The purpose of this study was to determinefaculty.washington.edu/dbooth/Henshaw and Booth...

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATIONVOL. 36, NO. 6 AMERICAN WATER RESOURCES ASSOCIATION DECEMBER 2000

NATURAL RESTABILIZATION OF STREAMCHANNELS IN URBAN WATERSHEDS'

Patricia C. Henshaw and Derek B. Booth2

ABSTRACT: Stream channels are known to change their form as aresult of watershed urbanization, but do they restabiize under sub-sequent conditions of constant urban land use? Streams in sevendeveloped and developing watersheds (drainage areas 5-35 km2) inthe Puget Sound lowlands were evaluated for their channel stabili-ty and degree of urbanization, using field and historical data. Pro-tocols for determining channel stability by visual assessment,calculated bed mobility at bankfull flows, and resurveyed cross-sec-tions were compared and yielded nearly identical results. We foundthat channel restabilization generally does occur within one or twodecades of constant watershed land use, but it is not universal.When (or if) an individual stream will restabilize depends on specif-ic hydrologic and geomorphic characteristics of the channel and itswatershed; observed stability is not well predicted by simply themagnitude of urban development or the rate of ongoing land-usechange. The tendency for channel restabilization suggests thatmanagement efforts focused primarily on maintaining stability,particularly in a still-urbanizing watershed, may not always be nec-essary. Yet physical stability alone is not a sufficient condition for abiologically healthy stream, and additional rehabilitation measureswill almost certainly be required to restore biological conditions inurban systems.(KEY TERMS: channel stability; urban channels; channel response;geomorphology; stability assessment; watershed urbanization;urban hydrology.)

INTRODUCTION

biological character of streams, often resulting indegraded water quality, loss or removal of streamhabitat, and physical alteration to the channel. Thelatter, which is the focus of this study, is commonlyexpressed by rapid erosion or deposition and changingchannel form. Although rebuilt habitat in a stabilizedurban stream may not provide the level of ecologicalintegrity required to maintain endangered salmonand other stream biota, physical stability is likely onenecessary component of a healthy stream.

The purpose of this study was to determinewhether eroding channels in urban watersheds arecapable of restabilizing over a period of years todecades. Based on a variety of field and historicaldata from a range of streams draining urban andurbanizing watersheds in the Puget Sound lowlandsof western Washington, we investigated the relation-ships between channel stability, watershed urbaniza-tion, and other factors that may also affect the timingor extent of geomorphic response.

BACKGROUND

Concern over the status of native salmon runs hasbrought renewed attention to the quality of urbanstreams in the Pacific Northwest. With fewer salmonreturning to the local streams to spawn, land man-agers have begun extensive programs to rehabilitateappropriate habitat that has been lost or degradeddue to urban development. Urbanization can haveprofound impacts on the physical, chemical, and

The deleterious influences of urbanization on thehydrology and geomorphology of small streams havebeen extensively explored and documented (Hammer,1972; Leopold, 1973; Arnold et al., 1982; Booth, 1990).The transition of a watershed from the natural,forested state characteristic of humid regions to apredominantly urban condition encompasses removalof vegetation and canopy, compaction of soils, creationof impervious surfaces, and alteration of natural

'Paper No. 99179 of the Journal of the American Water Resources Association. Discussions are open until August 1, 2001.'Center for Urban Water Resources Management, University of Washington, Box 352700, Seattle, Washington 98195 (present address for

Henshaw: Northwest Hydraulic Consultants, Inc., 16300 Christensen Road, Suite 350, Tukwila, Washington 98188) (E-Mail/Henshaw: [email protected]).

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 1219 JAWRA

Henshaw and Booth

drainage networks. These actions result in increasedsurface runoff and changes to sediment budgetswhich in turn commonly induce a geomorphicresponse that alters the cross-sectional geometry,channel morphology, profile, and planform of thestream. The emphasis of this work is on the specificresponse of channel cross-sectional form to watershedurbanization over time. Cross sections adjust rapidlyand are more directly influenced by discharge andsediment supply than are most other components ofchannel form and morphology, and they are alsoamong the most commonly collected data on streamchannels of any size and watershed setting (MacDon-ald et al., 1991).

Channel Stability

Although widely used in the scientific literatureand in engineering practice, the concept of "channelstability" is somewhat ambiguous, with individualauthors and disciplines offering different interpreta-tions of what constitutes a stable channel. This is par-ticularly true for urban streams, where perceptions ofstability (and especially instability) from a human orengineering perspective may be in direct conflict withmore traditional geomorphic views of a stable chan-nel. While most geomorphic definitions of stabilityattempt to account for the inherent variability ofchannel systems, engineers and public works man-agers have preferred to seek channels that are"unchanging in shape, dimensions, and pattern"(Schumm, 1977:131). This preference provides thejustification for the construction of channelized, con-crete-lined floodways but is unrealistic for naturalalluvial channels that are intended to support healthybiological communities.

The definition of a stable channel may best besummed up as one in which there is no progressiveadjustment in channel form (Schumm, 1977; Mont-gomery, 1994). Thus "stability" implies that channelform (in this case, cross-sectional shape and dimen-sions) is essentially constant, but not static, over adefined time scale and under a typical range of flowand sediment conditions.

Channel Response to Urbanization

In response to urban-induced changes in flow andsediment regimes (Wolman, 1967; Leopold, 1968; Hol-lis, 1975), streams are forced to alter their channels.In smaller watersheds, most channels are observedto enlarge, implying that increases in flood peakshave a greater influence than increases in watershed

sediment yield (Dunne and Leopold, 1978). Althoughsedimentation and channel contraction are sometimesobserved, particularly early in urbanization (Leopold,1973) or in very low-gradient systems (Odemerho,1992), the predominant responses reported in the lit-erature are channel erosion and enlargement.

Channel enlargement may occur through lateralerosion increasing channel width, bed erosion or (lesscommonly) overbank deposition increasing channeldepth, or some combination of these mechanisms. It isimportant to distinguish these types of moderatechannel expansion from catastrophic incision, where-in rapid bed degradation effectively separates the bedfrom the channel banks to create a nonalluvial, gully-like morphology. The latter process, described byBooth (1990) for western Washington streams, is lesscommon and is beyond the scope of this study.

Channel expansion in response to watershedurbanization has been observed in a variety of climat-ic and physiographic regions. Hammer (1972) relatedincreases in channel size to various land uses inurban and urbanizing watersheds in eastern Pennsyl-vania, finding enlargement ratios proportional to theincreases in mean annual flood determined byLeopold (1968). Hollis and Luckett (1976) found simi-lar, though less pronounced, responses in a number ofsmall catchments in southeast England. Urban-induced channel enlargement has also been reportedfrom New York and western Pennsylvania (Morisawaand LaFlure, 1979), north-central Texas (Allen andNarramore, 1985), New South Wales in Australia(Neller, 1988), and southwestern British Columbia(MacRae, 1997).

The magnitude and spatial extent of urban-inducedchannel enlargement is related to a number of factorsthat have been explored by individual authors. Theseinclude the type of development (Hammer, 1972) andits location in the watershed (Ebisemiju, 1989), theextent of stormwater drainage systems (Hammer,1972), bank composition (Roberts, 1989), and theextent of bank vegetation (Allen and Narramore,1985; Rowntree and Dollar, 1999).

Restabilization and Response Times

Although the initial effects of urbanization oncross-sectional geometry are well documented, longerterm channel response is less well understood.Assuming that stream channels are in a stable stateprior to watershed urbanization, the altered urbanhydrologic regime should disrupt that stability. Thisleads to the formation, at least temporarily, of thesorts of unstable channels that have been widelyreported in the literature (e.g. Leopold, 1973; Arnoldet al., 1982; Booth and Henshaw, in press). Existing

JAWRA 1220 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

Natural Restabilization of Stream Channels in Urban Watersheds

models of geomorphic response to changes in hydro-logic forcings (Graf, 1977; Wolman and Gerson, 1978;Bull, 1991; Simon, 1989) suggest that such disturbedchannels should adjust and develop a new, stableform over time, yet no systematic data have demon-strated this restabilization process in urban streamsystems.

Some prior studies have inferred that achievementof a stable, urban state occurs over a period of one toseveral decades of relatively constant land use in thewatershed (Hammer, 1972; Ebisemiju, 1989; Roberts,1989; Gregory et al., 1992), but others predict thatstream channels disturbed by urbanization willremain unstable indefinitely (Wolman, 1967; Arnoldet al., 1982). We have conducted this study to evaluatethe generic prediction of restabilization, and to estab-lish its timing in a particular geomorphic settingunder a type of disturbance, urbanization, that isincreasingly prevalent in the modern landscape.

Setting

The Puget Sound lowlands (PSL) of western Wash-ington extend from the foothills of the Olympic Moun-tains east to the Cascade Range foothills, and fromthe south end of Puget Sound north to the Canadianborder (Figure 1). The region is home to the majorityof the states population, including the cities of Seat-tle, Tacoma, Everett, and Olympia, and it is the cen-ter of rapid urban growth in the Pacific Northwest.

The PSL experience a cool maritime climate char-acterized by mild, wet winters and warm, drier sum-mers. Seattle receives an annual average of 97 cm ofprecipitation, most of which falls as rain. The rainyseason extends from October through May, with over85 percent of annual precipitation falling during thatperiod.

ANADAlinyham N

Il

W*E

Figure 1. Puget Sound Vicinity and Watershed Location Map.


STUDY AREA

?rett

C

0,

0U

U

0.

IGTON

—

Henshaw and Booth

The geology arid topography of the PSL reflect theinfluences of past glacial activity. Valleys are typicallyfloored in moderately permeable outwash, depositedby rivers and streams draining from the last conti-nental ice sheet about 15,000 years ago. These uncon-solidated deposits provide an essentially limitlesspotential source of coarse sediment to modernstreams. Upland plateaus are most commonly under-lain by thin and relatively impermeable glacial till, ahighly compacted and resistant substrate. Due to itsconsolidated structure, till has low infiltration rates,and where the overlying soil layer has been clearedfor development, surface runoff is generated witheven low-intensity rainfall.

In addition to large rivers that drain the flankingmountains, the region's rolling topography supportsan abundance of smaller streams. Typical PSLstreams drain small watersheds (up to a few tens ofkm2) at moderate gradients (0.5-3 percent slope).These small streams have historically providedimportant anadromous salmonid habitat and conse-quently have become a focal point in the response tothe recent Endangered Species Act listing of severaldwindling Puget Sound salmon runs.

At the present time, local stormwater regulationshave focused on controlling runoff to these streamsprimarily via retentionldetention facilities to limitpeak flow magnitudes. Booth and Jackson (1997)found that the existing regulations have been largelyineffective in protecting stream channels during theperiod of watershed urbanization, suggesting thatnew approaches to stream and watershed manage-ment are necessary.

Stream Descriptions

The study streams (Figure 1) were chosen to repre-sent typical low-order PSL streams. Their watersheds

encompass a range of development intensities (50 to> 90 percent developed area) and representative topo-graphic and geologic characteristics. Four primarystreams (May, Juanita, Kelsey, and Thornton Creeks)were the subject of detailed field study, consisting ofcurrent stability assessment at two or three alluvialsites along each channel. Field sites on three addi-tional streams (Madsen, Joes, and McAleer Creeks)provided a supplemental record of long-term channelmonitoring. These three secondary streams wereselected from a group of about 30 PSL streams thathave been surveyed on an annual to biannual basisover the past decade (Booth and Henshaw, in press),because they correspond most closely in gradient andwatershed area to the primary streams.

The level of urban development in each of theseseven watersheds was used to define the basins aseither "transitional," where development is ongoing orhas only recently stopped, or "post-development,"meaning that land cover has been essentially con-stant for at least the past decade. Watershed charac-teristics are summarized in Table 1.

In addition to these seven streams, the stability ofchannels in five less developed PSL basins — Bear,Jenkins, Upper Little Bear, Rock and Soos Creeks —was evaluated to broaden the range of urbanizationlevels covered by the seven study streams. Developedarea in these watersheds ranges between about 40 to70 percent, with much of the land cover change hav-ing occurred within the last ten years. Data from mul-tiple surveys of cross-sections on Boeing Creek,reported by Booth and Henshaw (in press), providedadditional information about restabilization responsein a post-development watershed following a majordisturbance.

TABLE 1. Watershed Characteristics.

CreekDrainage

Area (kin2)BasinSlope

DevelopmentLevel

DominantSurface Geology

StudyEmphasis

May 33 0.008 Transitional Till, Bedrock PrimaryMadsen 5 0.030 Transitional Till SecondaryJoes 8 0.028 Transitional Outwash, Till SecondaryJuanita 16 0.012 Transitional Outwash, Till PrimaryKelsey 26 0.010 Post-Development Till, Outwash PrimaryMcAleer 20 0.013 Post-Development Till SecondaryThornton 31 0.013 Post-Development Till Primary


Natural Restabiization of Stream Channels in Urban Watersheds

Approach

METHODS

Channel stability and levels of watershed urban-ization for each study stream were assessed to exam-ine the relationship between channel stability and themagnitude and rate of development in the watershed.Channel stability at each of the study sites was ana-lyzed using a combination of current observations andmeasurements and available historical recordsof channel dimensions, gathered from USGS cross-section surveys, rating curves and inspection reportsfrom nearby stream gages and bridges, and previousstudies. However, given the general lack of historicaldata, only current stability could be directly deter-mined in most cases. Comparison with decade-longrecords of channel form for the four secondarystreams provided an opportunity to assess the utilityof these short-term measurements as indicators oflong-term channel stability.

Hydrologic Data

Precipitation and discharge data were obtained toestablish a hydrologic context within which the water-shed and channel changes were occurring. Dailyprecipitation records for 1949-1997 were obtainedfrom the National Weather Service rain gage at theSeattle-Tacoma International Airport (Sea-Tac).Using these data, the maximum five-day precipitationevent was determined for each water year to assesschanges in precipitation over the period of record andover the ten-year period of observation for the sec-ondary streams. The five-day maximum event wasused because in the PSL region, under natural condi-tions, the largest flows are generated by rain onalready saturated surfaces.

Daily discharge and annual maximum flow datawere obtained from USGS and King County streamgage stations on May (USGS 12119600/King County37A), Juanita (USGS 12120500/King County 27A),Mercer (USGS 12120000), and Thornton (USGS12128000) Creeks to examine changes in streamfiowand hydrologic regime over the past few decades ofurbanization. Mercer Creek flow data were used as asurrogate for Kelsey Creek; the Kelsey Creek water-shed accounts for roughly 70 percent of the contribut-ing area at the Mercer Creek gage. Recent USGSrecords from Thornton Creek were not used due to theinfluence of a high-flow diversion constructed justupstream of the gage location. Additional ten-minutedata from a Seattle Public Utilities gage farther

upstream on Thornton Creek (SPU 42) were substi-tuted for 1998.

In addition to trend analysis on mean annual andannual maximum flows, discharge records were usedto determine a measure of the "flashiness" of the con-tributing watershed (Konrad, 2000). Flashiness wasdefined as the percent of flows (daily or ten-minuteaverage, as available) that exceeded the mean flow forthe year. Low values for this metric indicate a flashyhydrograph with short-duration peaks, whereas high-er values would occur in watersheds that producemore sustained peaks. Comparisons of flashinessbetween basins reflect differences in geology and landcover, both of which affect runoff generation and flowrouting.

Field Data

Field data were collected in the fall of 1998 andspring of 1999. Cross sections and sediment sizeswere measured at 10 sites on May, Juanita, Kelsey,and Thornton creeks (Table 2), where qualitative sta-bility indicators were also observed. Only observation-al data were collected for Joes, Madsen, and McAleerCreeks, supplementing the existing multi-year recordof cross-sectional measurements. Qualitative observa-tions were also made on Boeing Creek to supplementa two-year monitoring record of cross-sectionalresponse to a catastrophic sediment inflow (Booth andHenshaw, in press).

Site Selection. Sites on the four primary channelswere chosen on the basis of their capacity for alluvialresponse and their location relative to other availabledata to establish a context of historical channelchange. Field sites were located in straight, plane-bedreaches (Montgomery and Buffington, 1997) to limitthe effects of local geomorphic influences (such asmeander bends or riffle-pool sequences) on observedchannel stability. Observations were made in straightreaches to avoid confusing urban-induced instabilitywith natural processes of erosion and channel migra-tion in bends. Although this approach will miss lowlevels of increased channel erosion occurring only atthe outside of bends, the great range of observedchannel instability suggests that this limitation is notparticularly significant. Channelized and confined(i.e., riprapped) reaches were not considered becauseof the imposed restrictions on cross-sectionalresponse, which limited the number of potential studysites, particularly on the more urbanized streams.Sites below tributary junctions, pipe inflows, or wet-land reaches were also eliminated because of theadded complexity at these locations.


Henshaw and Booth

TABLE 2. Site Measurements.

Site

DrainageArea(kin2)

Median BedParticle

(mm)

WaterSurfaceSlope

Bankfull Widthat Cross Section

(m)

Bankfull FlowCross-Sectional

Area (m2)

May Creek

MlM2

M3

33

32

22

334727

0.0080.007

0.002

6.16.0

8.6

2.51.4

5.4

Juanita Creek

JiJ2J3

16

15

14

12

11

13

0.0007

0.002

0.002

4.25.75.4

3.7

3.5

2.3

Kelsey Creek

KiK2

208

43

140.002

0.0015.8

2.32.1

1.0

Thornton Creek

TiT2

31

324<2

0.0100.0004

5.25.3

2.51.7

Although grade adjustment is also a component ofalluvial response, slope typically adjusts more slowlythan channel dimensions and can be considered anindependent variable in geomorphic response overshorter time scales (Ferguson, 1986). Consequently,sites affected by downstream grade controls were notsummarily eliminated from consideration. Of theselected sites, the Joes Creek and Lower JuanitaCreek (Ji) sites are affected by rigid downstreamgrade controls. The Lower Kelsey Creek (K1) and MayCreek sites (Ml) are within 100 m of downstreambridges but are believed to be beyond the hydraulicinfluence of the structures.

Bank Conditions. On the four primary streams, arepresentative cross section at each site was surveyedto determine current channel dimensions. The bank-full channel boundaries were estimated by noting thelower edge of perennial vegetation and transition inbank slope from near-vertical to near-horizontal(Williams, 1978; Olsen et al., 1997).

To supplement quantitative field measurements,sites were grouped into four stability categories byvisual observation of indicators such as bank erosionand vegetation extent (Table 3 and Figure 2). The cat-egories were modeled after the bank stability portionof Gallis (1996) rapid stream assessment technique(RSAT). Bank stability categories at several of thesites were determined independently by two field

observers, and the assessments proved to be bothrapid and reproducible. Ratings were determined forthe five less urban streams (Konrad et al., 1998) inaddition to the seven study channels.

The bank stability ratings for sites on Madsen,Joes, and McAleer Creeks were compared with multi-pie cross-sections previously surveyed by Booth andothers over the past decade (Booth and Henshaw, inpress) to test the observational technique as an indi-cator of long-term channel instability.

Bed Conditions. Pebble counts were conducted todetermine the bed particle size distribution at eachsite. The pebble count procedure (after Wolman, 1954)consisted of random selection and measurement of theintermediate diameter of 100 bed surface particlesfrom evenly distributed locations across the reach.Particle sizes were used to classify the bed materialand to calculate bed shear stresses for evaluating bedmobility.

The relative bed stability (RBS) index, described byOlsen et al. (1997), defines bed stability as a ratio ofthe critical bed shear stress required to mobilize theD84-size particle (1c84) to bed shear stress at banklullflow (tbf). The critical bed shear stress is calculatedusing the form of the Shields equation in Olsen et al.(1997):

tc84 = (p5 - (1)


Class Description

W STABLE• Perennial vegetation to waterline• No raw or undercut banks (some erosion on outside of meander bends OK)• No recently exposed roots• No recent tree falls

III SLIGHTLY UNSTABLE• Perennial vegetation to waterline in most places• Some scalloping of banks• Minor erosion and/or bank undercutting• Recently exposed tree roots rare but present

MODERATELY UNSTABLE• Perennial vegetation to waterline sparse (mainly scoured or stripped by lateral erosion)• Bank held by hard points (trees, boulders) and eroded back elsewhere• Extensive erosion and bank undercutting• Recently exposed tree roots and fine root hairs common

COMPLETELY UNSTABLE

• No perennial vegetation at waterline• Banks held only by hard points• Severe erosion of both banks• Recently exposed tree roots common• Tree falls and/or severely undercut trees common

where p and p represent sediment and water densi-ties, respectively, and g is the acceleration of gravity.The dimensionless shear stress, t, is related to thebed particle size distribution by:

= (2)

where 0 and x are empirical values. Values of 0.045 (0)and 0.7 Cx) were used in this study after Olsen et al.(1997), who based these on data from Komar (1989).The bed shear stress at bankfull flow is determinedas:

Tbf = pgRS

where R is the hydraulic radius and S is the watersurface slope. The relative bed stability index is thendefined as:

RBS Tc84 / tbf

If the RBS is greater than 1, the bed is presumedto be fully mobilized only for events larger than bank-full and the channel is relatively stable. Conversely,if RBS is less than 1, the bed is mobilized at sub-bankfull flows and the channel is presumably unsta-ble.

Historical Data

Historical channel information was collected tohelp establish patterns of change over time and toassess long-term channel stability. Unfortunately, thehistorical records documenting stream channel condi-tions or changes were limited and generally of littlevalue for analysis or comparison. The most useful his-torical data consisted of previously measured crosssections, USGS rating curves, and bridge inspectionrecords.

(3) Land Cover

Changes in developed area were characterized fromaerial photographs (1970, 1971, 1980, and 1988) andpreviously classified 25-meter resolution Landsat

(4) Thematic Mapper images (1991 and 1995) for eachwatershed. For the Landsat images, training sites ofrelatively uniform land cover were selected fromaround the region to determine the reflectance for dif-ferent types of vegetative and urban ground cover.The Signature Editor module in ERDAS's "Imagine"software (version 8.4) was then used to extract spec-tral signatures for each training site, which in turn



TABLE 3. Streambank Stability Classification Criteria.

II

I

Henshaw and Booth

Figure 2a. Stable Reach (IV). Vegetation extends to thewaterline along the entire bank, and there is no

evidence of erosion (Kelsey Creek, Site K2).

Figure 2b. Slightly Unstable Reach (III). Vegetation extendsto waterline, but banks are slightly scalloped and grassroots indicate recent erosion (Juanita Creek, Site Ji).

were used to classify the entire Landsat image (Bots-ford et al., 1999). Using the 1991 classified image, weselected random classified pixels to compare with low-elevation 1991 orthophotos for validation of the classi-fied image. The categories, and their correspondingaverage land cover percentages, are given in Table 4.While these results generally support the use of thismethod for our broad categorization of developed andundeveloped areas, the variability within the classi-fied land-cover categories suggests that the classifica-tion scheme may not be reliable for more specificdistinctions between individual land covers.

Percent developed area consisted of residential,commercial, and industrial land uses and included all

Figure 2c. Moderately Unstable Reach (II). Minimal vegetationat the waterline and the left bank (right in picture) is

scalloped back except at hard points (Juanita Creek, Site J3).

Figure 2d. Completely Unstable Reach (I). Minimal bankvegetation and nearly vertical, severely eroded banks. Theright bank (left in picture) has been eroded back far enough

to almost undermine the large tree (Thornton Creek, Site Ti).

grassy areas, which Wigmosta and Burges (1997) andBurges et al. (1998) found to contribute a large por-tion of runoff in developed catchments. Results of theland-cover analyses are shown in Table 5, along withthe percent change in developed area between 1991and 1995 and between 1988 and 1995. Changes indeveloped area are shown for each watershed and forthe drainage area above each study site on the prima-ry streams. We estimate that differences of one or twopercent are not significant because of minor differ-ences in the definition of land-cover categoriesbetween the classification schemes for the 1991 and1995 land-cover data.



TABLE 4. Land-Cover Percentages of Developed and Undeveloped Watershed Areas.

Categories From the ClassifiedLandsat Image

Actual Land Cover FromOrthophotos (percent)OpenWater Trees

ShrubslGrass

Pavement orBare Earth

UNDEVELOPED

Open WaterConiferous VegetationDeciduous Vegetation

100 0— 91— 47

0849

01

4

DEVELOPED

Grassy/Shrubby VegetationForested Urban

— 87 39

6331

2923

Grassy UrbanIntense Urban

— 89 8

6121

3162

TABLE 5. Percent Developed Area Over Time.

Creek/Site 1970 1971Percent Developed

1980 1988 1991 1995Percent Change

1991-1995 1988-1995

May 23Ml —

M2 —

M3 —

—

—

—

—

36 37 42— — 42— — 41— — 32

50504941

19.0 35.119.019.528.1

Juanita 52Ji —

J2 —

J3 —

—

—

—

—

79 83 88— — 88— — 89— — 89

89899090

1.1 7.21.11.11.1

Kelsey —

Ki —

1(2 —

79—

—

83 82 84— — 85— — 86

858687

1.2 3.71.21.2

Thornton —

Ti —

T2 —

98—

—

98 98 97— — 97— — 96

979797

0.0 -1.00.01.0

Joes — — — 80 87 90 3.4 12.5

Madsen — — — 67 67 70 4.5 4.5

McAleer — — — 92 92 92 0.0 0.0

Bear — — — — 54 58 7.7 —

Jenkins — — — — 56 69 24.3 —

Little Bear — — — — 46 66 43.4 —

Rock — — — — 37 39 5.1 —

Soos — — — — 60 73 21.9 —

Hydrology

RESULTS

Rank-sum analysis (Helsel and Hirsch, 1992) of theannual maximum flood series from the May, Juanita,and Mercer Creek stream gages showed significant

increases in peak flows over the latter part of therecord on Mercer and Juanita Creeks, at 99 percentand 95 percent levels, respectively. May Creek, theleast developed of the study basins, did not show asignificant change in annual maximum discharge.These results are consistent with those reported byMoscrip and Montgomery (1997) for the samestreams. There were no accompanying changes in


Henshaw and Booth

180

16,0

140E

12.0a.U

10.0a.

8.09

6.0

40

20

..

•.

..

. ..

.. . ,—

. .. .

• ••.. . •. .•

.• •

..••

.•

. .... •

001940 1950 1960 1970 1980 1990 2000

Water Year

Figure 3. Percent of Gage Readings Above Mean Annual Flow.Lower values indicate flashier hydrographs. Thornton A is the

USGS gage station and Thornton B is the Seattle Public Utilitiesgage (Sources: USGS, King County, Seattle Public Utilities).

The annual five-day maximum rainfall from theSea-Tac precipitation record (Figure 4) was alsoanalyzed for trends using the Mann-Kendall test.There was no significant change in the five-day maxi-mum over the period of record nor over the decadespanned by the secondary stream monitoring, sug-gesting that increases in runoff or peak flow were notclimate-driven.

Channel Stability

Overall stability was determined for each channelas an aggregate of its available stability measures;with one exception, all measures indicated the samerelative stability in each watershed. Stability assess-ments for the four primary study streams were basedsolely on measures of current stability (the RBS indexand bank stability class), whereas the secondarystreams were assessed based on a combination of

Kelsey Creek. Kelsey Creek offers an example ofa highly developed watershed whose channel has pre-sumably undergone a period of change and has nowrestabilized. Survey results and bridge inspectionreports for a bridge about 50 m downstream of Kisuggest that both Kelsey Creek sites are currentlystable, and that channel change has been minimal forat least the past three years. The level of developmentin the watershed has been essentially constant since1980, following rapid development in the 1960s and1970s.

Thornton Creek. Despite having experienced vir-tually no change in land cover in the past threedecades, Thornton Creek is the least stable of thestudied channels, with the two field sites classified as"completely unstable" and "moderately unstable."Both the bed and banks are unstable; the channel atthe upstream site (T2, drainage area 3 km2) isextremely overwidened with a bankfull width virtual-ly the same as at site Ti (drainage area 31 km2).

Juanita Creek. Juanita Creek represents a tran-sitional basin where development has only recentlyslowed, so we speculate that the channel may still beadjusting to land-cover changes in the watershed.The channel displays recent instability, but there is


bank stability class and long-term channel behavior.Only bank stability data (Konrad et al., 1998) wereused for the five additional streams. In general, chan-nel stability evaluations show no systematic patternof channel stability with the magnitude, or the rate,of development in the watershed (Table 6 and Figure5).

mean annual flow for any of the basins. Records fromThornton Creek were not used for this analysis due toinsufficient record length.

Flashiness data (Figure 3) were analyzed forchange over time using the Mann-Kendall test (Helseland Hirsch, 1992). Both Mercer and Juanita Creeksexperienced increasingly flashy discharges, as shownby decreasing trends in the value of the metric, signif-icant at the 99 percent level. There was no significantchange in flashiness of the May Creek discharge, andtrends were not evaluated for Thornton Creek.

0.10

0.15 'C

0.20ojA

0.25 A £L A

A An4i. *A OA A0.30 AA 8

0.35 £ A . •0.40 0

0.45

1940 1950 1960 1970 1980 1990 2000

Water Year

• Intoy Juanita A tutorcer U Thornton A Ic Thornton B

Figure 4. Maximum Five-Day Precipitation by Water Year(Source: National Weather Service, Sea-Tac station).


TABLE 6. Summary of Results.

Land Cover Channe1 Stability

Creek*

DevelopedArea

(percent)(1995)

PercentChange

(1991-1995)

Current(average of stations)

Long-Term OverallBank Bed

(P1 = stable) (>1 = stable)

Thornton (1) 97 0 I-Il 0.6 n/a Completely Unstable

McAleer (2) 92 0 N n/a Stable Stable

Kelsey (1) 85 1.2 N 5.1 nla Stable

Juanita(1) 89 1.1 lI-Ill 1.8 n/a ModeratelyUnstableJoes (2) 90 3.4 III n/a Slightly Unstable Slightly Unstable

Madsen (2) 70 4.5 IV n/a Slightly Unstable Stable

Rock (3) 39 5.1 N n/a n/a Stable

Bear (3) 58 7.7 III n/a n/a Slightly Unstable

May (1) 50 19 Ill-N 2.6 n/a Slightly Unstable

Soos (3) 73 22 III n/a n/a Slightly UnstableJenkins (3) 70 24 N n/a n/a Stable

Little Bear (3) 66 43 IV n/a n/a Stable

*Study emphasis for each creek indicated as (1) primary, (2) secondary or (3) other.

00

I')1C)C,

C,

C-)

CUI-a-

0 Stable XSlightly Unstable £ Moderately Unstable •Completely Unstabl,e

Figure 5. Stability as a Function of Developed Area and Change in Development. Results are shown for eachfield site on May, Juanita, Kelsey, and Thornton Creeks and for the full watershed for other streams.

no basis to predict whether instability will persist, aswith Thornton Creek, or whether the channel willrestabilize, as with Kelsey Creek.

USGS cross section surveys at the Juanita Creekgage site indicate an increase in width sometime in

the 1960s, followed by consistent values through theend of the record in 1990 (Figure 6). Although thedata gap precludes further analysis of the path of thechannel adjustment (e.g., gradual linear increase ver-sus step change), these data do provide confirmation


450

40

35

300

25x

20 XX

15

10x

5 x0 X&o

30 40 50 60 70 80 90 100

Developed Area (%)

Henshaw and Booth

of the expected response of channel enlargement fol-lowing the onset of significant urbanization. The lackof change in channel width in the last 10 to 15 yearsof recorded cross sections is difficult to interpret, asthe limited response of the confined lower banks atthe gaging site may not accurately reflect conditionsin the fully alluvial reaches.

. .

••*1.0

0.0

the past decade. Results were quite consistent forthree of the four sites (the three secondary studystreams plus an additional site on a tributary toMcAleer Creek), with conditions of slight or no insta-bility readily identified by both long-term and short-term methods. The only exception was Madsen Creek,which was classified as "stable" based on bank stabili-ty observations despite significant measured channelchange over the past ten years (Figure 7). The sourceof this disparity is clear from the cross section: thesite experienced significant aggradation between the1989 and 1990 surveys, a result of massive erosionabout 400 m upstream during a large storm in Jan-uary 1990, and has since undergone bed degradation.The banks, however, restabilized more rapidly andhave remained essentially unchanged for the last fiveyears of the monitoring period. The channel adjust-ment in the long-term record is a result of a sedimentinflow from a discrete upstream event, rather thaninstability intrinsic to the site itself.

j1\ \\O'

Date

Figure 6. Flow Widths Measured at USGS Station 12120500on Juanita Creek, 1945-1990 (Source: USGS).

May Creek. Despite rapid development of thewatershed over the last decade, field observations andmeasurements indicate at most slight instability, andthat only in the lower reaches, of May Creek. Crosssection measurements made by the USGS in 1942(near Ml) and 1958 (near M2) showed flow widthsless than those determined from the current surveys.In the absence of a reference width (such as bankfull)or stage-discharge relation for comparison, however,there are insufficient data to evaluate the magnitudeof any channel enlargement over the last fourdecades.

Short-Term Versus Long-Term Channel Stability

The most accurate way to determine channel sta-bility is through long-term monitoring of one or morecross sections selected specifically for this purpose.However, in nearly all cases where such data areneeded for geomorphic assessment or for managementaction, this approach is infeasible. We therefore usedour three secondary streams to determine whethershort-term observations can accurately reflect long-term stability, by comparing current observed stabili-ty with the amount of cross-sectional adjustment over

Figure 7. Cross Sections of Madsen Creek Upstream From SR 169,1988-1998(1998 survey provided courtesy of R. W. Beck).

A two-year record of cross-sectional surveys forseven cross sections on Boeing Creek (Booth and Hen-shaw, in press) shows that all but one of the cross-sections restabilized within a year following acatastrophic debris flow in late December 1996, withtwo of the sections achieving a new stable form withinthree months of the event. All seven cross-sectionswere categorized as "stable" or "slightly unstable"using the bank stability classification scheme.


8.0

7.0

6.0

5.0

40

3.0

2.0

EC0

>a,w

0 2 4 6 8 10 12

Distance from Left Bank Datum Im)

H—*—1988 —44—1989 —ó—I990 —0-- 1993 —.._I998j


DISCUSSION

Urbanization and Channel Stability

Urbanization acts as the impulse for the hydrologicchanges that typically cause channel instability indeveloping watersheds. Once a watershed is devel-oped, however, does the urban setting continue to playa dominant role in channel response, or do the hydro-logic and geomorphic characteristics of the channeland watershed reassert control? Results of the cur-rent study suggest that the extent of channel instabil-ity and the likelihood of channel restabilization arenot determined strictly by the magnitude of distur-bance (i.e., the amount or rate of urbanization) butinstead depend jointly on the degree of urbanizationand the responsiveness of the channel and watershedto land-use change.

The level of urbanization in a watershed exerts atmost a coarse effect on the likelihood that the streamchannel will be stable, and the rate at which urbandevelopment is occurring shows no systematic influ-ence at all (Figure 5). In this study, channels in theless developed watersheds were stable or only slightlyunstable, and pronounced instability was observedonly in the most developed watersheds. Yet PSLwatersheds at even very low levels of developmentcan display major channel adjustments (Booth, 1990;Booth and Henshaw, in press), indicating that thelevel of channel response is more likely controlled byconditions in the channel and watershed other thangeneral levels of contributing urban land cover. Sim-ple correlations will not be universally applicable,even within the same region.

The circumstances for observed channel stabilityare not uniform. Stable channels in developingbasins (e.g., May, Bear, Jenkins, Upper Little Bear,Soos, and Rock Creeks) are not "restabilized," becausethere is no evidence or record of prior channel insta-bility and because land cover is still in such activeflux. Their stability is presumably a continued expres-sion of only gradual divergence from pre-urbanizationconditions. Conversely, the stable channels that havebeen affected by several decades of substantial water-shed urbanization (e.g., McAleer, Kelsey, and Madsen)almost certainly reflect restabilization followingurban disturbance, rather than mere persistence ofthe pre-development channel form.

Of our presently unstable channels, the prognosisfor future restabilization is uncertain. Theoreticalconcepts of the "equilibrium channel" (Leopold et al.,1964) and "graded stream" (Mackin, 1948), and multi-ple examples of stable urban channels, suggest thatsuch an outcome might be expected. Those stillexperiencing changes in watershed land cover (e.g.,

Juanita and Joes Creeks) may restabilize in time. YetThornton Creek remains highly unstable despite anurbanization level that has been virtually unchangedfor more than two decades, suggesting no reason toanticipate restabilization here in the near future.Again, a simple rule will not yield a universally cor-rect prediction.

Other Factors Controlling the RestabilizationResponse

If channel stability in developed watersheds isdetermined only weakly, if at all, by the magnitude orrate of land-cover change, other factors must play sig-nificant roles in determining whether a channel willdestabilize and/or restabilize over time. The potentialfor restabilization, in particular, depends on howchanges in the watershed affect the channel's flowand sediment regimes, and to what extent the chan-nel is capable of responding to these changes. Rele-vant factors must therefore exist at both thewatershed and channel scales.

Watershed-Scale Factors. The hydrologic regimeexperienced by the channel is closely tied to the levelof urbanization and impervious area in the water-shed, particularly where the underlying geologic sub-strate does not have a large infiltration capacity, as isthe case with the glacial till underlying parts of thestudy basins. Increased surface runoff in humid-cli-mate urban watersheds alters the natural subsurfaceflow regime, leading to larger, shorter-duration hydro-graph peaks. These changes affect streams by creat-ing more frequent competent flows with greatercapability for channel alteration. If the discharge pat-tern becomes too flashy, typical inter-storm flows can-not rework all of the sediment moved by the muchlarger storm discharges (Booth, 1991). Thus, channelform becomes dominated by the large events, and thechannel loses its ability to develop a stable formadjusted to smaller, more frequent discharges (e.g.,Leopold et al., 1964).

Thornton Creek is the prototype for such condi-tions. Despite draining the most established andunchanging watershed, it was the least stable of thechannels observed. With a 97 percent developedwatershed with essentially no detention or runoff-attenuation capability, Thornton Creek produced theflashiest of the gaged discharges. Discharge exceededmean annual flow less than 15 percent of the time,compared with 25 to 30 percent for Juanita, Mercer,and May Creeks.

In contrast, May Creek, which is stable to onlyslightly unstable, has not experienced the changes inpeak flows or flashiness typically seen in largely


Henshaw andBooth

urbanized basins. Compared to pre-development con-ditions, discharges for the two-year event haveincreased by relatively modest amounts, ranging from15 to 50 percent along the mainstem to only 5 to 20percent in the highland tributaries (King County,1995). Runoff generation mechanisms here are alsomore consistent with the natural condition, with thelargest peaks generated by multi-day storms ratherthan the one-day events that now generate the largestflows in the more urban Juanita and Mercer Creekbasins.

In addition to the hydrologic regime, topographicgradients might also be expected to influence channelstability. Yet across the range of slopes in the studywatersheds and streams, neither slope nor streampower showed any apparent relationship to observedchannel stability. Booth and Henshaw (in press)reported a similar lack of correlation across an evenwider range of channel gradients (from less than 0.01to 0.5).

Channel-Scale Factors. Several site-specific con-ditions contribute to the responsiveness of channels toland-use change. Most fundamentally, the erosionalcharacteristics of the geologic substrate surroundingthe channel can influence the extent to which thechannel form initially responds to watershed urban-ization, which in turn will influence how much adjust-ment is necessary to establish a new stable form.

The dominant geological materials in the PSL havequite contrasting properties: glacial till is a highlyresistant and relatively impermeable substrate,whereas glacial outwash is a less consolidated andmuch more erodible deposit. Outwash channels areparticularly prone to channel enlargement and insta-bility. Booth (1990) and Booth and Henshaw (in press)found that virtually all cases of extreme incisionoccurred in channels flowing over these deposits.Where less dramatic channel changes are evaluated,the pattern is less definitive but still consistent: slightto moderate instability is displayed by the outwashchannels in the present study, Juanita and Joescreeks.

In contrast, if the natural channel substrate ishighly resistant to erosion then the initial extent ofchannel response, and the difficulty of subsequentrestabilization, should be limited. The best exampleamong the study streams is found in the lower reach-es of McAleer Creek, which flow over extremely resis-tant fossilized peat beds. Even highly erosive,sediment-poor flows can produce only modest changesin this substrate, which helps to account for theextremely stable conditions observed on McAleerCreek despite its highly urbanized watershed.

Yet 'human actions at the channel scale can alternatural tendencies. On Thornton Creek, bank armor-ing and upland development have eliminated mostsources of coarse sediment (Trimble, 1997; Nelson,1999), resulting in a finer bed particle size distribu-tion than the other study streams despite high flows.The disruption of coarse sediment supply presumablyyields a sediment imbalance, forcing the stream tocontinue eroding any still-exposed banks and prevent-ing restabilization in the relict alluvial reaches.

Although channel gradient itself has no apparentrelation to channel stability, the extent of local con-trols on the gradient does appear to have some influ-ence. This condition is particularly relevant to urbanstreams, with their typically high densities of roadand utility crossings. Gradients just downstream ofthe study sites on Joes Creek and Lower JuanitaCreek (Ji) are rigidly controlled, limiting the abilityof the channels to incise. Both sites are only slightlyunstable, whereas the two upstream sites on JuanitaCreek, which are subject to less rigid and more widelyspaced grade controls, are both moderately unstable.Gradients on McAleer Creek, the most stable of thestreams despite a 92 percent developed watershed,are tightly controlled by a series of culverts along thelength of the channel, along with numerous points ofhighly resistant substrate.

In addition to the localized effects of substrate andgrade control on stability response, the extent of theriparian corridor may also influence responsiveness atthe channel scale. Although this factor was notexplored in detail, the most stable channels in thisstudy are generally associated with reaches with rela-tively intact natural riparian vegetation. Preservationof a natural riparian corridor reduces the area ofland-cover modification and so lessens the degree ofhydrologic change; it also discourages direct distur-bance of the channel and stabilizes (or allows morerapid restabilization of) potentially erodible banks.May, McAleer, and Madsen Creeks have the mostextensive natural riparian corridors and were eitherstable or only slightly unstable, whereas the moder-ately to completely unstable channels (Juanita andThornton Creeks) have limited, if any, naturalstreamside vegetation.

Time Scale of Restabilization

The observed stability of many established urbanwatersheds indicates that channels can restabilizewithout intervention. Results from these studystreams suggest that this restabilization occurs overone to two decades. Development in the McAleerCreek basin ceased around the late 1960s, and long-term monitoring results show that the channel has



been stable for at least the last ten years; land use inthe Kelsey Creek watershed has been roughly con-stant since the late 1970s, and both stream sitesappear to be quite stable. Yet no single generalityapplies to all urban channels: Thornton Creek has notrestabilized at all, even after more than 20 years ofvirtually unchanging land cover.

Under fortuitous stream and watershed conditions,channel restabilization following a major disturbancecan occur very quickly. Booth and Henshaw (in press)reported on the reestablishment of a stable channelon Boeing Creek within months of a major debris flow.On Madsen Creek, the banks were found to haverestabilized within three years of a major storm in1990 (Figure 7). Thus restabilization can occur in farless than a decade, even while land cover continues tochange.

As a general guideline, we judge that PSL streamswill tend to restabilize within one or two decades fol-lowing urbanization. Despite differences in hydrology,geology, and the extent and patterns of developmentin our study watersheds, the consistency of resultssuggests broad relevance to the variety of urban chan-nel conditions in the region. Yet no single period ofadjustment applies universally to restabilizing PSLstreams because of important differences in the deter-mining hydrologic and geomorphic characteristics.Only a case-by-case evaluation of watershed andstream conditions can show whether a 10-year to 20-year interval is appropriate for a given site, and anysimple prediction of channel response based on water-shed urbanization is guaranteed to yield spuriousresults.

Utility of Alternative Stability AssessmentTechniques

We used two alternative techniques, based on fielddata and observations, to assess the current state ofchannel stability: (1) we classified bank stabilitybased purely on observational data in order to rapidlygroup streams into several categories, and (2) we usedOlsen et al.'s (1997) relative bed stability (RBS) indexto characterize channel stability based on sedimenttransport at bankfull flows. Both methods producedsimilar results (Figure 8), and although higher bankstability classes did not uniformly yield greater RBSvalues, both measures clearly increased in consort.Thus, either one of these metrics is probably a suffi-cient indicator of relative stability in most cases,although the two in combination offer some increas-ing degree of discrimination.

Although useful as an indicator of relative stabilitybetween sites, the RBS index tended to overpredict

absolute stability. Only two sites (both on ThorntonCreek) had RBS values less than the defined stabilitythreshold of 1, even though two other sites were clas-sified as moderately unstable and seven of the tensites showed at least slight instabilities. This suggestseither that motion of bed surface particles smallerthan the D84 may be better associated with instabilityfor channels in this setting, or that the coarse parti-cles are more mobile than the critical shear stressesdetermined from Olsen et al.'s (1997) chosen dimen-sionless shear stress parameters would indicate.Studies of bed motion in gravel-bedded channels haveyielded a wide range of plausible values for the con-stants in the dimensionless shear stress function(Bufflngton and Montgomery, 1997); assumption of amore equally mobile bed would give lower criticalshear stresses and consequently lower RBS values.Using the bankfull discharge as the index for sedi-ment transport may also be unwarranted, as there isno certainty that these urban channels are adjustedto a typical 1.5-year to two-year recurrence bankfulldischarge (e.g. Wolman and Miller, 1960; Carling,1988).

.

. .

. ... ,

,

Channel Restabilization

Hydrologic, field, and historical data were used toclassify the stability of channel cross sections in sevenwatersheds and, where sufficient data existed, todetermine the rate and extent of change in channelform over time. The results indicate:

1. Restabilization of urban stream channels in thePSL can, and commonly does, occur even in highly


8

7

Cl,

41. 3a41

0

Bank Stability Class

Figure 8. Comparison of Stability Assessment Techniques.

CONCLUSIONS

Henshaw and Booth

urbanized watersheds. However, the degree of stabili-ty is not well predicted by either the magnitude ofdeveloped area or the rate of recent development.

2. There is no generalizable formula for channelrestabilization. When, and if, an individual channelwill restabilize depends on a combination of hydrolog-ic and geomorphic characteristics of the channel andits watershed, beyond simply the magnitude or rate ofurban development. The hydrologic regime and geo-logic setting appear to be important controlling fac-tors; extent of grade control and condition of theriparian corridor likely play noteworthy, but lessinfluential, roles.

Management Implications

Although slight instability was observed in chan-nels whose watersheds were as little as 50 percenturbanized, and previous work has documented thepotential for rapid changes at even lower levels ofurbanization, significant channel instability wasobserved in this study only in the most urban (greaterthan 90 percent developed) watersheds. Thus, theoverall physical form of many PSL streams can with-stand high levels of urbanization pressure, even atrapid development rates, and will also tend to restabi-lize even if morphologic changes have occurred. Fromthe perspective of basin stewardship, this suggeststhat management response to watershed urbanizationshould not emphasize the immediate imposition ofchannel stabilization measures, particularly if urban-ization levels are low or the watershed is still develop-ing. In many of these cases, restabilization is likely tooccur even without direct intervention, and bank pro-tection can actually accelerate bed degradation in anunstable channel (Simon and Downs, 1995).Resources may be better utilized during this period toprotect local habitat and mitigate increased flows anddownstream sediment load, rather than for intensivebank stabilization. So-called "remedial' measuresmay result in permanent riparian alteration withpotentially greater long-term biological consequences.Maintaining the channel and its riparian corridor inas natural a state as possible should be emphasized,because sites with better-preserved natural ripariancorridors appear to have a greater tendency towardstability.

Yet some channels will not restabilize quickly, if atall. Channels in extremely urban basins are at pro-gressively higher risk, and those bedded in intrinsi-cally erodible sediment, in particular, are prone tochannel enlargement or even incision. The presence ofrigid grade controls has some positive influence onchannel stability, although such in-channel measuresare likely to reduce the quality of other desirable

channel functions (such as biological habitat or aes-thetics).

Re stabilization does not imply a return of the chan-nel to its natural state, and restabilization alone isnot a sufficient goal for protecting aquatic communi-ties. A restabilized cross section will typically be larg-er and less geomorphically complex than thepre-urbanization channel form. This change affectsaquatic biology through loss of habitat and alteredflow patterns, water velocities, temperatures, andorganic inputs (Booth et al., 1997; Homer et al., 1997;Karr and Chu, 1999). Changes in short-term sedimentflux and scour, which are not well described by thestability indicators used here, are also biologicallyimportant (Montgomery et al., 1996). Further assess-ment and subsequent rehabilitation will be requiredto restore the biological integrity of the stream evenafter geomorphic stability is achieved, and the successof such additional efforts is by no means assured.

ACKNOWLEDGMENTS

Thanks to Chris Konrad, Stephen Burges, and David Mont-gomery of the University of Washington for assistance and reviews.We also appreciate the constructive suggestions from three anony-mous reviewers. Support for the first author was provided by theUniversity of Washington Valle Fellowship and the National Sci-ence Foundation Graduate Research Fellowship. Additional projectfunding was provided by the EPA Waters and Watersheds Program,Grant R825284-01-0, and the Center for Urban Water ResourcesManagement at the University of Washington.

LITERATURE CITED

Allen, P. M. and R. Narramore, 1985. Bedrock Controls on StreamChannel Enlargement with Urbanization, North Central Thxas.Water Resources Bulletin 21:1037-1048.

Arnold, C. L., P. J. Boison, and P. C. Patton, 1982. Sawmill Brook:An Example of Rapid Geomorphic Change Related to Urbaniza-tion. Journal of Geology 90:155-166.

Booth, D. B., 1990. Stream-Channel Incision Following DrainageBasin Urbanization. Water Resources Bulletin 26:407-4 17.

Booth, D. B., 1991. Urbanization and the Natural Drainage System— Impacts, Solutions, and Prognoses. Northwest EnvironmentalJournal 7:93-118.

Booth, D. B. and C. R. Jackson, 1997. Urbanization of Aquatic Sys-tems: Degradation Thresholds, Stormwater Detention, and theLimits of Mitigation. Journal of the American Water ResourcesAssociation 30:1077-1090.

Booth, D. B. and P. C. Henshaw (in press). Rates of Channel Ero-sion in Small Urban Streams. In: Influence of Urban and ForestLand Use on the Hydrologic-Geomorphic Responses of Water-sheds, M. S. Wigmosta (Editor). AGU Monograph Series, WaterScience and Applications, Volume 2.

Booth, D. B., D. R. Montgomery, and J. Bethel, 1997. Large WoodyDebris in Urban Streams of the Pacific Northwest. In: Effects ofWatershed Development and Management on Aquatic Ecosys-tems, L. A. Roesner (Editor). Engineering Foundation Confer-ence Proceedings, pp. 178-197.



Botsford, E., K. Hill, and D. Booth, 1999. A Rapid Land Cover Clas-sification Method for Use in Urban Watershed Analysis (online).Center for Urban Water Resources Management, University ofWashington, Seattle, Washington. Available from: http://depts.washington.edu/cuwrm/ (Accessed July 15, 2000).

Buffington, J. M. and D. R. Montgomery, 1997. A Systematic Analy-sis of Eight Decades of Incipient Motion Studies, With SpecialReference to Gravel-bedded Rivers. Water Resources Research33:1993-2029.

Bull, W. B., 1991. Geomorphic Responses to Climatic Change.Oxford University Press, New York, New York.

Burges, S. J., M. S. Wigmosta, and J. M. Meena, 1998. HydrologicEffects of Land-Use Change in a Zero-Order Catchment. Jour-nal of Hydrologic Engineering 3:86-97.

Caning, P., 1988. The Concept of Dominant Discharge Applied toTwo Gravel-Bed Streams in Relation to Channel StabilityThresholds. Earth Surface Processes and Landforms 13:355-367.

Dunne, T. and L. B. Leopold, 1978. Water in Environmental Plan-ning. W. H. Freeman, New York, New York.

Ebisemiju, F. 5., 1989. Patterns of Stream Channel Response toUrbanization in the Humid Tropics and Their Implications forUrban Land Use Planning: A Case Study From SouthwesternNigeria. Applied Geography 9:273-286.

Ferguson, R. I., 1986. Hydraulics and Hydraulic Geometry.Progress in Physical Geography 10:1-31.

Galli, J., 1996. Appendix A — Final Technical Memorandum: RapidStream Assessment Technique (RSAT) Field Methods.Metropolitan Washington Council of Gevernments, Washington,D.C.

Graf, W. L., 1977. The Rate Law in Fluvial Geomorphology. Ameri-can Journal of Science 277:178-191.

Gregory, K. J., R. J. Davis, and P. W. Downs, 1992. Identification ofRiver Channel Change Due to Urbanization. Applied Geography12:299-318.

Hammer, T. R., 1972. Stream Channel Enlargement Due to Urban-ization. Water Resources Research 8:1530-1540.

Helsel, D. R. and R. M. Hirsch, 1992. Statistical Methods in WaterResources. Elsevier, Amsterdam, The Netherlands.

Hollis, G. E., 1975. The Effect of Urbanization on Floods of Differ-ent Recurrence Interval. Water Resources Research 11:431-435.

Hollis, G. E. and J. K. Luckett, 1976. The Response of NaturalRiver Channels to Urbanization: Two Case Studies From South-east England. Journal of Hydrology 30:351-363.

Homer, R. R., D. B. Booth, A. A. Azous, and C. W. May, 1997.Watershed Determinants of Ecosystem Functioning. In: Effectsof Watershed Development and Management on Aquatic Ecosys-tems, L. A. Roesner (Editor). Engineering Foundation Confer-ence Proceedings, pp. 251-274.

Karr, J. R. and E. W. Chu, 1999. Restoring Life in Running Waters:Better Biological Monitoring. Island Press, Washington, D.C.

King County Surface Water Management Division, 1995. MayCreek Current and Future Conditions Report. Seattle, Washing-ton.

Komar, P. D., 1989. Flow-Competence Evaluations of the HydraulicParameters of Floods: An Assessment of the Technique. In:Floods: Hydrological, Sedimentological, and GeomorphologicalImplications, K. Beven and P. Carling (Editors). John Wiley andSons, Chichester, United Kingdom, pp. 107-134.

Konrad, C. P., 2000. Hydrologic Disturbances in Puget LowlandStreams. Doctoral Dissertation, University of Washington, Seat-tle, Washington.

Konrad, C., S. Morley, D. B. Booth, S. Burges, and J. R. Karr, 1998.Influences of Urban Development on Stream in the Puget SoundLowlands, Washington. EOS, American Geophysical Union,Spring Meeting, Vol. 79.

Leopold, L. B., 1968. Hydrology for Urban Land Planning — AGuidebook on the Hydrologic Effects of Urban Land Use. Geo-logical Survey Circular 554.

Leopold, L. B., 1973. River Channel Change With Time: An Exam-ple. Geological Society of America Bulletin 84:1845-1860.

Leopold, L. B., M. G. Wolman, and J. P. Miller, 1964. Fluvial Pro-cesses in Geomorphology. W. H. Freeman, San Francisco, 522pp.

MacDonald, L. H., A. W. Smart, and R. C. Wissmar, 1991. Monitor-ing Guidelines to Evaluate Effects of Forestry Activities onStreams in the Pacific Northwest and Alaska. U.S. Environmen-tal Protection Agency Region 10, Water Division, Seattle, Wash-ington, EPA/910/9-91-001, 166 pp.

Mackin, J. H., 1948. Concept of the Graded River. Geological Soci-ety of America Bulletin 59:463-512.

MacRae, C. R., 1997. Experience From Morphological Research onCanadian Streams: Is Control of the Two-year Frequency RunoffEvent the Best Basis for Stream Channel Protection? In: Effectsof Watershed Development and Management on Aquatic Ecosys-tems, L. A. Roesner (Editor). Proceedings of an EngineeringFoundation Conference, ASCE, New York, New York, pp. 144-162.

Montgomery, D. R., 1994. What Does Channel Stability Mean? In:Proceedings of the International Workshop on the Ecology andManagement of Aquatic-Terrestrial Ecotones, G.L. Link andR. Naiman (Editors). University of Washington, Seattle, Wash-ington, pp. 37-41.

Montgomery, D. R., J. M. Buffington, N. P. Peterson, D. Schuett-Hames, and T. P. Quinn, 1996. Streambed Scour, Egg BurialDepths and the Influence of Salmonid Spawning on Bed SurfaceMobility and Embryo Survival. Canadian Journal of Fisheriesand Aquatic Sciences 53:1061-1070.

Montgomery, D. R. and J. M. Buffmgton, 1997. Channel Reach Mor-phology in Mountain Drainage Basins. Geological Society ofAmerica Bulletin 109:596-611.

Morisawa, M. and E. LaFlure, 1979. Hydraulic Geometry, StreamEquilibrium, and Urbanization. In: Adjustments of the FluvialSystem: A Proceedings Volume of the Tenth Annual Geomor-phology Symposia Series, Binghamton, New York, D. D. Rhodesand G.P. Williams (Editors). Kendall/Hunt, Dubuque, Iowa, pp.333-350.

Moscrip, A. L. and D. R. Montgomery, 1997. Urbanization, FloodFrequency, and Salmon Abundance in Puget Sound LowlandStreams. Journal of the American Water Resources Association33:1289-1297.

Neller, R. J., 1988. A Comparison of Channel Erosion in SmallUrban and Rural Catchments, Armidale, New South Wales.Earth Surface Processes and Landforms 13:1-7.

Nelson, E. J., 1999. A Sediment Budget of the Issaquah CreekBasin. Masters Thesis, University of Washington, Seattle,Washington.

Odemerho, F. 0., 1992. Limited Downstream Response of StreamChannel Size to Urbanization in a Humid Tropical Basin. Pro-fessional Geographer 44:332-339.

Olsen, D. S., A. C. Whitaker, and D. F. Potts, 1997. AssessingStream Channel Stability Thresholds Using Flow CompetenceEstimates at Bankfull Stage. Journal of the American WaterResources Association 33:1197-1207.

Roberts, C. R., 1989. Flood Frequency and Urban-induced ChannelChange: Some British Examples. In: Floods: Hydrological, Sedi-mentological and Geomorphological Implications, K Beven andP. Carling (Editors). John Wiley and Sons, Chichester, UnitedKingdom, pp. 57-82.

Rowntree, K. M. and E. S. J. Dollar, 1999. Vegetation Controls onChannel Stability in the Bell River, Eastern Cape, South Africa.Earth Surface Processes and Landfonms 24:127-134.


Henshaw and Booth

Schumm, S. A., 1977. The Fluvial System. John Wiley and Sons,New York, New York.

Simon, A., 1989. A Model of Channel Response in Disturbed Allu-vial Channels. Earth Surface Processes and Landforms 14:11-26.

Simon, A. and P. W. Downs, 1995. An Interdisciplinary Approach toEvaluation of Potential Instability in Alluvial Channels. Geo-morphology 12:215-232.

Trimble, S. W., 1997. Contribution of Stream Channel Erosion toSediment Yield from an Urbanizing Watershed. Science278: 1442-1444.

Wigmosta, M. S. and S. J. Burges, 1997. An Adaptive Modeling andMonitoring Approach to Describe the

Hydrologic Behavior of Small Catchments. Journal of Hydrology202:48-77.

Williams, G. P., 1978. Bank-Full Discharge of Rivers. WaterResources Research 14:1141-1154.

Wolman, M. G., 1954. A Method of Sampling Coarse River-BedMaterial. American Geophysical Union Transactions 35:951-956.

Wolman, M. G., 1967. A Cycle of Sedimentation and Erosion inUrban River Channels. Geografiska Annaler 49A:385-395.

Wolman, M. G. and R. Gerson, 1978. Relative Scales of Time andEffectiveness of Climate in Watershed Geomorphology. EarthSurface Processes 3:189-208.

Wolman, M. G. and J. P. Miller, 1960. Magnitude and Frequency ofForces in Geomorphic Processes. Journal of Geology 68:54-74.

JAWRA 1 236 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

The purpose of this study was to determinefaculty.washington.edu/dbooth/Henshaw and Booth...

Documents

Transcript of The purpose of this study was to determinefaculty.washington.edu/dbooth/Henshaw and Booth...