitability is determined bystreamflow and the inundating dis-charges of plots in the riparian zone.The model can be used to predict ef-fects of alteration in streamflow or therelations of streamflow to plot moistureconditions on the suitability of the ri-parian zone for different types of vege-tation. Thus, the effects of flowalterations and changes in channel orbottomland topography proposed aspart of a stream restoration plan can beexamined in terms of changes in thesuitability of various locations in the ri-parian zone for different plant species.

Figure 7.39:Vegetation/waterrelationship. Soilmoisture conditionsoften determine theplant communitiesin riparian areas.Source: C. Zabawa.


Extreme Events and DisturbanceRequirements

Temporal variability is a particularly im-portant characteristic of many streamecosystems. Regular seasonal differencesin biological requirements are examplesof temporal variability that are oftenincorporated into biological analysesbased on habitat suitability and timeseries simulations. The need forepisodic extreme events is easy to

ignore because these are so widely per-ceived as destructive both of biota andof constructed river features. In reality,however, these extreme events seem tobe essential to physical channel mainte-nance and to the long-term suitabilityof the riverine ecosystem for distur-bance-dependent species. Cottonwoodin western riparian systems is one well-understood case of a disturbance-de-pendent species. Cottonwoodregeneration from seed is generally re-stricted to bare, moist sites. Creatingthese sites depends heavily on channelmovement (meandering, narrowing,avulsion) or new flood deposits at highelevations. In some western ripariansystems, channel movement and depo-sition tend to occur infrequently in as-sociation with floods. The same eventsare also responsible for destroyingstands of trees. Thus maintaining goodconditions for existing stands, or fixingthe location of a stream’s banks withstructural measures, tends to reduce theregeneration potential and the long-term importance of this disturbance-dependent species in the system as awhole.

There is a large body of information on the floodingtolerances of various plant species. Summaries ofthis literature include Whitlow and Harris (1979) andthe multivolume Impact of Water Level Changes onWoody Riparian and Wetland Communities (Teskeyand Hinckley 1978, Walters et al. 1978, Lee andHinckley 1982, Chapman et al. 1982). This type ofinformation can be coupled to site moisture condi-tions predicted by applying discharge estimates orflood frequency analyses to the inundating dis-charges of sites in the riparian zone. The resultingrelation can be used to describe the suitability of

sites for various plant species, e.g., relatively flood-prone sites will likely have relatively flood-tolerantplants. Inundating discharge is strongly related torelative elevation within the floodplain. Other thingsbeing equal (i.e., within a limited geographic areaand with roughly equivalent hydrologic regimes),elevation relative to a representative water surfaceline, such as bankfull discharge or the stage at meanannual flow, can thus provide a reasonable surro-gate for site moisture conditions. Locally determinedvegetation suitability can then be used to determinethe likely vegetation in various elevation zones.

There are a number of statistical procedures for estimat-ing the frequency and magnitude of extreme events(see flood frequency analysis section of chapter 8) anddescribing various aspects of hydrologic variation.Changing these flow characteristics will likely changesome aspect of the distribution and abundance of organ-isms. Analyzing more specific biological changes generallyrequires defining the requirements of target species;defining requirements of their food sources, competitors,and predators; and considering how those requirementsare influenced by episodic disturbance events.

7.a hydrologic and hydraulic processes

Documents