WATER RESOURCES RESEARCH, VOL. 36, NO. 10, PAGES 2957-2984, OCTOBER 2000
Regional interdisciplinary paleofiood approach to assess extreme flood potential
Robert D. Jarrett U.S. GeologicalSurvey, Denver, Colorado
Edward M. Tomlinson Applied Weather Associates,Monument, Colorado
Abstract. In the past decade,there has been a growinginterest of dam safetyofficials to incorporatea risk-basedanalysis for design-floodhydrology. Extreme or rare floods,with probabilitiesin therange of about10 -3 to 10-7 chanceof occurrenceper year, are of continuinginterest to the hydrologicand engineeringcommunities for purposesof planningand designof structuressuch as dams [NationalResearch Council, 1988]. The National ResearchCouncil stressesthat as much informationas possibleabout floods needsto be usedfor evaluationof the risk and consequencesof any decision.A regional interdisciplinarypaleoflood approach was developedto assistdam safetyofficials and floodplainmanagers in their assessmentsof the risk of large floods.The interdisciplinary componentsincluded documenting maximum paleofloods and a regionalanalyses of contemporaryextreme rainfall and flood data to complementa site-specificprobable maximumprecipitation study [Tomlinson and Solak, 1997]. The cost-effectiveapproach, which can be usedin many other hydrometeorologicsettings, was applied to Elkhead Reservoirin ElkheadCreek (531 km 2) in northwesternColorado; the regional study area Was10,900 km 2. Paleoflood data using bouldery flood deposits and noninundation surfaces for 88 streamswere used to documentmaximum flood dischargesthat have occurred during the Holocene.Several relative datingmethods were usedto determinethe age of paleoflooddeposits and noninundationsurfaces. No evidenceof substantialflooding was foundin thestudy area. The maximum paleoflood of 135m 3 s-1 for ElkheadCreek is about13% of thesite-specific probable maximum flood of 1020m 3 s-1. Flood-frequency relationsusing the expectedmoments algorithm, which better incorporatespaleoflood data, were developedto assessthe risk of extremefloods. Envelope curves encompassing maximumrainfall (181 sites)and floods(218 sites)were developedfor northwestern Coloradoto help define maximumcontemporary and Holoceneflooding in Elkhead Creek and in a regionalfrequency context. Study results for Elkhead Reservoirwere acceptedby the ColoradoState Engineerfor dam safetycertification.
1. Introduction tion of structures such as dams has included an estimate of the probablemaximum flood (PMF) [Cudworth,1989]. The PMF Worldwide,floods are amongthe mostdestructive events is an estimate of the maximum flood potential for a given related to meteorologicalprocesses. In the United Statesthe drainagebasin and is derivedfrom an analysisof the probable averageannual death toll of 125is accompaniedby about$2.4 maximum precipitation (PMP) [Cudworth,1989]. Prior to billion in damagesfrom floods [FederalEmergency Manage- about 1950, a variety of methods were used to obtain the ment Agency(FEMA), 1997]. Poor understandingof floods magnitudeof the designflood. The NRC [1988,1994] recog- contributesto unnecessaryloss of life and increasedflood nized (1) the limited hydroclimaticdata availableto estimate damagein somecases and, conversely,leads to costlyoverde- PMP/PMF values for mountain basins less than about 1050 sign of hydraulic structureslocated in floodplainsfor other km2, (2) thesubjectivity and variation of PMPestimates among situations[Jarrett, 1991, 1993; Baker, 1994]. Extreme or rare experiencedmeteorologists, (3) the criticalneed for regional floods,with probabilities in the rangeof about10 -3 to 10-7 chanceof occurrenceper year, are of continuinginterest to the analysesof extremeprecipitation and flooding, (4) theneed to use historicand paleoflooddata, and (5) the potentialuse of hydrologicand engineeringcommunities for purposesof plan- probability-basedmethods for providingan alternativeto the ning and designof structuressuch as dams[National Research PMP/PMF approach.In an evaluationof extreme floods and Council(NRC), 1988].However, estimating the magnitude and frequencyof extreme floods is difficult becauseof relatively the useof the PMF methodologyto estimatedesign floods, the short streamflow-gagingstation record lengths. InteragencyAdvisory Committee on WaterData [1986] raiseda For about the past 50 yearsthe designcriteria for construc- major concernabout PMFs beingeither dangerouslysmall or wastefullylarge, and they emphasizedthe importanceof accu- Copyright2000 by the AmericanGeophysical Union. rately estimatingthe risk of extremeflooding. Thus the objec- Paper number 2000WR900098. tive of a flood studyshould be to generateas much information 0043-1397/00/2000WR900098509.00 as practicableabout the flood potential at a site [NRC, 1988].
2957 2958 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
The NRC stresses that this should be the basis for evaluation haveoccurred in the regionsince the generalizedhydrometeo- of the risk and consequencesof any decision. rology report was published.Site-specific hydrometeorologic Efforts have been made to assigna frequencyto the PMF, studiesare being conductedbecause dam safetyofficials rec- but they are subjective[Interagency Committee on WaterData, ognizethe difficultproblems inherent in PMP estimatesin the 1986;NRC, 1988]. There is disagreementabout the range of Rocky Mountains.Utilization of an interdisciplinaryregional frequencyfor the PMF and whether any frequencyshould be paleofloodstudy provides additional supportinginformation assignedto a deterministicPMF estimate.Extensions of flood- for understandingthe magnitudeof the largestcontemporary frequencyrelations to rare floods(e.g., 10,000-yearrecurrence floodsand paleofloodswith estimatesof the PMF potentialfor interval) are tenuousfor short streamflowrecords but are a particularbasin. enhanced when paleoflood data are included in flood- frequencyanalysis [Kochel and Baker,1982; NRC, 1988;Jarrett, 1987; Jarrett and Costa, 1988; Levish et al., 1994; Ostenaaand 2. Background Levish,1995]. Jarrett and Costa[1988] proposed using regional Substantialuncertainty and controversyexists in estimating flood-frequencyestimates, w,hich incorporated paleoflood in- flood magnitudesand frequencies,particularly those of ex- formation, to estimate recurrence intervals for rare floods in- treme floods in the Rocky Mountains. This results from a cludingPMF values.These extensions provide an approachto misunderstandingof the complexhydrometeorological pro- place PMF estimatesin perspectivewith regionalgaged and cesses involved and a lack of data on extreme rainstorms and p•aleofloodestimates of flood potential [Jarrettand Costa, flooding.Jarrett [1993] made a systematicevaluation of flood- 881. frequencyestimates for 25 long-term streamflow-gagingsta- Paleofioodhydrology is the studyof past or ancientfloods tions throughoutColorado where publishedFederal Emer- [Baker,1987]. Kochel and Baker [1982],Gregory [1983],Baker et gencyManagement Agency (FEMA) floodplainreports, which al. [1988], Costa[1987c], Stedinger and Baker [1987],Stedinger were derivedby variousmethods, also were available.On av- and Cohn [1986],Hupp [1988], andJarrett [1987, 1990b,1991] erage,the publishedFEMA 100-yearflood is 35% greaterthan provide summariesof paleofioodhydrology. Although most the gaged100-year flood obtainedby fitting a log-Pearsontype studiesinvolve prehistoric floods, the methodologyis applica- III distribution to streamflow data for the 25 stations. Similar ble to historic or modern floods [Baker et al., 1988; Jarrett, differenceswere identifiedfor the 10-, 50-, and 500-yearfloods 1990b].Paleofiood studies provide important information that and demonstratethe needto improveflood-frequency estimates can be used in risk assessments and in the assessment of cli- usedfor floodplainmanagement and other usesin Colorado. matic changeon floodingand droughts[Jarrett, 1991]. Paleo- Interdisciplinaryflood researchin Colorado [Jarrettand flood data are particularlyuseful in providingupper limits of Costa, 1983, 1988; Jarrett, 1987, 1990a, b; Grimm, 1993; Pitlick, the largestfloods that have occurredin a river basinin long 1994; Waythomasand Jarrett, 1994; Pruess, 1996; Capesius, time spans[Jarrett, 1990b; Enzel et al., 1993]. 1996;Pruess et al., 1998;McKee and Doesken,1997] and in the A regionalinterdisciplinary paleoflood study was conducted Rocky Mountains [Hertz, 1991; Jarrett, 1993; Jensen,1995; in northwesternColorado (Figure 1) to help assessthe flood Buckley,1995; Eastwood, 1995; Brien, 1996; Parrerr, 1997, 1998] hydrologyfor Elkhead Reservoirin Elkhead Creek basinnear providesnew insight into the hydrometeorologyof extreme Craig. The objectiveof the paleofloodstudy was to estimate flooding in the Rocky Mountains. In an analysisof USGS prior maximumflooding during the Holocene from evidence streamflow-gagingstation and paleoflooddata in the Rocky preservedin the floodplain.The interdisciplinarycomponents Mountain region,which included stations in New Mexico,Col- included documentingmaximum paleofloodsand regional orado, Wyoming, Idaho, and Montana, envelope curvesof analysesof contemporary(---155 years in Colorado)extreme maximumunit discharges(maximum peak flow dividedby rainfall and flood data in and near Elkhead basin.The major drainagearea) and gageelevation identified elevation limits drainageswithin the regionalstudy area are the Yampa River for largeunit discharges[Jarrett, 1993]. In Coloradothe eleva- and White River basins.Hydroclimatic conditions are rela- tionlimit is about2300 m for anenvelope curve value of 1 m3 tively homogeneousin northwesternColorado [Miller et al., s-• km-: for basinsless than about 10 km:; as basinsize 1973;Kircher et al., 1985].A primaryfocus of U.S. Geological increases,unit dischargedecreases. Such low-magnitude unit Survey(USGS) interdisciplinaryresearch is to developcost- dischargesresult in only minor flooding in Colorado; flows effectivepaleoflood techniques that can be used to comple- seldomexceed the top of the main-channelbanks, and when ment meteorologic,hydrologic, and engineeringmethods to they do, flow depths usually are insufficientto modify the improve estimationof the magnitude,frequency, and risk of floodplain. These peak dischargesare causedprimarily by floods. snowmeltrunoff, relativelysmall amounts of rainfall (as com- The paleofloodstudy was conducted for the ColoradoRiver pared to lower-elevationrainfall amounts),or a combination Water ConservationDistrict to complement a site-specific of rainfall on snowmelt. Below about 2300 rn in eastern Col- PMP by Tomlinsonand Solak[1997] and a PMF studyby Ayres orado,unit dischargesas largeas 38 m3 s-1 km-2 haveoc- Associates,Inc. in Fort Collins, Colorado, for Elkhead Reser- curred. Above about 2400 m in Colorado, maximum observed voir. Elkhead Reservoirwas being recertifiedfor hydrologic 6-hour rainfall is about 100 mm; in eastern Colorado at lower safety by the Colorado State Engineer. PMP estimatesare elevations, maximum observed 6-hour rainfall is about 610 consideredof lesserreliability along the Continental Divide, mm, which also is the maximum24-hour value [Hansenet al., which includesthe upper Yampa River basin [Hansenet al., 1978]. 1977]. Therefore a site-specificPMP studywas conductedto The interdisciplinaryresearch cited abovehas providedde- addressissues raised by the NRC [1988,1994] pertaining to the finitive informationthat substantialflooding in Coloradohas hydrometeorologyfor the basin and the surroundinggeo- not beenobserved above an elevationof about2300 m [Jarrett, graphicallyand climatologicallysimilar region. Inherent in a 1987, 1990b, 1993; Jarrett and Costa, 1983, 1988; Grimm et al., site-specificPMP studyare analysesof extreme stormsthat 1995; Pruess,1996]. The limit varies somewhatbecause of JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2959
109 ø WYOMING 108 ø 41 ø •. 107 ø 106 ø II COLORADO
LARIMER
MOFFAT Cache ia Poudre River I¸ Maybell Fort Collins Reservoir i ßLay Craig OLYMPUS RABBIT EARS Estes Hayden Milner PASS DAMLoveland SteamboatSprings i STUDYAREA/"•\ i ROU'I-F DIVIDE I- \ RIO BLANCO GRAND i Yellow \Meeker BOULDER 4oo_
"'
CLEAR z CREEK Creek i Rifle GlenwoodSprings EAGLE GARFIELD I CONTINENTAL i PITKIN PARK 39øF Grand i MESA DELTA I CHAF•EE•' ) I, GUNNISON 0 25 50 75 100 MILES I I I I I 0 25 75 100 KILOMETERS • .Junction•1 "1'"-Maparea COLORADO
Figure 1. Locationof theregional interdisciplinary paleoflood study area for northwesternColorado, which ishighlighted bythe bold dashed line. The northwestern (NW) and southwestern (SW) Colorado topographic boundaryformed by the ColoradoPlateau (CP) are labeledand shown as hatched area. The WhiteRiver Plateau(WRP) and Gore (G), RabbitEars (RE), andPark (P) mountainranges are labeled and denoted with different patterns.
local/regionalhydroclimatic variations and is somewhatlower mates are 250 mm in 6 hours and 510 mm in 24 hours in the in selectedbasins east of the ContinentalDivide [Jarrett, upperYampa River basin[Hansen and Schwarz,1981; Hansen 1990b]and most basins west of theContinental Divide [Jarrett, et al., 1988].For comparison,in easternColorado (Denver), 1993].The elevationlimit of about 1 m3 s-1 km-2 seemsto PMP values are 675 mm in 6 hours and 920 mm in 24 hours vary from about 2400 m in the southern Rockies to about [Hansenet al., 1988].J. F. Henz(Review of probablemaximum 1700m in the northernU.S. RockyMountains [Jarrett, 1993]. precipitationin Wyoming--LevelII, Phase1 Report, draft, M. Quick(University of BritishColumbia, unpublished data, preparedfor Wyoming'sState Engineer'sOffice, 1991) re- 1993)indicated that the elevation limit in theCanadian Rocky viewedthe use and applicabilityof currentPMP methodolo- Mountains is somewhat lower than the elevation limit for the gies in Wyoming[Hansen et al., 1988]. He concludedthat northernU.S. RockyMountains. additionalmeteorological research is neededto improveesti- Theseresults, which are supportedby analysesof extreme matesof extremeprecipitation in the Rocky Mountainsof rainfalldata and paleoflood studies [Jarrett, 1987, 1990b; Jarrett Wyoming.Buckley [1995] concluded that thereare no signifi- and Costa,1988; Grimm, 1993;Waythomas and Jarrett,1994; cantrainstorms even remotely comparable to themagnitude of Pruess,1996; Brien, 1996] contrast dramatically from published PMP estimatesfor mountainsin Wyoming.Eastwood [1995] valuesof PMP andPMF for the RockyMountains. PMP esti- developedregional relations to estimatethe frequencyof ex- 2960 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
treme precipitationin the mountainsof Wyoming.Tomlinson maximumelevation of 3808m) and flowswesterly through the and Solak [1994, 1997] developeda site-specificmethodology Gore (3295 m), Rabbit Ears (3748 m), and Park (3725 m) to determine PMP/PMF estimates.Jensen [1995] developed mountainranges (Figure 1). The White River originateson the new criteria for computingPMP estimatesfor short-duration, White River Plateau and also flows westerly.The boundary small-area storms in Utah. His study resulted in significant between northwestern and southwestern Colorado is defined decreasesin extremeprecipitation for Utah comparedto cur- by the topographicdivide between the White River and Col- rently usedPMP estimates[Hansen et al., 1977]. orado River basins(Figure 1), which has elevationsranging Large differencesin estimatesof extremerainfall and flood- from about 2500 to 3800 m. Elevations at the downstream ing have substantialeffects on dam safety. For example, a studylimit are 1804 m at Maybell in the Yampa River basin paleofloodstudy was conducted for the Bureauof Reclamation and 1898m at Meeker in the White River basin.Major Yampa for OlympusDam in EstesPark, Colorado,on the Big Thomp- tributaries include the Elk and Little Snake Rivers and Elk- sonRiver [Jarrettand Costa,1988]. OlympusDam is locatedat head and FortificationCreeks. The regionalstudy area is ap- an elevation of 2300 m, and the spillwaywas designedfor a proximately10,900 km 2. PMF of 637 m3 s-1. However,a revisedPMF (Bureauof Elkhead Creek has its headwaters in the Elkhead Mountains Reclamation,written communication,1988), based on the re- and flows southwesterlyto its confluencewith the Yampa visedPMP estimates[Hansen et ai., i988], is 2380m 3 s-•-. River about ...... 1[/ 1•II1 east of '•--:-•.•1 (Figure 1). Elkhead Creek Paleofloodinvestigations by Jarrettand Costa[1988] indicated basinhas a drainagearea of 531knl 2 at Elkheaddam. Eleva- that the largestnatural flood flow in the Big ThompsonRiver tionsin the basinrange from about3307 m at the highestpeak upstreamfrom Olympus Dam is 142m 3 s-1 (6% of therevised of the Elkhead Mountains to about 1890 m at its confluence PMF) duringat leastthe past 10,000years (since glaciation). with the Yampa River. The elevationof Elkhead Reservoiris This paleofloodinformation and a review of the existingand about 1950 m. Distinct mountainsand ridgesdefine the north revisedPMF valuesby the Bureau of Reclamationresulted in (-2900 m), east(-2400 m), and west (-2300 m) boundaries a decisionnot to modify the spillwayfor OlympusDam at an of the basin.The topographyis rollinghills andvalleys, except estimatedcost of $10 million (Bureauof Reclamation,written in the steeper,mountainous headwater areas. Elkhead Creek communication,1988). and numeroustributaries drain the mountainsforming the In part becauseof the interdisciplinaryresearch, concerns basinboundary. Most streamsin the studyarea are of higher and questionsof extreme rainfall and flood designvalues for gradientwith slopes greater than 0.002 m m-• [Jarrett,1984], structureslocated in floodplainshave been raisedby stateand except the lower reaches of Elkhead Creek and the lower federal dam safety officialsfor the Rocky Mountains. Most Yampa River. Cobble-and boulder-sized material make up the state agenciesin the Rocky Mountain region have ongoing stream bed and fine-grainedsediments compose the flood- hydrometeorologicand paleofloodstudies to revisemethodol- plain. Somelower-elevation tributary valleys to the Yampa and ogiesto estimateextreme precipitation and floodingfor dam White Rivers are predominantlyfine-grained alluvial fill. safetybecause of recognizeddeficiencies in PMP estimatesin Elkhead Creek basin is underlain by Cretaceousand Ter- mountainousareas. Colorado began studiesto developnew tiary rocks(shale, sandstone, conglomerate, and coals)in the methodsto estimateextreme rainfall in the mountains[McKee Lance, Lewis, Wasatch, Browns Park, Fort Union, and Iles and Doesken,1997], and the second,30-month phase recently formations;upper Tertiary intrusiverock, primarily porhyries began(A. Pearson,Colorado Dam SafetyOffice, written com- of intermediate and basalticcomposition, cover parts of the munication,1999). The Bureauof Reclamationrecently began basin [Tweto, 1976]. Within the general study area, similar a programto use a risk-basedassessment, which incorporates geologicformations as in Elkhead Creek basinwith Precam- paleoflood investigationsto provide estimatesof the magni- brian rocks (granite, Quartz monzonite,granodiorite, Quartz tude and frequencyof extremefloods, to assistwith dam safety diorite, and gabbro)and biotite and hornblendgneisses occur decisionmaking [Levish et al., 1994;Ostenaa and Levish,1995]. in the Park Range. Tertiary andesiticand basalticlava flows The U.S. Army Corps of Engineers is implementinga risk from the Flattopsand Elkhead Mountainswith someintrusive assessmentmethod to evaluatepotential safetyproblems for rocksin the studyarea [Tweto,1976]. Most of the Park Range, its more than 550 damsto aid decisionmakers in prioritizing upper Elkhead Creek basin, and the Flattops experiencedat investmentdecisions [Foster, 1999]. In 1999 the American So- leastthree Pleistoceneglaciations [Madole, 1982, 1989, 1991a, ciety of Civil Engineersbegan a task committeeon paleoflood 1991b,1991c]. Flights of Pleistoceneterraces along the Yampa hydrologyas it relates to dam safety and risk-basedassess- River from about SteamboatSprings downstream to about ments.The NRC [1988, p. 111] statesthat "Nonetheless,the Craig, probablyfrom glacial processes,range from about a expenseof suchstudies is minor in relation to planningcosts meter (early Holocene)to 183 m (620 ka) abovethe present for major high-riskprojects such as nuclear power plants or floodplain;the average incision rate since 600 ka is0.11 m ka-• large dams. At presentthese opportunitiesare largely being [Madole, 1991a]. Unglaciated tributaries lack the well- ignored.... For criticalprojects the paleoflooddata shouldat developedPleistocene terraces, and Holocene terracesare rel- least be collected,appropriately weighed, and consideredin ativelyclose to the valleyfloors [Madole,1991a]. Loess, typi- the overall decisionprocess leading to design."Thus it is im- cally 1.3 to 2 m thick, of at least two agesis widespreadin the portant to developmethodologies that can be used by dam Yampa River basinwith the latest depositionin the late Pleis- safety officialsto make decisionsabout the probabilitiesof toceneand possiblyearly Holocene [Madole,1991a]. extreme floods. The majority of Elkhead Creek basin and regional study area hasbeen mapped as low to moderatelywell-drained soils, exceptin limited higherelevation areas where bedrockis at or 3. Study Area near the groundsurface [Soil ConservationService, 1982; Nat- The Yampa River in northwesternColorado originateson uralResources Conservation Service, 2000]. At higherelevations the White River Plateau (also known as the Flattopswith a in Elkhead Creek and the upper Yampa River and White JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2961
River basins,subalpine forests consisting of aspen,lodgepole 1976 Big Thompsonfloods were imbeddedin one of the worst pine,Douglas fir, andEngelmann spruce are common.In most droughts in contemporary records, whereas other extreme lower parts of Elkhead Creek basin and Yampa River basin floodshave occurred during wet periods,such as the 1965Plum below SteamboatSprings, vegetation consists primarily of pi- Creek flood [Collinset al., 1991]. The paleoflood data de- non pine, juniper, sagebrush,rabbit brush,and native grasses. scribedin this report provide a meansto assessthe effectsof Mean annual precipitationin the basinvaries from about climate changeon large floodsduring the Holocene. 405 mm at Elkhead reservoirto slightlymore than 760 mm near the headwaters[Doesken et al., 1984]. Most annualpre- cipitationfalls as snowin the winter months.The largestpre- 4. Methodology cipitationamounts are limitedto smallparts of the basinat the This sectiondescribes methods used to estimatepaleoflood highestelevations around the eastand north rims of the basin. discharge,determine the paleoflood chronology,analyze re- Within the larger context of the regional study area, mean gional precipitationand streamflowdata, and conductflood- annual precipitationranges from over 1525 mm in the Park frequencyanalyses. The strategyof a paleofloodinvestigation Range(the wettestarea in Colorado)to 305 mm near Maybell is to visit the most likely placeswhere evidenceof substantial to 406 mm at Meeker [Doeskenet al., 1984].Frequent, local- flooding, if any, might be preserved.Because glaciation and ized convectiverainstorms occur during the summer months. glacial outwash"erases" evidence of floods, paleoflood evi- Convectivestorms have producedmoderate flooding in small, dencegenerally can be no older than the time sincethe last steepbasins with little vegetationat lower elevationsin north- period of glaciationor about 12,000 years ago in the Rocky westernColorado [Jarrett,1987]. These basinsare located in Mountains[Madole, 1991a]. For unglaciatedbasins the objec- the western parts of the regional study area, particularly tive was to identify the largestflood that has occurredwithin Piceance Creek and Yellow Creek basins. General rainstorms the longesttime period but limited to the Holocenewhen the in northwesternColorado can cover large areas but have not climate variability has been relatively constant.Thus paleo- producedsubstantial flooding in historictimes. flood investigationscan identify physicalevidence for the oc- Flood-frequencyanalysis has long been done assumingsta- currenceor nonoccurrenceof substantialfloods for very long tionarityof climaticand hydrologicprocesses [Interagency Ad- time periods.Paleoflood data were collectedfor siteson Elk- visoryCommittee on WaterData, 1981;NRC, 1999]. There is head Creek and its tributaries and from numerous streams in growingconcern that climate naturallyvaries with time and the Yampa, White, and Little Snake Rivers basinsin north- that climatemay be respondingto anthropogeniceffects; how- western Colorado. ever, at present (2000), there is little informationto assess theseimpacts on averageor extremeconditions [NRC, 1999]. 4.1. Paleofiood Discharge Lins and Slack[1999] analyzed contemporary peak streamflow Floodsleave distinctivedeposits and landformsin and along data in the United Statesand were not able to detect signifi- stream channels,as well as botanic evidence [Baker, 1987; cant trends.Knox [1993] documentedincreases in flood mag- Baker et al., 1988; Jarrett, 1987, 1990b, 1991; Hupp, 1988]. nitude due to climate changein the upper MississippiRiver Slack-waterdeposits of sand-sizedparticles [Kochel and Baker, basin,but these resultsare site-specific.Ely [1997] suggests 1982;Baker et al., 1988],flood scarson trees,erosion scars, and evidencefor climate and flood flow changeover the last 1000 boulderyflood-bar depositscommonly are used as indicators years, particularly the last 2 centuriesin the southwestern of pastflood levelscalled paleostage indicators (PSIs) (Figure United States.England [1998] noted that no explanationwas 2). When flowsare large enough,streambed and bank mate- providedfor the apparentincrease in floodsand that only six rials are mobilizedand transported[Costa, 1983; Komar, 1987; of Ely's [1997] 19 siteshave recordsequal to or greater than Wilcock,1992]. Such mobilization and transportare a function 1000years. Of equalimportance, is that Ely's [1997]study does of channelgradient. As gradient increases,smaller velocities not account for large uncertaintiesin estimatingflood dis- and depths are required to move sedimenton the bed of a chargeaveraging 60% as noted by Jarrett[1986, 1987, 1994]. stream[Costa, 1983]. When streamvelocity, depth, and slope Few climate changestudies have been done for the Rocky decrease,flowing water often is no longer competentto trans- Mountainregion. Paleoclimatic studies by Madole [1986],Elias port sediments,which are depositedas flood bars and slack- [1996],Menounos and Reasoner[1997], Valero-Garceset al. waterdeposits in the channelor on the floodplain(e.g., Figures [1997], Fredlundand Tieszen[1997], Smith and Betancourt 3 and 4). The typesof siteswhere flow competencedecreases [1998],and Vierling[1998] used dendrochronology, pollen, di- and flood depositscommonly are found and studied include atoms,fossil bettle assemblages,carbonate geochemistry, iso- (1) locationsof rapid energydissipation, where transported topic evidence,and sedimentstratigraphy, respectively, to doc- sedimentswould be deposited,such as tributary junctions, umentmean temperature and precipitationfluctuations. These reachesof decreasedchannel gradient, abrupt channelexpan- researchresults are in good agreementthat warmingtemper- sions,or reachesof increasedflow depth;(2) locationsalong aturesresulted in deglaciationof high mountainareas between the sides of valleys in wide, expandingreaches where fine- about 13,000 and 12,000years B.P. and that modern climate grainedsediments or slack-waterdeposits would likely be de- began about 9500 to 9000 yrs B.P. In this period of modern posited;(3) pondedareas upstream from channelcontractions; climate,average temperatures varied _+1 ø to 2øCfrom contem- (4) the insideof bendsor overbankareas on the outsideof porary mean summer temperatures.Average annual precipi- bends;and (5) locationsdownstream from morainesacross tation and average annual streamflowhave varied by up to valleyfloors where large floods would likely depositsediments about 50% of modern mean values,but nearly similar varia- eroded from the moraines.Flood-transported sediments and tions have occurred during contemporaryrecords [Jarrett, woodydebris can scartrees (Figure 5) and alsoaccumulate on 1991]. However, inferring extremevariations from studiesof trees and other obstructionsto provide a good indicator of averageclimate changeis problematic.For example,extreme flood height. The height of tree scarsand the top of woody floods in Colorado such as resulted from the 1935 Hale and debrisare usedas indicators of approximateflood height[Har- 2962 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
/,Flood stager I'!••i•'"x•rree bG;ra ':•!•J• LøW:nW•ter
Figure 2. Diagrammaticsection across a streamchannel showing a hypotheticalmaximum paleofiood and variousflood featurespreserved as paleostageindicators [Jarrett, 1991]. risonand Reid, 1967;Hupp, 1988; Gottesfeld,1996]. A lack of channel-bedelevation. Using the estimatedflood depth and scarringis an indicator that floodinghas not occurredsince channelgeometry, the width and cross-sectionalarea belowthe establishmentof trees on the floodplain.In semiaridstreams, PSI elevation was determined. Because most streams in the woody flood debris typically lasts 60 years or longer before studyarea are highergradient (>0.002 m m-•), paleoflood completelydecaying (D. Levish,Bureau of Reclamation,per- discharge was estimated using the critical-depth method sonalcommunication, 1998). [Barnesand Davidian, 1978],which hasbeen suggestedfor use Paleoflood dischargewas determined from estimates of becauseflow in these streamsusually is very near critical or flood width and depth correspondingto the elevation of the slightlysubcritical, particularly for large floods [Jarrett,1984, top of flood-depositedsediments (or PSIs) and channelslope 1986; Triesteand Jarrett,1987]. The slope-conveyancemethod obtained during on-site visits to streams.Hydraulic calcula- [Barnesand Davidian, 1978] was used to estimatepaleoflood tions are similar to indirect dischargeestimates using high- dischargein the few lower-gradientrivers (mostly mainstream water marksfollowing floods, except PSI usuallyare older and Yampa River downstreamfrom Craig). Flow-resistancecoef- may have greater uncertainty.Flood depth was estimatedby ficientsfor lower-gradientrivers were estimatedfrom analysis usingthe PSIs in the channelor on the floodplainabove the of data for Coloradorivers [Jarrett, 1984, 1985].Although flow-
Figure 3. Upstreamview of flood-bardeposit for Lefthand Creek near Boulder, Colorado,with 1995 flood watersabout 20 cmbelow the peak stage. Peak discharge was 34 m3 s-], andthe elevation of thetop of flood bar (and a separateslack-water deposit of sandin far right center) equaledmaximum flood elevation. JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2963
Figure 4. Cross-streamview of Roaring River alluvial-fandeposit from the 1982 Lawn Lake dam-failure flood.Peak discharge was approximately 340 m 3 s-•. competencerelations can be used to estimatethe flow depth these are consideredflow depthswhich do not produceflood for surface-sedimentsizes [Costa, 1983; Levish et al., 1994], erosionalor depositionalevidence based on the author's ex- they do not take into accountvegetation and localizedturbu- tensivepostflood studies in the Rocky Mountain region. lence,particularly on floodplainsurfaces, and generally do not In paleofloodinvestigations, lack of physicalevidence of the provideconsistent results [Pruess, 1996]. Thus, in thisstudy, for occurrenceof floods is as important as discoveringtangible streamsthat haveno overbankflood evidence(flood barsand on-site evidence of such floods [Stedingerand Cohn, 1986; erosionalfeatures), a PSI flow depthof lessthan 0.5 m (higher Jarrett, 1987, 1990b; Jarrett and Costa, 1988; Levish et al., 1994; gradients)to 1 m (lowergradients) on the floodplainwas used; Ostenaaand Levish,1995]. The geomorphicevidence of floods
•-.?•ß'..::.;.':•:: ....,,•I.:' :' * ".i:;i:..:::!5 "??'"..;1':.:??:?g•}, I;;:'•"?::....:::? ::;* ':•:'*"'":'...... ' ...... ***'
..... ß...... / '.:; '•:;•:.'•,•;:•'- ?..;•'•*•fa:.'-.'• B•>5;:;:.'%/.(.'•:':;•. .. ..:;.7%;>;;•:•' ,:?...... ,..,.,..-...... •,.:,,•..
...... ,-.- ..-• ...... ::.:...-*:-:'*:-**.-'-.:-'.*.:•:."-.--:.-:.:*:,-."*'"*•*•}..<'%•,a***.;;::•.";.':'":5 .,,..:,...... **.>.:;.;** ...... •.,...... :-"•b• •" .:•,••:..:,'-•::•::•-: •'"':-• ...... :-:--::;-:-...... •::,;- -..•':.•;<&•:,**•w**:':•,?*a.*•-;:w•;?•::*:½::•%:•:g *:*;½-/-'•,..,...½..'..., .... ß...... ':',.*'...... ::,"'*::::';:.**::<::?c?*::?,a,,**.?:.:,.-, -'---**"-'-'-',:.,:,-"....
;•.. .,.•<*.•*`•*`:•*}•:}•***•*•::•..:..:•::•:...*•;:*.•;:•:::•a:•*•:.;.;;;*;.•**.*:.•-::--'"'*:::- ..:'..:....-'.":""..-'"':.%.::'--
'*'•*:•;.ß,..,...... •:•***'•8***•'**::?::%::;..'•*•;';,'" :...... yX.•.....:. *''• • .::..•....:.::..::..;,:.'•:,z;,:• ...... •;C...... ß . '.%%:.':':-' ...•;...... ,:,..•:....•...... ,•,:.....:..:•......
":...... , ...... ;,..,..,..;•.x •..... _./ ,a• , -.,.,.:,,**•....:•,..,.:?.....-,•:.•...... •...... 5.•;::•.,.•,...,•.:,.X ...... ' ß .-.:..,:..:?.....:
-'• ......
Figure 5. Downstreamview of flood scarsand woodydebris from the 1982 Lawn Lake dam-failureflood (380m 3 s-•); flowdepth at surveyrod (1.5 m) isapproximately 30cm above ground. Flood scars average about 1 m higher than the 1982 flood height. 2964 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
in steepmountain basins [Boner and Stermitz,1967; Snipeset the higheststage reached by a flood and primarilyconsist of al., 1974; Schwarzet al., 1975; McCain et al., 1979; Costa, 1983; fine woodydebris, leaves, grass, needles, other floatablemate- Jarrettand Costa, 1986;Jarrett, 1987, 1990b, 1991; Grimm et al., rials and mud lines. These streamshave drainageareas that 1995;Waythomas and Jarrett, 1994] is unequivocal(e.g., Figures rangefrom about0.004 km 2 to more than 6000km 2. Peak 3-5). Paleofloodevidence is relativelyeasy to recognizeand dischargeranged from about0.03 to 2500m 3 s-•, and the longlasting because of the quantity,morphology, and structure majoritywere larger than 100-yearfloods. Stream gradient at and size of sedimentsdeposited by floods.Lack of movement thesesites ranges from about 0.0007 to 0.35m m-•. Thesize of of all sizes of clasts on a streambed with an unlimited size of flood-depositedsediments ranged from silt to bouldersmore clastsis a direct measureof flow competence[Costa, 1983; than 4 m in diameter;only thosedeposits considered to have Komar, 1987;Bull, 1990; Wilcock,1992] and an indirect mea- long-termpreservation potential were documented.Analysis sureof flood discharge.In manychannels in Colorado,there is of the differences in PSIs and HWMs indicates that the eleva- an "unlimited" size of clastspresent in channelsbecause of tionsof the top of flood-depositedsediments (PSIs) generally pastglacial outwash or rockfallthat are availablefor transport. are within _+0.2 m of flood HWM elevations. Therefore use of Point, longitudinal,and transversebars are built up of layersof the top of flood-depositedsediments as PSIs for streamsin this silt, sand,gravel, cobble, and bouldersfrom bed load to about studyprovides a reliable estimateof the maximumpaleoflood the height of maximumflood water [jarrettet ai., 1996]. For a depththat is usedto reconstructthe dischargeof paleofloods. givenreach of channelthe relation of maximumclast size in a Good HWMs for the 1995 near-recordpeak flows during flood depositto the availableclast size in the upstreamchannel fieldwork were established,which were not that much smaller was noted. Clasts show little reworking on the stream bed than maximumpaleofloods. Paleoflood techniques were used (random pattern due to winnowingof finer materials)when to estimatepeak dischargefor the 1995 flood where stream- floods are small; however,large floodstend to producewell- flow-gagingstation estimates were availablebut without prior developed, fluvial-depositionalfeatures. In addition, when dischargeknowledge. Although this comparativeapproach flowsexceed the main channelby evensmall depths (<0.5 m), does not considerall the uncertainties,the comparisondoes coarser-grainedbed material is transportedonto the floodplain providean estimateof the uncertaintyof the criticaldepth and and deposited[Jarrett et al., 1996] suchas shownin Figure 3; slope-conveyancemethods for estimatingpeak dischargein the thus lack of coarsematerial on floodplain surfacesindicated study. minimal inundation. There are three main sourcesof uncertaintyin paleoflood 4.2. Paleoflood Chronology reconstructions[Jarrett and Malde, 1987]: selectionof flow- In this study,a variety of relative-dating(RD) techniques resistancecoefficients, channel changes,and representative- were usedto estimatethe approximatelength of time (age) nessof PSIs of flood height. Selectingflow-resistance coeffi- correspondingto the highestflood depositsemplaced during cients,whether Manning'sn value as in this studyor other the Holocenefor subsequentuse in flood-frequencyanalysis. resistanceforms (e.g., Darcy'sor Chezy's),can be problematic RD techniques based on landform modification, rock- for flood studies,particularly for paleofloodestimates when weatheringfeatures, and soildevelopment have long been used riparian vegetationmay have been different and unquantifi- to differentiateand map Quaternarydeposits in the western able. Jarrett[1986] and Triesteand Jarrett[1987] indicate for United Stateswith emphasison datingglacial deposits [Birke- most higher-gradientstreams and many lower-gradientrivers, land et al., 1979; Colman and Pierce,1983; Birkeland, 1990]. flow is about critical or slightlysubcritical even in channels Rock-weathering,soil, and geomorphicparameters all change havingsubstantial channel and floodplain vegetation prior to a with time [Colman and Pierce, 1983; Birkelandet al., 1979]. flood.Large floodstend to removemost vegetation prior to the Althoughmuch datinghas involvedfluvial deposits,ages pri- peak stage [Matthai, 1969; McCain et al., 1979; Phillips and marily are determinedwith absolute-datingmethods such as Ingersoll,1998], and becauseflow is nearlycritical, some of the •4C, thermoluminesce,and dendrochronology[Kochel and uncertaintyin estimatingn valuesdue to unquantifiablepast Baker, 1982; Hupp, 1988; Colman and Pierce, 1991]. Some vegetationis removed. investigatorshave useda variety of RD methods,which utilize To minimize effects due to channel changes,the second the collectivestrengths of each method [Burkeand Birkeland, sourceof uncertainty,cross sections were located in bedrock 1979; Harden, 1986, 1990; Waythomasand Jarrett,1994; Mills reachesor in relativelystable alluvial reaches. Although most andAllison, 1995]. The chancesof arrivingat a valid approxi- of the Yampa River is alluvial,there are good constraintson mate age is greatlyenhanced if a variety of RD methodsare relativestability or slightrates of incisionduring the Holocene measured[Birkeland et al., 1979]. [Madole,1991a]. For alluvialtributary streams that may have RD methodsapplied to surficialdeposits are basedon post- undergonecyclical aggradation and degradationor if these depositionalmodifications that varywith age [Birkelandet al., cycleswere difficultto ascertain,then very restrictive(conser- 1979].Field evidenceof age is usuallyderived from soil prop- vative) ageswere assignedto the paleofloodestimate, usually erties, rock-weatheringcharacteristics, changes in landform lessthan 100 years. morphology,and lichenometry.Their usefulnessfor character- The third sourceof uncertaintyin paleofloodreconstruc- izing surficialdeposits is related to the degreeto which they tions is maximum flood height inferred from PSIs. In most canbe quantifiedand to their rate of change[Birkeland et al., paleofloodstudies, PSIs have been assumed to be slightlylower 1979]. Episodicflooding produces deposits of different ages than maximumflood height [Kocheland Baker, 1982;NRC, that can be separatedby long periods;thus the depositshave 1988;Baker et al., 1988].New researchon the elevationof the distinctiveproperties that are amenableto measurementby top of flood-depositedsediment (new PSIs) and high-water RD methods.It is assumedthat when flood-depositedsedi- marks(HWMs) of recentflooding in 90 streamsprimarily in ments undergo transport and reworking during high-energy the westernUnited Statesthat has been done [Jarrettet al., flood transport,the weathering"clock" essentiallyis reset to 1996] challengesthis assumption.HWMs are the evidenceof zero [Waythomasand Jarrett,1994]. Certainly,a limitation is JARRETI' AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2965
Table 1. Descriptionof Relative Dating Methods Used in NorthwesternColorado
Numerical Rating and Descriptiona Type of Relative Dating Method 0-3 4-6 7-10
Soil horizons C (increasingO/A) O/A/C O/A/Btj/C Rock weathering fresh partly weathered very weathered Pitting, % <10, rare/incipient 30-70 >75, common Grain relief, mm <0.5 0.5-1 Boulder burial, % 0-25 25-75 >75 Surfacemorphology Terrace scarp angular moderatelyrounded well rounded Slope steep moderatelymuted extremelymuted Terrace tread fresh longitudinal moderatetransverse extensivetransverse flood evidence rills and gullies rills and gullies Lichenometry Largest thalli, mm 0-100 > 150 > 150 Rock coverage,% <75 >75 >75
Abbreviationtj indicatesincipient accumulationof silicateclay that has either formed in situ or is alluvial [Birkeland,1984]. aA ratingof 0 is modernor 0 years;10 is earlyHolocene or about!0,000 yearsor older.The rating valuesare approximate,nonlinear, and applicablefor northwesternColorado river valleys.
the occurrenceof large floodsseparated by a short time span 4.2.1. Soils. There is a strongrelation between degree of whereRD techniques(and evenabsolute techniques) may not soildevelopment and time, althoughrates of soildevelopment be able to differentiate between several individual older de- varywidely [Birkeland,1984; Harden, 1986, 1990]. Correlation posits. of soil-profiledevelopment with soil chronosequences,dated The most important factor for using RD techniquesfor with numericaltechniques, is crucialwhen determiningthe fluvial depositsis to comparethe depositswith flood deposits relativeage of surfaces[Birkeland, 1984; Bull, 1990].Soils show and other surfacesimmediately adjacent (upslope and down- a systematicand generallyslow progressive soil-profile devel- slope)in a shortreach (site). Within-stream factors would have opmentwith age.Readily availablesoil surveysprovide useful zero age,whereas deposits and surfaceswith increasingheight data used in conjunctionwith field checking.The degree of abovethe channelhave increasing age. Thus state factors (e.g., developmentof the local soil profile was determinedin the lithology,microclimate, climate, vegetation, and topography) field by trenchesand cut-bankexposures of flood depositsand are assumed to be held constant; therefore differences in terrace deposits.Age diagnosticparameters include the fol- weatheringand soil-profiledevelopment are directlyrelated to lowing:thickening of the total soilsand developmentof the B time at any individual site, which is the justificationof RD horizon;increasing enrichment of the B horizonsin secondary methods[Burke and Birkeland, 1979; Harden, 1982; Waythomas clay(an argillichorizon); presence, abundance, and thickness and Jarrett,1994]. Also, it is important to obtain agesusing of clay films; increasinglydiffuse horizon boundaries;abun- these different RD techniquesfor several sampleswithin a dance of calciumcarbonate; oxidation depth; pan develop- reach of channel. Similar ages derived from different tech- ments; and rubificationof the B and C horizons[Bilzi and niquesresult in increasedconfidence in the estimate[Burke Ciolkosz, 1977; Harden, 1982; Colman and Pierce, 1983; Birke- andBirkeland, 1979; Waythomas and Jarrett, 1994]. Because the land, 1984].Conditions that increasethe rate of soil develop- mostlimiting factor in age-datingstudies is small samplesize ment includethe following:warm, humid climate;forest veg- [Harden,1990], dating numerousdeposits along rivers helps etation; high permeability;and flat topography.Conditions increasethe reliabilityof age determinations. that tend to retard soil developmentare cold, dry climates, The focusof this studywas to identifygross changes in RD grassvegetation, low permeability,and steepslopes. features[Burke and Birkeland,1979], which were then usedto Alluvial soilsare often thoughtof as beingyoung or unde- estimate maximum age of flood or noninundationsurfaces. veloped,but this is not alwaystrue [Gerrard,1981]. Soils on Age estimatesfor paleoflooddeposits are basedon relative- river terracesare often interpretedas alluvial and considered agecriteria as proposed by Burke and Birkeland [1979], Colman young,but sincethe majority of river terracesare of Pleisto- and Pierce[1983], Harden [1982, 1986, 1990], and Waythomas cene age, many suchsoils are well developed[Gerrard, 1981]. and Jarrett [1994]. RD techniquesused for this studywere For example,Madole [1991a]identified numerous alluvial ter- degreeof soildevelopment (S), surface-rockweathering (W), racesup to 183 m above the presentchannel, but they have surfacemorphology (M), lichenometry(L), and boulder been dated to about 650 ka. Thus incision rates and time since burial(B), althoughnot all methodscould be usedat eachsite. inundationare importantin developingflood chronology. For each of these criteria a numerical value from 1 to 10 was 4.2.2. Surface-rockweathering. Ratios of fresh to weath- assigned,1 representingmodern channeldeposits and 10 ex- ered material,the abundanceof pitting, and pit depth on rock hibitinggreatest age correspondingto early Holoceneor older clastsare usefulfor subdividingHolocene deposits by an ageof (Table 1). For all RD methodsas discussedfor each except 1000years or more [Benedict,1968]. It is assumedthat abrasion lichenometry,the rating scaleis 1 for ---0 to 1000 years to a of granite, rhyolite, and basalt clastsduring flood transport value of 10 for ---10,000years. For lichenometry,1 is modern removesmost previouseffects of weathering.The degree of and 10 is probably3000 yearsor less.Finally, an averageage surface-rock weathering of 25-50 of the largest flood- and range of age uncertaintywas determined. depositedclasts of the samelithology provides an indicationof 2966 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD their relative age. Older depositshave extensivesurface pit- rock surfaces[White, 1971], take about 50 years to become ting, rougher surfaces,and increasinggrain relief becauseof established.Environmental factors known to affectthe growth differential weatheringof minerals.As lessresistant minerals of R. geographicuminclude rock type, shading,temperature, decompose,quartz and feldspar grains tend to stand out in moisture,and stabilityof the substrate,and thus they need to relief. Pitting is commonif presenton more than 75% of the be factored into age assignments.Growth increaseswith clastsurfaces and is rare (or incipient)if presenton lessthan coarsertexture, moisture, temperature. Abrasion during sedi- 10% of the clastsurfaces. Flood-deposited clasts are compared ment transportin higher-energystreams common to floodsin with end-memberclasts from the streambed(fresh) and clasts mountainregions removes most lichen thus essentiallyreset- much higheron the land surface(extremely weathered). ting the time "clock"to zero for lichengrowth. Benedict [1967, 4.2.3. Boulder burial. Many recent flood deposits in 1968] indicatedthat growthcurves are fairly constantbetween Colorado(e.g., Figure 3) and other mountainrivers consist of 3125m and 4047 m (correspondingto a mean air temperature little or modestamounts of matrix-supportedcobble and boul- changeof about 6.6øC) for his sites along the crest of the dery deposits,particularly around surface clasts [e.g., Matthai, ContinentalDivide in Colorado.Benedict's [1967, 1968] sites 1969;McCain et al., 1979; Costa,1983; Waythomasand Jarrett, are located about 80 km southeastof this studyarea. Lichen 1994;Jarrett et al., 1996]. Boulder burial refers to the percent- growthcurves have a maximum age utility of about 3000 years age •f th• total boulder surfaceexposed above ...... • For in Colorado [Benedict,1967, !968], probablya shortertime at flood depositsthe amount of cobble and boulder burial by lower elevations in Colorado where climate is more conducive additionof newersediment at the site (colluvium,eolian, and to faster growth rates. Maximum thalli diameter and percent slopewash) also can be usedto estimatethe relativeage of a lichen on similar rock types for 25-50 clastson the flood deposit[Waythomas aml •arrett, 199•]. Thus time or age since depositand other rock surfaceswere made at each site. flood depositionis inferredfrom depthof burial. The older the depositis, the greater is the percentageof theseclasts that are 4.3. Regional Analyses of Maximum Rainfall and Flood Data coveredby postflooddeposition. 4.2.4. Surfacemorphology. Formation of streamterraces A lack of flood evidence, particularly of extremely rare involveschanges in the behaviorof a fluvialsystem [Bull, 1990]. floods, in one basin such as Elkhead Creek basin could result Terracesmay form becauseof a variety of internal or external from pure chance.Thus it is essentialto ascertainthe flood changes,climate, tectonics,base level, slope, complex re- historyfor other basinsin the region [NRC, 1988]. Regional sponse,and thresholds[Patton aml Schumm,1975, 1981; Wom- analysisextends hydrometeorologic records and provides a ack and Schumm,1977; Bull, 1990]. Remnantsof the former tool to estimatedischarge at ungagedsites [Jarrett and Costa, streambedare preservedas terrace treads. Terrace features 1988;NRC, 1988;Hosking and Wallis,1998]. In addition,re- suchas riser angle becomemuted with time by faunal, water, gional analysesprovide improved estimatesof precipitation and wind action,and ratesof changedepend on factorssuch as and streamflowcharacteristics for gaged sites by decreasing cohesionof material,vegetation, and flow stress[Lewin, 1978; time-samplingerrors for relativelyindependent samples. Birkelandet al., 1979]. Younger terracestend to be more an- Predictingthe upper limits to the magnitudesof floodsin a gular and have steepslopes; with increasingage, terrace scarp specificregion has been a long-standingchallenge in flood slopesbecome flatter unlessmaintained by cut-bankerosion. hydrology.Envelope curves encompassingmaximum rainfall Along rivers in glaciatedbasins in the Yampa River basin, [Linsleyet al., 1982;Jarrett, 1987, 1990b]and floodsin a homo- early Holocene to late Pleistoceneterraces, which are 1-2 m geneoushydrometeorologic region have long been used in abovethe presentfloodplain, are coveredwith eolian (loess) flood hydrology[Crippen and Bue, 1977; Costa,1987a; Jarrett, deposits[Madole, 1991a]. These surfacesare relativelyeasily 1987, 1990b;Enzel et al., 1993].Utilization of envelopecurves erodible if flooded thus providingunique sitesfor paleoflood for a hydrometeorologicregion can be evaluatedby examining investigations.Over time, localhillslope runoff produces trans- maximum floods in nearby basins.A premise for envelope verse gullies or channelson terraces and colluvial surfaces; curvesis that not all basinsin the region are expectedto have greater developmentgenerally requires longer time. Also, if had the maximumflood, but no basinhas yet had a flood that alluvial (or colluvial)surfaces are inundatedby flood waters, exceedsthe envelopecurve for the specificregion. The primary microchannelsand surface depositsare producedlongitudi- limiting factorsfor extreme floods are amount, intensity,du- nally, which are somewhatsimilar to crevassechannels and ration, and spatialdistribution of rainfall, which includesoro- splays[Lewin, 1978]. With time, geomorphicexpression of graphic enhancementeffects and basin slope [Costa,1987b; these features is muted. Lack of such flood features or noni- Pitlick, 1994].Incorporating paleoflood data for variousbasins nundationsurfaces [Jarrett and Costa,1988; Levish et al., 1994; in the region providesan opportunityto add a new level of Ostenaaand Levish, 1995] providesan upper bound of flood confidenceto envelopecurves [NRC, 1988;Enzel et al., 1993]. height on a surfaceof known age. 4.2.5. Lichenometry. A commonRD techniqueused for 4.4. Rainfall Data dating glacialdeposits is lichenometry[Benedict, 1966, 1967, Extreme rainfall data for the last 100 yearswere compiled 1968; White, 1971;Beschel, 1973]. Its use has primarily been from 181 official precipitation gagesand numeroussupple- high-elevationor arcticclimates, and the transfervalue is lim- mental rainfall-bucketsurveys in western Colorado [McKee ited, particularlyfor growthcurves. Rhizocarpon geographicum, and Doesken,1997]. Four, long-term precipitationstations in the most commonlyused lichen for dating, growsthroughout the study area, at Steamboat Springs, Hayden, Lay, and most of Colorado. Most lichenometric studies use a combina- Meeker (Figure 1), have been operatedfrom 61 to 94 years. tion of maximum thallus diameter and percentagecover on Rainfall-bucketsurvey data, primarily collectedby the Na- clastsurfaces to determine the agesof late Holocene deposits tional Weather Service,Army Corpsof Engineers,and Bureau [Benedict,1967, 1968; Beschel,1973; Birkelandet al., 1979]. of Reclamation,were compiled for Colorado [Jarrett,1987, Lichens,which grow on all but freshly exposedor deposited 1990b]and updated through 1997 for thisstudy. Although very JARRETI' AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2967
little bucket data have been collectedin recent years, several Censoringbelow a thresholdcan accountfor an assumeddis- extreme rainstorms have been documented in northwestern tribution not fitting the "true" distributionat a site [NRC, Colorado sincethe early 1900s. 1999]. Low outliers in Colorado often result from modest streamflowdiversions for irrigation of hay meadowsduring 4.5. Streamflow Data low-flowyears, but in normalyears these diversions have min- Streamflowdata for 218 sitesin the Yampa River and White imal (<5%) effecton peakflows [Jarrett, 1987]. Historical and River basinswere compiledand usedto assessextreme flood- paleoflooddata also are censoredsamples because only the ing in the region. U.S. GeologicalSurvey and Colorado State largestfloods are recorded.Paleoflood data (magnitudeand Engineerstreamflow records through 1998 are availablefrom ages)were incorporatedinto the flood-frequencyanalysis to the U.S. GeologicalSurvey's NWlS-W Data Retrieval system extendthe gagedrecord. In the EMA analysisthe paleoflood (http://waterdata.usgs.gov).These data were then usedto de- dischargewas specifiedas a range, and EMA runswere made velop an envelopecurve of peak dischargeversus drainage for the rangein age (Table 2) for a site. area for northwesternColorado. Unit dischargecan be usedto infer both maximumrainfall intensitiesand spatial extent of 5. Results rainstorms[Jarrett, 1990b]; the data were usedto developan envelopecurve for unit dischargeversus elevation. 5.1. Paleoflood Investigations Paleoflooddata from on-sitestudies of 88 sitesthroughout 4.6. Flood-FrequencyRelations the studyarea are providedin Table 2. Althoughnot all trib- To help facilitate risk assessmentsof rare floods for dam utary streamsin the northwesternColorado were documented safetyofficials and floodplainmanagers, flood-frequency rela- (becauseof inaccessibilityto privateproperty), sites were se- tions were developedfrom an analysisof annual peak flows lected such that paleoflood data were collected downstream through1998 for selectedstreamflow-gaging stations in north- from tributaries on the main stream or similar to "nested" westernColorado where paleoflooddata are available.A va- sites.For each site, drainagearea, elevation,channel slope, riety of distribution functions and estimation methods are type of PSI (either flood bar (FB) or noninundationsurface availablefor estimatinga flood-frequencydistribution [NRC, (NI), width, depth,velocity, flood dischargecorresponding to 1999].Flood-frequency relations normally are developedusing the maximumPSI (e.g., Figure 6), and sediment-sizedata, a log-Pearsontype III frequencydistribution [Interagency Ad- where available,are presented.Site selectionwas dependent visoryCommittee on WaterData, 1981] referred to as Bulletin on findinggood PSIs (FB and NI) throughouta 50 to 300 m 17B (B17B) and the expectedmoments algorithm (EMA) length(length dependent on sizeof channel).Ideally, each site [Cohnet al., 1997;England, 1998]. B17B guidelineswere es- would have a FB and a NI surface.Sites were primarily in tablishedto provideconsistency in federal flood-riskmanage- straight,uniform reacheswhere local aggradationor degrada- ment for handlinglow and high outliers,for recognizingthe tion would be least. General scourappears to have been small need for regionalizedskew, and for zero-flowadjustment, for duringthe Holocene.According to Madole [1991a],there has example. been about a meter of degradationalong most the Yampa EMA is an efficient approachfor incorporatinghistorical River during the Holocene. Paleoflood evidence along the and paleoflooddata and usesthe log-PearsonIII distribution upper Yampa River primarilywas midchannelbars. These bars [Cohnet al., 1997;NRC, 1999].The NRC [1999]recognized the (islands)were interpretedas erosionalremnants of the late need to follow the spirit of the guidelinessuch as when using Pleistocenechannel rather than depositionalfeatures, and thus EMA. The EMA is used as the generalizationof the conven- maximumpaleofloods using present channel geometry may be tional log-spacemethod of momentsand makesmore effective slightlyoverestimated. Some streamshave had little bed ma- use of historicaland paleoflooddata in a censored-dataframe- terial movement(Table 2), and only NI evidenceexists, but if work [England,1998; NRC, 1999].EMA explicitlyincorporates the NI surface is consistent in the reach of river, then the the number of known and unknown dischargesabove and maximumpaleoflood discharge was consideredreliable. If a below a threshold,number of yearsin the historic/paleoflood reachof the samechannel had similarpaleoflood discharges at period, and knowledgeof the numberof yearswhen no large severalsites, the dischargeestimate was consideredmore re- floodshave occurred[Cohn et al., 1997;England, 1998]. The liable. For alluvialchannels the dischargewas consideredless difference between B17B and EMA is the treatment of historic reliable and reflectedin an assignedshort age of 100 years, andpaleoflood data [England,1998]. For B17B the gagerecord which is when muchof the channelarroyos developed in Col- is usedto fill in the censored(unknown) floods, whereas EMA orado [Pattonand Schumm,1975, 1981; Womack and Schumm, computesthe expectationsfor flow data below the threshold 1977]. and weightsthis value by the numberof censoredvalues [En- RD data using the criteria shownin Table 1 also are sum- gland, 1998].A comprehensivereview of the EMA, censoring marizedin Table 2. For eachsite the weatheringcharacteristics thresholdsand analysesof contemporaryand paleoflooddata for each of the RD methodsused, their numerical rating, is providedby England[1998]. estimatedage, and reliability (range) of age estimates,are For this study,several alternative flood-frequency distribu- provided.Assigning a numericalrating, an age, and range in tion estimateswere determined for the study sites in north- age for each RD method used is subjective;however, if all westernColorado using EMA to betteruse the longpaleoflood methodsused suggest similar numerical ratings, then the com- records.This analysiswas done for various combinationsof positeage estimate is likely more reliable.Although individual gage and paleoflooddata availableat each site. Low outliers RD agesare rather crude and may provide different relative were adjustedusing the B17B procedureto externallyelimi- agesof a surface,a compositerelative age using several meth- nate low outliersthat affectedthe fit of the upper end of the odsclearly enables one to distinguishdeposits of variousages. curveto the data, which is similar to dischargethreshold cen- Althoughthe error of individualRD agescan be +50% [Bir- soring [Cohn et al., 1997; Levish et al., 1994; NRC, 1999]. keland,1990], composite age is likely more accurate.For use 2968 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
Table 2. PaleofioodEstimates for Streamsin Elkhead Creek Basin and Nearby Streamsin the Regional StudyArea in
DA, Elevation, S, Width, Depth, Site Stream km 2 m m m- • m m
Yampa River 1 Yampa River at SteamboatSprings 1564 2041 0.01 38.6 2.6 Yampa River at SteamboatSprings 1995 peak 33.5 1.2 Yampa River at SteamboatSprings 1995 peak 2 Yampa River near Hayden--total 3704 1929 119 Yampa River near Hayden--MC 0.004 61 3.6 Yampa River near Hayden--OB 58 0.5 3 Yampa River at Craig--total 4481 1885 137 Yampa River at Craig--MC 0.003 76.2 3.1 Yampa River at Craig--OB 0.003 61 0.6 Yampa River at Craig 1995 peak 76.2 2.3 Yampa River at Craig 1995 peak 4 Yampa River near Maybell total 8806 1795 0.001 183 Yampa River near Maybell--MC 0.001 67.1 4 Yampa River near Maybell--LB 0.001 122 0.6 Yampa River near Maybell 1995 peak 0.001 51.8 3 Yampa River near Maybell 1995 peak
Williams Fork 5 Williams Fork near Hamilton total 1036 1905 Williams Fork near Hamilton--MC 0.01 30.5 1.8 Williams Fork near Hamilton--OB 22.9 0.6 Williams Fork near Hamilton 1995 peak 22.9 1.4 Williams Fork near Hamilton 1995 peak 6 Sulphur Gulch near mouth 5.7 1884 0.01 4.6 0.9 7 Ute Gulch near mouth 4.9 1899 0.01 6.1 0.6 8 Castor Gulch near mouth 10 1905 0.01 3 0.9 9 Morapos Creek near mouth 161 1905 0.01 9.1 1.5 10 Morapos Creek below Deer Creek 148 1945 0.01 9.1 1.5 11 Williams Fork aboveMorapos Creek total 829 1908 0.01 Williams Fork aboveMorapos Creek--MC 22.9 2.1 Williams Fork aboveMorapos Creek--OB 18.3 0.6 12 Williams Fork aboveMorapos Creek total 803 1914 0.01 Williams Fork below West Gulch--MC 22.9 1.8 Williams Fork below West Gulch--OB 15.2 0.5 13 West Gulch near mouth 4.7 1920 0.01 6.1 0.6 14 Waddle Creek near mouth 62 2012 0.01 9.1 0.6 15 Deal Gulch near mouth 7.8 1945 0.01 6.1 0.6 16 Jeffway Gulch near mouth 18 1954 0.01 6.1 0.6 17 Spring Gulch near mouth 9.1 1958 0.01 7.6 0.9 18 Williams Fork above Spring Gulch total 699 1958 0.01 Williams Fork aboveSpring Gulch--MC 15.2 1.8 Williams Fork aboveSpring Gulch--OB 16.8 0.8
Tributaries to 19 Dry Creek near Hayden 135 1932 0.01 7.6 1.5 20 StokesGulch near Hayden 32 1943 0.01 6.1 1.5 21 Sage Creek upstreamDam near Hayden 9.1 2243 0.01 3 0.6 22 Sage Creek dam failure 60 m downstreamtotal 0.02 Sage Creek--LB 9.1 0.9 Sage Creek--MC 12.2 2.7 23 Sage Creek dam failure 120 m downstream 0.01 30.5 1.5 24 Boxelder Gulch near Axial 39 1844 0.01 4.6 1.5 25 Milk Creek near Axial 272 1902 0.01 6.1 1.2 26 Stinking Gulch near Iles Grove 62 1939 0.01 15.2 0.9 27 Good Spring Creek near Axial total 104 1935 0.01 Good SpringCreek near Axial--MC 6.1 1.2 Good Spring Creek near Axial--OB 15.2 0.5 28 Good SpringCreek near Mount Streeter 91 1999 0.01 9.1 0.9 29 Wilson Creek near Axial 70 1920 0.01 6.1 1.2 30 Morgan Gulch tributary near Lay 9.8 1878 0.01 6.1 1.2 31 Morgan Gulch near Lay total 169 1850 0.01 Morgan Gulch near Lay--MC 12.2 1.5 Morgan Gulch near Lay--OB 15.2 0.9 32 Big Gulch near Lay 210 1890 0.003 15.2 1.2 33 Lay Creek near Lay at U.S. ,Highway40 259 1885 0.01 13.7 1.5 34 Lay Creek at Lay 479 1875 0.01 33.5 1.5 35 Pine Ridge Gulch near Craig 23 1899 0.01 9.1 1.5 36 Cedar Mountain Gulch near Craig 12 1897 0.01 9.1 1.2 37 Fortification Creek at Craig--1984 total 668 1887 0.002 FortificationCreek at Craig--1984 MC 10.7 3 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2969
Northwestern Colorado
Qgage Velocity, Q, Difference, Q, Q/A, Dbed, DFB, Age, Reliability, in s-1 m 3 s-1 % m 3 s-1 km-2 mm mm Type RD Method years years Remarks a
Basin 3.1 311 25 0.2 610 305 FB, NI S9, W9, M9, L9, B8 9OOO _+2000 2.4 100 -4.8 HWM 105 gage 651 25 0.2 FB, NI S9, W8, M7, L7, B7 9000 _+2000 2.7 593 1,2 2 58 687 30 0.2 NI S8, W8, M8, B9 5OOO 3OOO 2.6 614 1,2 2 73 2.1 372 11.4 HWM 334 gage 865 30 0.1 NI S8, M8, B10 5OOO 3OOO 1 2.7 729 1,2 1.8 136 1 2.1 337 -10.6 HWM 1 377 gage
Basin 178 30 0.2 FB, NI S6, W8, M6, L8, B5 5OOO _+1000 2.7 153 610 152 1.8 25 2.1 67 -10.7 HWM 75 gage 2.4 10 25 1.8 610 152 FB, NI S5, W10, M9, B9 5OOO _+1000 1,4 2.1 8 25 1.6 914 152 FB, NI S5, W10, M9, B9 5000 _+1000 1, 4 2.1 6 25 0.6 610 152 FB, NI S5, W8, M7, B8 5000 ñ1000 1 2.1 30 30 0.2 3O5 152 NI S5, W6, M5, B7 1000 _+500 1 1.8 25 30 0.2 3O5 152 NI S6, W5, M6, B6 1000 _+500 1 139 30 0.2 NI S5, W6, M5, B8 1000 _+500 2.4 119 610 152 1.8 20 127 30 0.2 NI S4, W7, M6, B7 1000 _+500 2.7 115 610 152 1.8 13 2.4 9 25 1.9 610 152 FB, NI S9, W8, M7, B7 5OOO ñ1000 1,4 1.5 8 25 0.1 3O5 102 FB, NI S8, W8, M7, B7 5000 ñ1000 1, 4 2.1 8 25 1 3O5 102 FB, NI S9, W8, M7, B7 5000 _+1000 1, 4 2.1 8 25 0.5 457 102 FB, NI S7, W8, M7, B7 5000 _+1000 1, 4 1.8 13 25 1.4 610 102 FB, NI S9, W7, M8, B7 5000 _+1000 1, 4 91 30 0.1 FB, NI S6, W8, M7, B8 5000 ñ1000 2.4 68 610 2O3 1.8 23
Yampa River 2.1 25 30 0.2 NI S8, W7, M6, B7 1000 ñ500 1,4 2.1 20 30 0.6 NI S9, W6, M7, B8 1000 ñ500 1, 2, 4 1.5 3 30 0.3 305 76 NI S7, W7, M6, B8 1000 _+500 1, 4 166 25 FB, NI S1, W2, M1, B1 3 2.7 23 3 4.3 143 762 762 3 4 184 25 762 762 FB, NI S1, W1, M1, B1 3 1.8 13 30 0.3 305 102 NI S8, W9, M7, B6 100 1000 1 2.1 16 30 0.1 305 102 NI S8, W8, M7, B5 100 1000 1, 4 2.1 30 30 0.5 305 152 NI S8, W9, M7, B7 100 1000 1, 4 26 30 0.3 NI S9, W8, M8, B6 100 1000 1 2.1 16 610 152 1 1.5 11 1.5 13 30 0.1 610 127 NI S7, W9, M8, B7 100 lOOO 1 2.1 16 30 0.2 NI S8, W9, M8, B7 100 lOOO 1, 2, 4 2.1 16 25 1.6 FB, NI S8, W9, M7, B9 5OO lOOO 1, 2, 4 75 25 0.4 FB, NI S9, W9, M7, B7 5OO lOOO 2.4 45 1,2 2.1 30 2.1 40 30 0.2 FB, NI S10, W9, M8, B9 5OO lOOO 1,2 2.4 51 30 0.2 NI S8, W9, M7, B7 5OO lOOO 1,2 2.4 125 30 0.3 NI S9, W9, M6, B8 5OO lOOO 1,2 2.4 34 30 1.5 NI S8, W9, M8, B7 5OO lOOO 1, 2, 4 2.4 27 30 2.3 NI S8, W8, M7, B6 5OO lOOO 1, 2, 4 80 25 0.1 HWM 2.1 69 1,2 2970 JARRETI' AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
Table 2. (continued)
DA, Elevation, S, Width, Depth, Site Stream km 2 m mm-1 m m
Tributaries to FortificationCreek at Craig--1984 RB 19.8 0.5 FortificationCreek at Craig 1995peak 129 1887 0.002 9.1 2.3 FortificationCreek upstreamCraig--1995 total 124 1893 0.003 27.4 FortificationCreek upstreamCraig--1995 MC 9.1 1.5 FortificationCreek upstreamCraig--1995 OB 18.3 0.3 39 Coal Bank Gulch near Hayden 2.1 1966 0.01 6.1 0.3 40 Trout Creek near Milner 518 1990 0.01 19.8 1.5 41 Walton Creek near SteamboatSprings 109 2161 0.03 18.3 1.5 42 BurgessCreek near SteamboatSprings 7.8 2225 0.05 9.1 1.2 43 SpringCreek near SteamboatSprings 18 2195 0.03 16.8 0.9 44 Soda Creek near SteamboatSprings 52 2057 0.01 13.7 1.2 45 ButcherknifeCreek near SteamboatSprings 10 2134 0.02 10.7 0.9 46 Fish Creek near SteamboatSprings 62 2188 0.04 12 1.2 Fish Creek near SteamboatSprings 1995 peak 13.7 0.8 Fish Creek near SteamboatSprings 1995 peak
Elkhead Creek 47 Elkhead Creek downstreamU.S. Highway40 near Craig 645 1905 0.003 15.2 2.7 Elkhead Creek downstreamU.S. Highway40 near Craig--1995 peak 15.2 2 ElkheadReservoir (1995 peak outflow) 48 Elkhead Creek downstream Elkhead Reservoir total 634 1907 0.002 70.1 Elkhead Creek downstream Elkhead Reservoir--LB 45.7 0.6 Elkhead Creek downstream Elkhead Reservoir--MC 15.2 2.1 Elkhead Creek downstream Elkhead Reservoir--RB 9.1 0.9 49 Brown Gulch near Craig 2.3 1948 0.01 3 0.6 5O Wadell Gulch near Craig 4.7 1957 0.01 9.1 0.6 51 Little CottonwoodCreek near Craig 2.1 1996 0.01 3 0.3 52 CottonwoodGulch near Craig 5.4 1987 0.01 3 0.6 53 Long Gulch at CountyRoad 18 near Craig 5.2 2012 0.01 3 0.8 54 Long Gulch at CountyRoad 29 near Craig 5.7 1996 0.01 3 0.9 55 Long Gulch tributaryat CountyRoad 18 near Craig 2.3 2012 0.01 3 0.5 56 Long Gulch at CountyRoad 36 near Craig 18 1914 0.01 3 0.9 57 Elkhead Creek upstreamElkhead Reservoir 443 1954 0.01 18.3 2.1 58 Calf Creek near Craig 30 2015 0.01 6.1 1.8 59 Elkhead Creek at CountyRoad 56 near Craig 246 2015 0.01 18.3 1.5 60 Elkhead Creek downstream North Fork Elkhead Creek 1 231 2076 0.02 15.2 1.8 61 Elkhead Creek downstream North Fork Elkhead Creek 2 total 231 2073 0.01 26.2 Elkhead Creek downstream North Fork Elkhead Creek 2--MC 12.2 1.7 Elkhead Creek downstream North Fork Elkhead Creek 2--OB 14 0.5
Elk River 62 Slate Creek near Milner 3.6 2085 0.02 3 0.6 63 Hot SpringCreek near Mad Creek 25 2042 0.01 6.1 1.2 64 Mad Creek near Mad Creek 97 2057 0.02 21.3 1.4 65 Big Creek near Mad Creek 98 2060 0.02 12.2 1.4 66 Elk River at Clark at gage 559 2213 0.01 30.5 2 67 Willow Creek near Elk Ridge 189 2256 0.01 15.2 1.5 68 Willow Creek downstream Steamboat Lake 130 2408 0.01 10.7 1.5 69 Hahns Peak tributary near Hahn's Peak 0.8 2694 0.01 1.2 0.6 70 Willow Creek downstream Hahns Peak Lake 18 2484 0.01 6.1 1.2 71 Willow Creek tributarynear Hahn's Peak 6.5 2493 0.01 2.4 0.9 72 Ways Gulch at Hahn's Peak 6.5 2460 0.01 3 0.8 73 Elk River at Elk Ridge 275 2268 0.01 19.8 2.4 74 Hinman Creek near Elk Ridge 35 2316 0.01 7.6 0.9 75 CoultonCreek near Elk Ridge 12 2310 0.01 6.1 0.9 76 Elk River near Hinman Campground 259 2329 0.01 18.3 2.1 77 North Fork Elk River at Middle Fork 104 2435 0.02 12.2 1.8 78 Middle Fork Elk River upstreamNorth Fork 54 2451 0.01 15.2 1.2 79 Elk River near Milner 1070 2045 0.01 52.9 1.8 Elk River near Milner 1997 peak 1075 2009 0.003 38.0 1.6 Elk River near Milner 1997 peak 1075 2009 0.003
Little Snake 8O King SolomonCreek near Columbine 30 2576 0.01 6.1 1.2 81 IndependenceCreek near Columbine 6.5 2621 0.01 3 0.9 82 IndependenceCreek tributary near Columbine 1.3 2624 0.02 2.4 0.6 83 Hahns Peak tributaryto IndependenceCreek 0.8 2688 0.1 2.4 0.3
White River 84 Piceance Creek at Rio Blanco 23 2219 0.01 10.7 0.6 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2971
Qgage Velocity, Q, Difference, Q, Q/A, Dbed, DFB, Age, Reliability, ms -! m 3 S-1 % % m3 s-• km-2 mm mm Type RD Method years years Remarks a
Yampa River 1.2 11 1.8 38 O.3 HWM 37 25 O.3 FB, HWM 2.1 30 1 1.2 7 1 1.5 3 30 1.4 NI S8, W9, M7, B6 100 lOOO 1, 2, 4 2.1 64 25 0.1 FB, NI S6, W6, M8, L8, B7 1000 lOOO 1, 2 3.8 lO5 25 1 914 305 FB, NI S8, W9, M9, L10, B8 9000 ñ2000 1, 2 3 34 25 4.4 FB, NI S7, W9, M9, L9, B6 6000 ñ1ooo 1,2 2.4 37 25 2.1 FB, NI S8, W9, M9, L10, B7 6000 ñ1ooo 1, 2 2.7 46 25 0.9 FB, NI S8, W9, M9, L10, B8 6000 ñ1ooo 1, 2 2.4 24 25 2.3 FB, NI S7, W9, M9, L8, B7 6000 ñ1ooo 1,2 3.1 45 25 0.7 1067 305 FB, NI S8, W9, M9, L10, B8 9000 ñ2000 1 2.4 25 4.2 HWM 24 gage
Basin 3 127 30 0.2 305 152 FB, NI S8, W9, M9, B7 5000 +_lOOO 1.8 55 -6.8 HWM 59 gage 135 30 0.2 FB, NI S8, W9, M9, B7 5000 _ lOOO 1.5 42 1 2.4 79 305 76 1 1.5 13 1 2.1 4 30 1.7 NI S6, W7, M6, L4, B8 100 lOOO 1, 2, 4 2.1 12 30 2.6 NI S6, W7, M5, L3, B8 100 lOOO 1,2,4 1.5 1 30 0.7 NI S8, W7, M6, L5, B5 100 lOOO 1, 2, 4 2.1 4 30 0.7 NI S6, W7, M6, L4, B8 100 lOOO 1, 2, 4 1.5 4 30 0.7 NI S7, W7, M6, L4, B6 100 lOOO 1, 2, 4 1.5 4 30 0.7 NI S5, W7, M6, L4, B5 100 lOOO 1, 2, 4 1.8 3 30 1.1 NI S6, W7, M6, L4, B8 100 lOOO 1, 2, 4 2.4 7 30 0.4 NI S6, W7, M6, L4, B7 100 lOOO 1,2,4 2.4 95 30 0.2 610 152 FB, NI S9, W9, M8, L8, B10 8000 ñ2000 1,2 2.1 24 30 0.8 NI S6, W7, M6, L4, B9 1000 ñ500 1, 2 3.1 85 25 0.3 762 260 FB, NI S9, W7, M6, L8, B8 5000 ñ1000 1, 2 3 85 30 0.4 914 254 FB, NI S8, W7, M8, L8, B9 5000 ñ1000 1 95 30 0.4 3.9 81 762 152 FB, NI S6, W8, M9, L4, B8 5000 _+1000 2 14
Basin 2.1 4 30 1.1 FB, NI S6, W7, M6, B8 500 1000 1,2,4 2.4 18 25 0.7 FB, NI S9, W9, M8, L9, B9 8000 ñ2000 1,2 2.4 71 25 0.7 914 305 FB, NI S7, W8, M9, L10, B7 8000 ñ2000 1 2.4 41 25 0.4 762 305 FB, NI S8, W7, M6, L8, B9 8000 ñ2000 1 4.4 268 25 0.5 762 381 FB, NI S9, W9, M9, L9, B9 9000 ñ2000 1 2.7 64 30 0.3 FB, NI S7, W8, M7, L9, B8 7000 ñ2000 1,2 2.1 35 30 0.3 FB, NI S6, W8, M9, L7, B9 5000 ñ1000 1,2 1.8 1 30 1.7 FB, NI S6, W7, M6, L4, B9 1000 1000 1, 2 2.1 16 30 0.9 NI S5, W7, M8, L8, B8 1000 1000 1, 2 2.1 5 30 0.7 FB, NI S5, W7, M6, L4, B6 1000 1000 1, 2, 4 1.8 4 30 0.7 FB, NI S4, W8, M6, L8, B9 1000 1000 1, 2, 4 3 147 30 0.5 FB, NI S8, W7, M7, L10, B8 7000 ñ2000 1, 2 2.4 17 25 0.5 FB, NI S9, W8, M9, L8, B8 7000 ñ2000 1, 2 2.4 14 25 1.1 FB, NI S8, W8, M6, L7, B9 6000 ñ1ooo 1,2 3 119 25 0.5 1524 254 FB, NI S9, W7, M8, L9, B8 8000 ñ2000 1 3 68 25 0.7 FB, NI S7, W8, M9, L10, B7 8000 ñ2000 1,2 2.7 51 30 0.9 FB, NI S6, W7, M6, L7, B8 8000 ñ2000 1,2 2.5 238 30 0.2 FB, NI S6, W8, M9, L7, B9 5000 ñ1000 2 3 182 11.7 25 0.2 HWM 2 163 gage
River Basin 2.1 16 30 0.5 FB, NI S8, W7, M9, L7, B8 5000 ñ 1000 1, 2 1.5 4 30 0.7 FB, NI S6, W6, M7, L8, B7 5000 _+1000 1, 2 1.5 2 30 1.8 FB, NI S8, W7, M7, L7, B9 5000 _ 1000 1,2 1.8 1 30 1.7 305 76 FB, NI S5, W7, M6, L8, B6 3000 ñ 1000 1
Basin 2.1 14 30 0.6 NI S5, W7, M4, L5, B4 500 lOOO 1, 2, 4 2972 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
Table 2. (continued)
DA, Elevation, S, Width, Depth, Site Stream km 2 m m m -t m m
White River 85 Sheep Creek near Meeker 49 1905 0.01 7.6 1.5 86 Flag Creek near Meeker 135 !920 0.01 9.1 1.5 87 White River near Meeker 2093 1905 0.003 45.7 2.4 White River near Meeker 1995peak 2093 1905 41.1 1.7 White River near Meeker 1995peak 2093 1905 88 Curtis Creek near Meeker 41 1948 0.01 9.1 0.9 Abbreviationsare as follows: S isestimated water slope at site.Percent difference isQgage percent difference of 1995peak discharge estimated usingpaleoflood techniques from high-watermarks and at the streamflow-gagingstation independently obtained from a well-definedstage- dischargerelation. Q% is estimatedtotal uncertainty of paleoflooddischarge estimate given in percent.Q/A is unitdischarge. Dbed is maximum particlesize on the streambedavailable for transport.DF• 3 is maximumparticle size transported to the floodbar. Type is typeof evidenceused to determinepeak discharge. FB is boulderyflood bar, NI is noninundationsurface, HWM, high-watermark, and GAGE is streamflow-gaging station.RD methodis the method used which considers degree of soildevelopment (S), surface-rockweathering (W), surfacemorphology (M), lichenometry(L), andboulder burial (B), althoughnot all methodscould be usedat eachsite. Numerical values from Table i are listedfor each RD method.Age is the compositeage for thepaleoflood record length. Reliability is the estimateduncertainty of compositeage; positive value onlyindicates age may be longerby thisamount. Abbreviations with streamsare asfollows: DA, drainagearea; MC, mainchannel; OB, overbank; and LB, left bank. aDefinitionsof numeralsare asfollows: 1, no substantialflooding on floodplain;2, watertoo deepto estimateparticle size or unavailable;3, flood causedby dam failure; and 4, underfit stream. here in attemptingto identifythe paleofloodrecord length for velopmentsometimes were mappedas freshalluvial deposits. the largestflood during the Holocene,such uncertainties can Thiswas attributed to the fact that the primarypurpose of soil be addressedin the EMA flood-frequencyanalyses. surveysis the determinationof agriculturalproductivity or Soil development,boulder weathering,and surface mor- buildingsite potential. Thus floodplains may have received less phologyseemed to have the best consistency(Table 2) and attention during mapping. Boulder burial had a fair consis- thus potential for assigningages either individuallyor com- tency in the studyarea. However, it appearedto have more bined.Available soil and surficialgeology reports for the study variability;because compared to boulder burial data in the area [Soil ConservationService, 1982; Madole, 1982, 1989, Front Rangeof Colorado[Waythomas and Jarrett, 1994], there 1991b, 1991c;Natural ResourcesConservation Service, 2000] is usuallymuch more fines depositedaround the cobble and were helpful in assigningages, particularly for NI surfaces. boulderymaterial during flooding. Thus there is some"age" However,near-stream alluvial deposits with extensivesoil de- immediatelyafter flooding.Lichenometry, while having poten-
..
:::
.....;. .,,:.:.,,....,:..::..: :z•. ':"•';'::•:'•%:'...... ¾..'"%.,, :...:::.:...... •...... ;•;...... •:.?-,;:?.•:;:•!-:-::-':!•:??:;,:.::',•:.--:...... ,...:':...::::""-:-:-%(•:.:...... :•..<;:•...•,:•,:..::..•.:•...... :,•...... •: ..... ::.;"'"'•'.:•½•'4.•--.:•4•---•--%•:::.;.5'•'."-5::""'-?;:•;• ?½;--•.:..:-•. .... ' ::'::':::'::•"::•½•;•*•..'•:::,•:.--..,: ..... ß::• .. '..... •t:?.½:;:•;•2;•-•:.f%•:::.-,,.•.-•;•:.•. :':'::%•"½. "*.•:... •-:..;•::"• ::....::.,•..•;;• .- ...... ::•. .;::;;-:•'.:...:
;•"•'•.•'%.•&:D:½•,::'--:,½•':•,.,•.-..:a%.?½ •:•.,,.•:• ...... :• :'"' •;..,.:q'•t•,:,•:-,•;:,;: ..... 5•;;•;•;•;•:;•..%..;•.•::..•:W:&•*`•,•4•X•?i;•;•:%•..•.•.•.•.. :'*:'½;•-'•:•½•ii'i'i' %?' ,..,½•.:,•...... •....•:,.<.. . :½•:•:::::::':.•.,...•(•;•:.•.:•*.•.•;•.•.•..:.•:.....•.•.•3..•.•.•..::•:•:m• .:•,• :,½:z•::.,•:-:•.•..%•.•. :5...... •'• '½½'"•:'•'••:•-:-.,-.-.--:-::::-...... •?:.•:•;•::4•:•<½i•::•. .•,.•-•:;..:.-•..%:;• ....•',•:'":•'•-"'•'• (•.:r..., :• ::.•.'"•":'•-•;•::;•:::.,:---&:::*:,-•:k:-:••.:½.•}•:•.:•8:--•---•..-:.:.•.:..
...,.,•:•::.•;•::• ...... :.:..... •.::•½;•.•.....•., .. .•...... •.::.,.,...... ,:,.½...... :"":'"'::':':"•:':•:'"'""':'"'"....:...... :. %ß/½• " ...:-.
Figure 6. Small,low-relief, flood-bar deposits define the maximumpaleostage indicator (PSI) for Elkhead Creek downstreamfrom North Fork Elkhead Creek. The view is downstreamand toward the left bank; line denoteslocation of crosssection in Figure7. Well-developedalluvial and colluvial soils (Table 2) on thevalley floordefine the noninundation surface forthe maximum PSIrange. Paleoflood discharge is85 m 3 s- • (_+25 %) in about5000 years(+ 1000years). JARRET'F AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2973
(continued)
Qgage Velocity, Q, Difference, Q, Q/A, Dbed, DFB, Age, Reliability, in s-1 m 3 s-1 % % m3 s-• km-2 mm mm Type RD Method years years Remarks a
Basin 2.4 28 30 0.6 NI S9, W5, M6, IA, B8 100 1000 1, 2 2.1 30 30 0.2 NI S8, W7, M6, L7, B9 100 1000 1, 2, 4 3.1 340 25 0.2 FB, NI S8, W9, M8, L7, B9 5000 + 1000 1, 2 2.3 158 5.3 HWM 150 gage 2.4 20 30 0.5 NI S5, W8, M6, L5, B7 100 1000 1, 2, 4
tial for datingrecent flood surfaces,reveals the least informa- teorologicevent. This is supportedby the maximumpaleoflood tion because of numerous factors affecting lichen growth. dataof about3 m3 s-• for SageCreek immediately upstream Thesefactors made it difficultto estimaterelative ages, and the from the dam (site 21 in Table 2). method is likely limited to agesconsiderably less than 3000 Intermittent streams in northwestern Colorado show little years for most of the study area. The weathering of flood evidence of substantial runoff and are underfit streams for the bouldersand lichen colonizationof boulder surfacesmay be basin size and broad valleys.An underfit stream is one that complicatedby the effectsof forestand rangefires [Birkeland, appearstoo small to have eroded the valley in which it flows. 1984; Bierman and Gillespie,1991]. If the outer part of the Valley bottoms are relativelybroad and completelycovered boulderspalls after a fire, the boulderweathering "clock" is with native grassesand often have well-developedsoils. It is essentiallyreset, and rock-weatheringfeatures record the time unlikelythat a channelcould undergodegradation and aggra- elapsed since the last episode of fire. If spall evidence is dationwithout leavingany evidencesuch as terracesand with- present,then rock-weatheringand lichen-coverdata provide out showingevidence of substantialage (thousandsof years) only minimum age estimates.No spalledsurfaces on rocksor on the valleyfloor. A lack of channeldevelopment is due to (1) spall detrituswas evident at sitesin the studyarea. Age reli- little seasonalsnowpack [Doesken et al., 1984] and thus rela- ability(ranges) for alluvialchannels with arroyodevelopment tivelylittle snowmeltrunoff versushigh mountainstreams and (e.g.,site 24 in Table 2) is an estimateof the conservativeage. (2) the basinlocation above the elevationfor substantialrain- Becauseof the scopeof this investigation,use of field soil fall runoff. morphologyand developmentindexes [Bilzi and Ciolkosz, It is particularlynoteworthy that for many channelshaving 1977; Burke and Birkeland, 1979; Meixner and Singer, 1981; coarse-grainedbed material, these sedimentshave not been Harden,1982] were not used,but theycould provide better age mobilizedand depositedas flood bars and slack-waterdepos- control. its. Data on maximumparticle size in the channels(Dbed) and No evidenceof substantialflooding was found in any inves- on flood bars (DFB) presentedin Table 2 help demonstrate tigatedstream in Elkhead Creek basinor streamsin the north- lack of flow competence.In all but the very fined-grained westernColorado study area. If substantialflooding were com- channels,the largestparticles in flood bars are smaller than mon in the study area, evidence should be present. The particlesavailable for transportin the channels,suggesting that maximumpaleoflood discharge for four sitesin Elkhead Creek large floodshave not occurredduring the Holocene.The few upstreamfrom Elkhead Reservoir ranged from 79 to 95m 3 s- • in-channelbars that exist are small, have low relief, and par- (sites57, 59, 60, and 61 in Table 2). Consideringthe estimated ticle sizesof cobbleor smaller,and suggestinsufficient flow to total uncertaintyassociated with each paleoflood discharge mobilizereadily available streambed material (Figures6 and ("Q, %" in Table2), thebest estimate is 85 m3 s-• +_25% in 7). Similarly, for fine-grain material streamsthe in-channel abou[5000 years (_+ 1000 years). No tree scarringor flood- bars also are poorly developedand exhibit low relief. Coarse transportedwoody debris was identified on floodplainsurfaces, material in the streambedof many streamsin northwestern exceptassociated with the SageCreek Dam failure flood near Colorado (derived from conglomerate,basalt, and Precam- Hayden(sites 22 and 23 in Table 2) andin lowlandareas along brian rocks)is slightlyreworked glacial outwash gravel (e.g., the lower Elkhead Creek and Yampa River downstreamfrom Figure 6). Had substantial flooding taken place, coarse aboutMilner. The peak dischargeresulting from failure of the streambedmaterial would be transportedonto the floodplain damwas about 175 m 3 s-• (averagefor sites22 and23), which [e.g.,McCain et al., 1979;Jarrett and Costa,1988; Jarrett, 1990b; providesan analogfor PSIs for a largeflood in the studyarea. Waythomasand Jarrett,1994] such as shownin Figures3 and 4, Local residentsreported that the dam failed in the mid-1980s whichwould be preserveduntil a largerflood emplacedhigher from seepagethrough the dam and was not related to a me- deposits. 2974 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
through 1990 [Jarrett,1990b] and was updated through 1997
Colluvial (angular) with recentdata [McKeeand Doesken,1997]. Although there boulders hasbeen an extensiveprogram for documentingextreme rain- 5 • J Well.developed•,•/ storms in Colorado, there have been few intense flood- / producingrainstorms documented in northwesternColorado c• [] X / alluvialsoils % / - (triangleson Figure8). The maximum24-hour rainfall data for 181 stationsand bucketsurveys in southwesternColorado also ß> •2•7;_2•;;' v•:'.7 "• are shown on Figure 8 for comparisonto help define the northwesternColorado region. The maximum24-hour amount for northwesternColorado is 82 mm (Meeker). Maximum 2 • oøO• øoo• monthlyvalues for northwesternColorado only slightly exceed oo o o o •o record maximum 24-hour amounts for southwestern Colorado I • Pleistocene-outwashgravels 0 10 20 30 40 (Figure 8), dramaticevidence of large relative differencein Stationing, m flood-producingrainfall from northwesternto southwestern Colorado. The maximum rainfall amount for southwestern Figure 7. Graph of channelcross section for El•cad Creek Colorado is about 150 mm in a few hours for Sweetwater •11•L1•111• ...... •11•111 •*•1N•lLll Fork E•ead Creek Dl1•11• .... •1:- •1•U1••: .... 6. Creek (1976) and for Dove Creek in 24 hours(1972) [McKee and Doesken,1997]; these areas are influencedby the flow of moist air from the southwest[Collins et al., 1991].Maximum During on-sitevisits, peak dischargeusing 1995 HWMs also 24-hour rainfall in southwesternColorado is somewhatlarger was estimated for seven sites at streamflow-gagingstations than in northwesternColorado but is substantiallyless than the (Table 2, type is "gage").These estimatesusing the critical- maximum6-hour rainfall amountsof up to 610 mm in eastern depth and slope-conveyancemethods were comparedto the Colorado. Of particular interest in western Colorado,maxi- peak dischargefor 1995 obtainedfrom gagerecords to help mum 24-hourprecipitation amounts fell as snowand are pre- assessthe accuracyof paleodischargeestimation methods. The sented as snow-waterequivalent (SWE) (Figure 8, open estimateddischarges generally are within about 10% of the squares). gageddischarge (Table 2, "Q" and "Qgage Difference" col- A subjective,indirect indicator of the occurrenceof intense umns)thus showing that the methodsused to estimatepaleo- rainfall and associatedflooding is the developmentof rill flood dischargeare reliable. However, additionalsources of (light) and or gully(deep) erosion on hillslopes[Jarrett, 1990b; uncertainties(e.g., channel change) are more difficultto quan- Jarrettand Browning,1999]. Generally, erosionpotential in- tify. Attemptsto estimateother uncertaintiesin dischargeare creasesas slopesteepens; steeper slopes usually require com- reflectedin terms of dischargeuncertainty (Table 2, "Q, %" parativelysmall amounts of rainfall before substantialerosion column)and age (Table 2, age). occurs,although erosiveness also dependson the type of soil, Paleoflood estimatesin bedrock channelsgenerally were assignedan uncertaintyof 25%, whereasalluvial channels were assignedan uncertaintyof 30%. Paleofloodestimates in allu- 600 ! ! vial channelswith minor arroyo development,though the age ß NWCO 24-hr of NI surfacesare very old, were assignedan age of 100 years 24-hr PMP ß NWCOmonth 500 o SWCO24-hr to accountfor historicalarroyo development.Additional con- [] W CO 24-hr SWE fidencein paleofloodestimates is exhibitedwhen multiple sites • 100-yr, 24-hr 400 ..... 10-yr, 24-hr are used and resultsare similar. For example,for Elkhead E Creek downstreamfrom Elkhead Reservoir,paleoflood esti- E 1976 Sweetwater matesranged from 127 m 3 s-• to 135m 3 s-• (sites47 and48 • 300 in Table 2), a differenceof about 3%. Similarly,paleoflood ,_ •' 200 estimatesalong Elkhead Creek upstreamfrom the reservoir 6-hrPMP / • ßo 1972DoveCreek increaseconsistently from 79 m3 s-• to 135m 3 s-1 with in- ß [] _ [] 100 o o o(• -- o ..,,.•_ bl Pi creasingdrainage area. g •, o•o__.•o••,.&." ...... 9.. [] Streams in northwestern Colorado have few coarse flood 0 ' ,o ,o depositson the floodplain.Where present,the depositsare 1000 2000 3000 4000 either associatedwith the recordsnowmelt flooding in 1984,or Elevation, m the depositsare very old (Table 2). The fact that streamsin Figure 8. Maximum24-hour and maximummonthly precip- northwesternColorado have no substantialpaleoflood evi- itation for northwestern(NW) Colorado and maximum24- denceis important,because it indicatesthe lack of substantial hour precipitationfor southwestern(SW) Colorado [Jarrett, flooding in Elkhead Creek basin is not due to chance.The 1987, 1990b;McKee and Doesken,1997]. Two of the largest paleoflooddata then were used to help define the regional southwesternColorado rainstorms (SweetwaterCreek and maximumflooding and for flood-frequencyanalyses in north- Dove Creek) are noted.It is importantto note that numerous western Colorado. large snowstormsreported as snow-waterequivalent (SWE) accountfor someof the largest24-hour precipitationamounts 5.2. RegionalAnalyses of Maximum Rainfall and Flood Data in all of western (W) Colorado.The 10-year and 100-year, 24-hour duration rainfall amounts [Miller et al., 1973] are 5.2.1. Maximum rainfall. A relation between maximum shownto place contemporaryrainfall data into a frequency 24-hour rainfall and elevationfor the studyarea in northwest- context.The 6-hour and 24-hour duration probablemaximum ern Coloradois presentedin Figure 8; this relation was con- precipitation(PMP) values[Hansen et al., 1977]are alsoshown structed from documented rainstorms from about 1900 for comparison. JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2975
10000 ...... i ...... ! ...... i ...... i ...... i ...... United States E. Colorado (Costa)
1000
E---...... <2,300 100
10 j/ •., • ,.•,•x•,•"xxx -i• NWColo. rary) //•. •.'Z... -•"•,.&7,•, ' --" // ," ""-' _._x.... 1
x • x x •! Ix gage x • • I' ungagedsite x x '• i I' paleoflood<5000yrs f IApaleoflood >5000yrs. .01 " ß1 I I 0 100 1000 10000 100000 Drainagearea, km 2 Figure 9. Relation betweencontemporary and paleofloodpeak dischargeand drainagearea with envelope curvesfor northwesternColorado. Envelope curves of maximumflooding for easternColorado [Jarrett, 1990b] and for the United States[Costa, 1987a] are shownfor comparison. vegetationcover, and infiltrationrate [Gilleyet al., 1993]. For similar to basins at lower elevations in eastern Colorado that example,as little as 25 to 50 mm of rain in a few hours can are subjectto large, intenserainstorms. produce rill erosion on bare, poorly drained soils on steep Rainfall-frequencyrelations developed for Colorado[Miller slopes[Hadley and Lusby, 1967;McCain et al., 1979;Jarrett, et al., 1973] can be usedto assessthe frequencyof contempo- 1990b;Jarrett and Browning,1999]. Thus a lack of rill erosionis rary rainfall data. Superimposedon Figure 8 are the 10-year a good indicator that intense rainfall is uncommon or that and 100-year, 24-hour duration rainfall frequencyrelations erosionhealing rates are high. Extensiverill and deep gully developedfrom Miller et al. [1973]along an east-westtransect erosion are commonin areas subjectto intenserainfall. For from the crestof the Park Rangeto Maybell. For comparative example,gullies up to 2 m deepformed on hillslopesin the Big purposesthe 6-hour and 24-hour PMP estimatesfor Elkhead ThompsonRiver basinwhere rainfall exceeded150 mm in a Reservoirare shownon Figure 8. Hansenet al. [1977, Figure few hoursduring the rainstormof July31, 1976 [McCainet al., 5.7] comparedthe ratio of the 24-hourPMP estimatesto 100- 1979].Jarrett and Browning [1999] used geomorphic techniques year, 24-hour rainfall frequency estimatesfor the western to relate hillslope erosionwith rainfall data for an extreme United States(e.g., for Coloradousing Miller et al. [1973]); rainstorm on July 12, 1996, in Buffalo Creek, located in the they suggestreasonable ratios between 2.8 and 5. For Elkhead foothillsnear Denver. Their geomorphicestimated maximum Reservoirthe ratio is 8.4 (510 mm/61 mm). Although there hourly rainfall of 115 mm, which was determinedimmediately may be someuncertainty in estimatingrainfall frequencies,the after the storm,compared with 130 mm independentlyderived ratio adds further supportto the conclusionthat PMP esti- in 1998 from Doppler radar signaturesand upper air observa- matesfor the ColoradoRockies may be too large. High moun- tions [Henz, 1998]. Part of the subjectivityin using hillslope tain barriers(Figure 1) reducethe availableatmospheric mois- erosionis estimatingthe time rills and gulliesremain [Jarrett ture from the Pacific and Gulf of Mexico to northwestern and Browning,1999]. Rill and gully networksformed during Colorado[Tomlinson and Solak, 1997]. extreme rainstorms such as in 1965 and 1976 in eastern Colo- 5.2.2. Maximum flooding. Recordsfrom 198 streamflow- rado [McCain et al., 1979; Matthai, 1969] and west central gaging stationsin northwesternColorado, primarily in the Colorado[Jarrett, 1990b] have changed little in the intervening Yampa River and White River basins,were analyzed;some years.Conversely, in regionsof the Rocky Mountainsnot sub- have peak-flow data since the early 1900s. These gagesare ject to intenserainfall, there is a general sparsityof hillslope fairly uniformly distributedin the studyarea. To help define erosionevidence on slopeswhere evidenceshould have been the maximumflood potentialfor northwesternColorado, flood preservedhad large rainstormsoccurred. data from 20 ungagedsites that definemaximum flooding from On-siteinspection indicated that rills and gulliesare smallor intense,localized rainstorms in northwesternColorado [Jarrett, nonexistent in basins above about 2000 m in northwestern 1987, 1990b;data availableat http://waterdata.usgs.gov],also Colorado.Lack of rilling throughoutsuch a largearea provides were incorporatedinto the database.Maximum peak discharge additionalsupporting evidence of the absencesubstantial rain- is 940m 3 s-•, drainageareas ranged from 0.21 to 19,840km 2, stormsin recent times.Hillsides having sparse vegetation and gageelevation ranged from 1595 to 3200 m, and there were a comprisedof sand or finer-grainedsoils such as in Piceance total of 3512 stationyears of record. A relation of maximum Creek and Yellow Creek basinshave rill and gullyerosion, but discharge,including paleoflood data, and drainagearea with these basinsare in the far southwesternpart of the regional the envelope curve for northwesternColorado is shown in study area. However, no hillslopeshave gully development Figure9. Thelargest gaged rainfall-produced flood of 190m 3 2976 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
4O ...... i ...... , ...... l ...... i ...... i ...... extremerainstorms (intensity, amount, and size) and thusse- ...... vere flooding.Above about 2300 m in northwesternColorado, • [' ungaged unit dischargesare slightlyhigher than east of the Continental • Ea•]•rrrnett•o'o.[_• _paleofloo.••fl _. Divide, which reflects the maximum snowmelt runoff from the E 30[ Park Range. E
= 20 5.3. Flood-FrequencyAnalysis Flood-frequencyrelations with EMA for selectedstreams in ._• northwesternColorado (Table 3) were developedusing the
'•= 10 PiceanceCrTxrib. • recordedannual peak-flow data and paleoflooddata (Table 2). Yellow_Cr.•4 • NWColorado LongLakeInlet Flood-frequencyanalyses were done usingthe paleofiooddis- charge,which was varied by the estimateduncertainty (e.g., site
o ...... l.•_m .... 61 in Table2, 95 m3 s-• _+25% for theElkhead Creek gage). 10oo 1500 2000 2500 3000 3500 4000 The paleofloodrecord length of time (age) and the age reli- Elevation, m ability(range) for the maximumpaleoflood (Table 2) wasused Figure 10. Relation between maximum unit dischargeand to define the paleofloodrecord length in the analysis(e.g., elevationwith envelopecurve for northwesternColorado and 5000 _+1000 yearsfor the Elkhead Creek gage). easternColorado [Jarrett, 1990b]. Becauseregional skew estimates [Interagency Advisory Com- mitteeon WaterData, 1981]were developedover 30 yearsago and do not incorporatepaleoflood data, two EMA runswere s-• occurredin Yellow Creek near Rangley(streamflow- made.The first EMA runsused station skew (Table 3). Then, gagingstation 09306255, drainage area of 679km2). This flood to assessif regional skew may affect results, station skews was a hyperconcentratedflow that resultedfrom a localized (Table 2) were reviewedto assessif regional skew relations rainstormstorm over lessthan about50 km2 of the steep, with stationdrainage area, period of record (stationand pa- sparselyvegetated basin (U.S. GeologicalSurvey, unpublished leoflood record length), and gage elevation.There were no data, 1978). For riversdraining higher mountain areas in the statisticallysignificant relations, perhaps because of usingonly studyarea, peak flowsare dominatedby snowmeltrunoff. For eight stations,homogeneity of the study area, or the narrow comparison,the envelopecurves for streamsbelow 2300 m in range of skew about zero (-0.15 to +0.17) for these sites, easternColorado [Jarrett,1990b] and for the United States which suggestspaleoflood data may provide a stable at-site [Costa,1987a] also shownon Figure 9 help demonstratethe skew.Therefore a secondset of EMA runswas made using the lower-magnitudeflooding in northwesternColorado. Maxi- arithmeticaverage of the at-siteskew values as a regionalskew mum floodingin easternColorado is about3 timeslarger than (-0.03) (Table 3), whichis essentiallya lognormaldistribution, for similarlysized streams (>-3 km2) in northwesternColo- and these are consideredthe preferred curves. rado. Maximum flooding in eastern Colorado streams is The 95% confidencelimits were approximatedusing the slightlysmaller than maximum flooding in the United States B17B approach.Confidence limits only reflect parameterand (Figure 9). peak dischargeuncertainties; uncertainties such as best model, The envelopecurve (Figure 9) of maximumflooding can be representativedata, proper identificationof censoringthresh- usedto estimatethe hypotheticalmaximum flood for Elkhead olds,and selectionof properskew are more difficultto quan- Creek at Elkhead Reservoir.For a drainage-basinsize at the tify and were not included.In addition, the effectsof climate reservoir(531 km 2) the corresponding maximum flood is about change(natural or anthropogenic)may be the greatestsource 240m 3 s-•. The maximumpaleoflood estimate of 135m 3 s-• of uncertainty,and they are difficult to quantify[NRC, 1999]. (Table 2, site 48) for Elkhead Creek downstreamfrom Elk- Thus confidencelimits do not reflect the total uncertaintyin head Reservoiris 56% of the envelopecurve value. the frequencyanalysis (limits are too narrow), but no method The maximumunit dischargefor streamsin northwestern is currentlyavailable to make suchan assessment. Colorado(Figure 10) is 5.2 m3 s-• km-2 for PiceanceCreek Estimated flood quantileslisted in Table 3 for the eight tributary(2.8 km 2) near Rio Blanco(streamflow-gaging station streamflow-gagingstations reflect the averageage and average 09306042)resulting from a localizedrainstorm [Jarrett, 1987]. dischargefor the maximum paleoflood (Table 2). Flood- Four other smallstreams have had unit dischargesgreater than frequencyrelations for EMA for Elkhead Creek near Elkhead 3 m3 s -• km-2 resultingfrom intense, localized rainfall (Figure incorporatingpaleoflood data (the rectanglebrackets the likely 10). The largestunit dischargein the highestmountains in the rangeof dischargeand age for the maximumpaleoflood) and Park Range(Long Lake Inlet, Figure 10) is locatedin the area stationskew are shownon Figure 11. EMA resultsusing the of maximumsnowfall and representsmaximum snowmelt run- regional(average) skew of -0.03 alsoare listedin Table 3 and off in northwesternColorado. For comparison,maximum unit shownon Figure 11. It is not surprisingthat EMA frequency dischargeis about38 m3 s-• km-2 for smallstreams (<--•10 relationsusing the regional skew are not that different from km2) below2300 m in easternColorado; the envelopecurve the stationrelations (-4% differencefor the 10,000-yearflood for eastern Colorado is provided for comparison [Jarrett, in Table 3) becausethe at-site skewvalues have smallvaria- 1990b]. Such a small maximum unit dischargein northwest tion. Coloradois significantin that the stormoccurred in the Yellow A flood-frequencyrelation is needed at Elkhead Reservoir; Creek and PiceanceCreek basin where steep hillslopeswith however,there is no streamflow-gagingstation close enough to sparsevegetation exacerbate runoff. Althoughmaximum unit the reservoirto transfer(scale) the gagedfrequency estimates dischargegradually decreases with elevationin northwestern with the commonlyused drainage-area ratio approach[Hosk- Colorado(Figure 10), the decreaseis muchmore pronounced ingand Wallis,1997]. Regional flood-frequency relations avail- in eastern Colorado, where lower elevationsare subjectto able for western Coloradowere developedby Kircheret al. JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2977 2978 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
10000
Peak Discharge .... 95% confidence limits
PMF 1000- Maxpaleofloods Kircheret al. Reservoir envelopecurve for ' ,• .... (at dam) Gage \ .... '1 ElkheadCreek (fig. 9) ,•yr•= o...... t'•E ...... ;,•..,...•.•_ ...... o) • 100
._•
n • EMA(P=5,000 yrs; • ave.skew=-0.03• EMA(P=5,000 yrs; 10 ' stationskew)
i ...... ' , ß , , , , , 11 , , , , .... i ...... 1 10 100 1000 10000 Recurrence Interval, years Figure 11. Flood-frequencyrelations for Elkhead Creek at Elkhead Reservoir[Kircher et al., 1985]with paleoflooddata (shownas a rectangle)at the reservoir;near Elkhead gage(0924500) with paleoflooddata (shownas a rectangle)using at-site and averageskew with 95% confidencelimits. P denotesthe paleoflood lengthof record.The envelopecurve value (Figure 9) for maximumflooding in the Holoceneis alsoshown. The flood-frequencyrelation and probablemaximum flood for Elkhead Reservoir(Ayres Associates,Inc., Fort Collins,Colorado, unpublished data, 1996) are shownfor comparison.
[1985] that can be usedto estimatethese relations at an un- (Figure 11 and Table 3), whichhas about 30% of the drainage gaged site such as Elkhead Reservoir.Kircher et al. [1985] area at the reservoir,is similar to the regionalrelation. The developedregression relations for 33 flow characteristics, similarityof the relationsare likely becausethe majorityof the which include mean annual and mean monthly discharges, basin betweenthe gage and the reservoircontributes little flow-durationseries, peak discharge,and minimumand maxi- additionalpeak runoff from snowmeltand rainfall. mum 7-day dischargesfor recurrenceintervals from 2 to 500 Also shownon Figure 11 is the regionalenvelope value of years.These relations were developedfor four hydrologically maximumflooding for Elkhead Reservoirfrom Figure 9. For distinctregions in westernColorado usingrecords from 264 comparativepurposes the PMF estimateshown for Elkhead streamflow-gagingstations by relatinggage flow characteristics Creek at the dam (AyresAssociates Inc., written communica- with basin physiographicand climatic explanatoryvariables. tion, 1996)is 1020m 3 s-•, whichis 4 timeslarger than the The regressionrelations developed from 67 gagesin north- envelopecurve discharge value for ElkheadReservoir, and the westernColorado [Kircheret al., 1985, Table 8], which are a recurrenceinterval for a flood the magnitudeof the PMF functionof drainagearea and mean annualprecipitation, were exceeds10,000 years (Figure 11). The maximumpaleoflood in usedto estimateflood-frequency relation for ElkheadCreek at the last 5000 yearsfor Elkhead Creek near Elkhead Reservoir Elkhead Reservoir. The regional regressionrelations for is about 13% of the site-specificPMF estimate. northwestern Colorado have a mean standard error estimate of 63% [Kitchefet al., 1985]. At ElkheadReservoir the drainage area is 531 km 2, and the 6. Discussion mean annualprecipitation for the basinis 635 mm. The flood- Paleofloodtechniques and rainfall-runoffmodeling (includ- frequencyrelation for Elkhead Creek at Elkhead Reservoir ing PMP/PMF methods)have inherentassumptions and limi- (Figure 11 and Table 3) was estimatedusing the work of tationsthat produceuncertain flood estimates.Although pa- Kircheret al. [1985]. The frequencycurve in Figure 11 was leoflood estimatesalso involve uncertainties, the estimatesare extendedlinearly. The 100-yearflood estimatefor Elkhead basedon interpretationsof physicaldata preserved in channels Creekis about61 m3 s- • at ElkheadReservoir [Kircher et al., and on floodplainsduring the past 5000 to 10,000years in 1985]and compares with 64 m3 s-1 at the gageusing EMA northwesternColorado. Paleoflood uncertainties primarily are (Table 3). The estimatedrecurrence interval for the maximum relatedto possiblepostflood changes in channelgeometry and paleoflood(135 m 3 S-•) at ElkheadReservoir is slightlymore flood heightsinterpreted from PSIs. Where possible,paleo- than 10,000years (Figure 11) [Kircheret al., 1985], and the flood estimates are obtained in bedrock-controlled channels 10,000-yearflood is about110 m 3 s-• (Table3). The 10,000- that minimizechanges in channelgeometry; there is little ev- yearflood estimate is 104m 3 s-• at theElkhead gage (Table idencethat major changesin channelgeometry have occurred 3). The flood-frequencyrelation for the Elkhead Creek gage in alluvial channelsin the study area. Becausebeds of most JARRETT AND TOMLINSON: REGIONAL iNTERDISCIPLINARY PALEOFLOOD METHOD 2979 riversin the studyarea are relativelyarmored with cobbleand ducelarge floodsin the RockyMountains [FEMA, 1976;Han- bouldersand floodplainsediments typically are fine-grained sen et al., 1977, 1988] has implicationsfor dam safety and but stablefor long periods(Table 2), paleofloodreconstruc- floodplainmanagement. Although a number of streamflow- tions reflectrelatively stable conditions. The HWM-PSI rela- gagingstations in the Yampa River basinhad over75 yearsof tions developedfrom recent floods in the western United record,but no largerainfall floods,these long-term gaged data States[Jarrett et al., 1996],which included several documented were assumednot to be representativeof extremeflood po- 1995 floodsin northwesternColorado, help to reducethe un- tentialfrom rainfallby FEMA [1976].Thus the floodhydrology certaintyof paleodischargeestimates. In addition,when using for somestudies was basedon transposingdistant, large rain- paleofloodtechniques to estimatepeak dischargeof recent storms from Arizona, New Mexico, and southwesternColo- largefloods where gaged flood data were availableto assessthe rado into northwesternColorado and using rainfall-runoff reliability,the paleofloodestimates were within about 10% of modelingto adjustthe upperend of the gagedflood-frequency large gagedfloods and further documentthe value of the relation [FEMA, 1976].The flood-frequencyrelation for Elk- critical-depthmethod (Table 2). Therefore paleofloodesti- head Creek at Elkhead Reservoirdeveloped by Ayres Associ- matesfor this studyare believedto havetotal uncertaintiesof ates,Inc. (writtencommunication, 1996) (Figure 11 andTable about 25 to 30%. 3) essentiallyis the sameas the flood-frequencyrelations from The greatestsources of uncertaintyon flood variabilityare this studyup to about the 20-yearflood. The Ayres relation natural or anthropogenicclimate change(variability) effects. sharplyincreases above the 20-year flood, falls outsidethe Paleoflood estimates incorporate the effects of climatic confidencelimits of the regional flood-frequencyrelations changeson hydrologyduring the period of the paleoflood abovethe 50-yearflood, exceeds the maximumpaleoflood for record[Jarrett, 1991]. Certainly, moderate climate changes (or the basin at a recurrence interval of about 150 years, and other changessuch as wildfire effectson floodingor vegetation exceedsthe envelope curve value of 250m 3 s-•, whichis not changes)have occurred during the Holocene;however, these reasonablehydrologically. effectsare reflectedin the maximumflood preservedat a site. Similar to Elkhead Creek, the FEMA and gaged flood- Paleoflood data where the maximum age during which the frequencyrelations for the Yampa River at SteamboatSprings flood occurredis at least 5000 years are denotedwith large, (Table 3 and Figure12), wheredata collection began in 1904, solid triangles,and small, solid trianglesdenote a maximum have good agreementto about the 50-yearflood. For larger ageof lessthan 5000years (Figure 9). The envelopecurve of recurrenceintervals the FEMA relation increasessharply and maximumflooding incorporating the paleoflooddata (Figure does not fall within the 95% confidence limits for the flood- 9) is about 20 to 25% larger than contemporarymaximum frequencyrelation based on streamflowdata. In addition,the floodingin aboutthe past 100years since streamflow monitor- FEMA 500-yearflood is almostdouble the maximumpaleo- ing began(Figure 9). This modestincrease likely is due to the floodestimate of 311 m3 s-•. The PMF for StagecoachRes- large spatial extent of the database and relatively low- ervoirlocated on the Yampa River upstreamfrom Steamboat magnitudeflooding in northwesternColorado. Variability in Springs(Figure 12) alsofar exceedsa 10,000-yearrecurrence climate and basin conditionsduring the Holocene does not interval. Similar results for Walton Creek near Steamboat appearto havehad a largeimpact on floodmagnitude, and the Springsare listedin Table 3. assumptionof stationaritymay be valid for the upper end of The differencefor larger recurrenceintervals primarily re- the flood-frequencycurves in the studyarea. Thus the enve- sultsfrom transpositionof distantrainstorms over basinsin lope curve (Figure 9) probablyreflects an upper bound of northwesternColorado and then usingrainfall-runoff model- floodingduring the Holocenein northwesternColorado. ing to estimatethe upper end of flood-frequencyrelation as More quantification(e.g., using one-dimensionalor two- well as the PMF. The gageand paleoflooddata provideinfor- dimensionalhydraulic modeling to calculatepaleoflood dis- mation that canbe usedto refine assumptionsused to estimate charges,using absolute-age dating of flood deposits,more ro- extremeflooding using storm transposition and rainfall-runoff bust flood-frequency parameter estimation procedures, modelingto at least a recurrenceinterval of 5000 years.The regionalflood-frequency analysis with paleoflooddata, etc.) paleoflooddata provideno supportfor sharpupward slope wouldimprove the accuracyof individualpaleoflood estimates increaseof the frequencycurve. and better quantificationof uncertainties.However, the inter- To help place the flood and paleoflooddata in a regional pretationthat no substantialflooding has occurred during the probabilistic context, the EMA relations (with average/ Holocenein northwesternColorado, including Elkhead Creek, regionalskew) for the eight stations(Table 3) were plotted would not differ. While use of complexprocedures might pro- versusdrainage area (Figure 13). Althoughflooding results vide a moreprecise quantitative description of the data,dis- from severalfactors (basin slope, precipitation indices, vege- chargeand frequencyestimates of extremefloods in a basin tation, etc.) other than drainagearea, there is a fairly good may be readily estimatedby the paleofloodtechniques de- relation betweengaged sites. In addition,the envelopecurve scribedabove that providea cost-effectiveapproach. definedby the paleoflooddata alsocan be placedin a proba- A critical assumptionfor calculationof PMP estimatesis bility context. geographictransposition of stormevents from geographically The site-specificPMP study conductedfor the Elkhead and climatologicallysimilar locations to watershedof interest. Creek drainagebasin west of the ContinentalDivide in north- However, the NRC [1994] cautionsthat storm transposition western Colorado revisited various issues related to the PMP and moisturemaximization need to be for a slightlydifferent underthe explicitconditions which exist at ElkheadReservoir locationin the sameclimatic region. Regional analyses of rain- and other reservoirsin northern Colorado [Tomlinsonand fall, streamflow,and paleoflooddata in the presentstudy pro- Solak,1997]. These issues included a physicalaccounting of the vide informationto evaluatethe assumptionsabout large rain- effect of topographyon storm transpositioning,downslope storms in northwestern Colorado. wind flows under PMP storm conditions, and high-altitude The assumptionthat large rainstormsor rain on snowpro- moisturedepletion. The combinedresults of the hydrologic 2980 JARRETI' AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
1000 PMF, Stagecoach Reservoir Max Paleoflood
FEMA Envelope curve for Yampa River (fig. 9)
EMA (P=9,000 yrs; station skew) 100
EMA (P=9,000 yrs; ave. skew=-0.03)
I x PeakDischarge I ...... 95ø/øcønfidence limits
lO ...... i .... , i i , I , , i i i , i II i ! i , , i i , 1 10 100 1000 10000
Recurence Interval, years Figure 12. Flood-frequencyrelations for the Yampa River at SteamboatSprings gage (09239500) and paleoflooddata (shown as rectangle) using at-site and average skew with 95% confidencelimits. P denotesthe paleofloodlength of record.The envelopecurve value (Figure 9) for maximumflooding in the Holoceneis alsoshown. The flood-frequencyrelation for the YampaRiver at SteamboatSprings derived from rainfall- runoffmodeling [Federal Emergency Management Agency, 1976] and the probablemaximum flood for Stage- coachReservoir on the YampaRiver upstream from Steamboat Springs are shownfor comparison. modeling which routed the excessrainfall and snowmelt 7. Conclusions through Elkhead Reservoir and paleoflood study results showedthat the ElkheadDam wouldnot be overtoppedfrom A regional,interdisciplinary paleoflood approach provides a the site-specificPMP. Theseresults were acceptedby the Col- more thoroughassessment of floodingand with site-specific oradoState Engineer for damsafety certification with no mod- PMP/PMF studiesprovides dam safetyofficials with new in- ificationsto the existingstructure. formationto assessextreme flood potential.Interdisciplinary
lOOO x •lage...... ß ungagedsite ß paleoflood<5000yrs Envelopecurve for 4 paleoflood>5000 yrs NWColorado 100-yr • 500-yr • . • _ _
lOO
t. "-.2.(;' A ' A x • x x xX --x• •,,x xß • X X X
X X XX X ß X X A A' x x xX ß * Ax X • X ß . x ß . A A. ß X X •xxx x x x .... • ' ,, , ...... • ß ß , • , , ,,I I 10 100 1000 10000 Drainagearea, km 2 Figure 13. Relationbetween contemporary and paleofloodpeak dischargeand drainagearea with flood- frequencycurves for eightstations (Table 3) superimposedfor northwesternColorado. JARRETY AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2981
componentsinclude documentingmaximum palcofloodsand ing just the 162 high-riskdams in Colorado to the PMP stan- regional analysesof contemporaryextreme rainfall and flood dards[Hansen et al., 1988]to be approximately$184 million. data in a basinand in a broaderregional setting. Site-specific This modificationcost appears low as the estimatedmodifica- PMP studies were conducted to better understand extreme tion costfor proposedmodifications of the Cherry Creek dam rainfall processesby analyzingthe rainstormswith similar hy- are ashigh as $250 million for CherryCreek dam near Denver, droclimaticconditions [Tomlinson and Solak, 1997]. The ap- Colorado (U.S. Army Corps of Engineers,written communi- proachprovides scientific information to help determinethe cation,1997). There are over 10,000dams in the RockyMoun- delicate balance between cost of infrastructureand public tain region that may need to be modified for current PMP safety. criteria duringdam safetyrecertification. Thus, giventhe large The approachwas applied to ElkheadReservoir on Elkhead differencesin maximumpalcoflood and PMF values in the Creek(531 km 2) near Craig in theColorado Rocky Mountains. Rocky Mountains, it seemsprudent to conductadditional hy- On-site palcofloodinvestigations to determinemaximum pa- drometeorologicand palcofloodresearch to help reduce the lcofloodmagnitudes and regionalanalyses of extremerainfall uncertaintyin estimatesof maximumflood potential. This re- and flood data in northwesternColorado, primarily in the gional interdisciplinarypalcoflood approach,which is cost- YampaRiver and White River basins (10,900 km2), were con- effective,can be used in other hydrometeorologicsettings to ducted. Bouldery flood depositsand slack-watersediments, improveflood-frequency relations and provideinformation for which are preservedfor manythousands of years,were usedto a risk-basedapproach for hydrologicaspects of dam safety. estimateflow depthof palcofloods.A varietyof relativedating techniques(degree of soil development,surface-rock weather- Acknowledgments.We gratefullyacknowledge partial supportand ing, surfacemorphology, lichenometry, and boulder burial) assistancefrom Ray Tenney and Dave Merritt, ColoradoRiver Water were used to determinethe palcofloodrecord length for pa- ConservationDistrict, and along with Jim Pankonin, City of Craig, lcoflood depositsand noninundationsurfaces. Peak discharge Colorado,and Doug Laiho,Ayres Associates, help enhancingtechnol- for a palcoflooddeposit was obtained primarily using the crit- ogy transfer. Helpful review commentsfrom Charles Parrett, Pat ical-depthmethod, which had a dischargeuncertainty of about Glancy,Gerhard Kuhn, Lisa Ely, and Dean Ostenaaare appreciated. Conversationswith JohnEngland Jr. and Jery Stedingerwere partic- 25-30% for most study sites.Maximum palcofloodsprovide ularly helpful with flood-frequencyanalysis; they provided excellent physicalevidence of an upper bound on maximumpeak dis- technicalreview commentsas well. Jenny Curtis and Gary D'Urso chargefor any combinationof rainfall or snowmeltrunoff in provided excellentsuggestions with relative dating techniquesand northwesternColorado in at leastthe last 5000 to 10,000years. provided thorough reviewsof the flood chronologysection of the manuscript. Envelope curvesof maximum rainfall and flood data were developedfor contemporarydata and for the palcoflooddata. Maximum 24-hour rainfall for northwestern Colorado is about References 150 mm in aboutthe past100 years, which provides additional supportfor the lack of flood and palcofloodevidence. Maxi- Baker, V. R., Palcofloodhydrology and extraordinaryflood events,J. Hydrol.,6(1-4), 79-99, 1987. mum rainfall and flooding in northwesternColorado is sub- Baker, V. R., Geomorphologicalunderstanding of floods, Geomor- stantiallyless than in easternColorado, which is subjectto phology,10, 139-156, 1994. some of the most extreme rainfall flooding in the United Baker, V. R., R. C. Kochel,and P. C. Patton (Eds.),Flood Geomor- States. Large floods, if as hypothesizedby transpositionof phology,John Wiley, 503 pp., New York, 1988. large rainstormsinto northwesternColorado, would have left Barnes,H. H., Jr., and J. Davidian, Indirect methods,in Hydrometry-- Principlesand Practices, edited by R. W. Hershey,pp. 189-190, John palcofloodevidence in at least one of the streamsstudied. Wiley, New York, 1978. The envelopecurve of palcoflooddata is about20 to 25% of Benedict, J. B., Radiocarbon dates from a stone-bankedterrace in the the envelopecurve defined with contemporarydata alone.This Colorado Rocky Mountains,U.S.A., Geogr.Ann., 48, 24-31, 1966. suggeststhat effectsof climatechange and other factors(wild- Benedict,J. B., Recentglacial history of an alpinearea in the Colorado Front Range,I, Establishinga lichengrowth curve, J. Glaciol.,6(48), fire andvegetation changes) during the Holocenehave not had 817-832, 1967. a dramaticimpact on maximumflooding in northwesternCol- Benedict,J. B., Recentglacial history of alpine area in the Colorado orado. Flood-frequencyanalyses were made for eight gages Front Range, U.S.A., II, Dating the glacial deposits,J. Glaciol., with the expectedmoments algorithm, which makes better use 7(49), 77-87, 1968. of historicaland palcoflooddata. Frequencydata were super- Beschel,R. E., Lichens as a measureof the age of recent moraines, Arct.Alp. Res.,5(4), 303-309, 1973. imposedon the envelopecurves to help place the contempo- Bierman, P., and A. Gillespie,Range fires: A significantfactor in rary and palcoflooddata and associatedenvelope curves in a exposure-agedetermination and geomorphicsurface evolution, Ge- probabilitycontext. The maximumpalcoflood of 135m 3 s-1 ology,19, 641- 644, 1991. for Elkhead Creek at Elkhead Reservoir is about 13% of the Bilzi, A. F., and E. J. Ciolkosz,A field morphologyrating scalefor evaluatingpedological development, Soil Sci.,124(1), 45-48, 1977. site-specificPMF of 1020m 3 s-1. The estimated10,000-year Birkeland, P. W., Soilsand Geornorphology,372 pp., Oxford Univ. flood is about 170 m 3 s-• at Elkhead Reservoir. The lack of Press, New York, 1984. substantial rainstorms and flood evidence in northwestern Col- Birkeland,P. W., Soil-geomorphicresearch--A selectedreview, Geo~ orado probablyis explainedby high mountainbarriers, which morphology,3, 207-224, 1990. Birkeland,P. W., S. M. Colman,R. M. Burke,R. R. Shroba,and T. C. substantiallyreduce the availableatmospheric moisture from Meierding,Nomenclature of alpineglacial deposits, or, what'sin a the Pacific Ocean or Gulf of Mexico. The results of the site- name?,Geology, 7, 532-536, 1979. specificPMP/PMF studyand the regionalinterdisciplinary pa- Boner, F. C., and F. Stermitz, Floods of June 1964 in northwestern leoflood studyshowed that Elkhead Dam would not be over- Montana, U.S. Geol.Surv. Water Supply Pap., 1840-B,242 pp., 1967. topped from the site-specific PMP. These results were Brien, D. L., Palcofloodand streamflowdata to describethe spatial occurrenceof rainfall and snowmeltfloods in Wyoming,M.S. thesis, acceptedby the ColoradoState Engineerfor dam safetycer- 82 pp., Univ. of Wyo., Laramie, 1996. tificationwith no modificationsto the existingstructure. BuckIcy,J. L., Evaluationof designcriteria for hazardousdams, M.S. Changnonand McKee [1986]estimated the costfor modify- thesis,93 pp., Univ. of Wyo., Laramie, 1995. 2982 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
Bull, W. B., Stream-terracegenesis, implications for soil development, Gerrard, A. J., Soilsand Landforms,An Integration Geomorphology and Geomorphology,3, 351-367, 1990. Pedology,219 pp., Allen and Unwin, Winchester,Mass., 1981. Burke, R. M., and P. W. Birkeland, Reevaluationof multiparameter Gilley, J. E., W. J. Elliot, J. M. Laflen, and J. R. Simanton,Critical relative dating techniquesand their applicationto the generalse- shearstress and critical flow rates for initiation of rilling, J. Hydrol., quencealong the easternescarpment of the Sierra Nevada,Califor- 142, 251-271, 1993. nia, Quat. Res., 11, 21-51, 1979. Gottesfeld, A. S., British Columbia flood scars: Maximum flood- Capesius,J.P., A regionalanalysis of easternColorado streams and a indicators,Geomorphology, 14, 319-325, 1996. paleofioodanalysis of First Creek, Rocky Mountain Arsenal, Com- Gregory, K. J. (Ed.), Backgroundto Palaeohydrology--APerspective, merceCity, Colorado,M.S. thesis,178 pp., Dep. of Geogr.,Univ. of 486 pp., John Wiley, New York, 1983. Denver, Denver, Colo., 1996. Grimm, M. M., Paleofloodhistory and geomorphologyof Bear Creek Changnon,D., and T. B. McKee, Economic impacts and analysis basin,Colorado, M.S. thesis,126 pp., Colo. StateUniv., Fort Collins, methodsof extremeprecipitation estimates for easternColorado, 1993. CIRA Pap. 5, 75 pp., Coop. Inst. for Res. in the Atmos., Colo. State Grimm, M. M., E. E. Wohl, and R. D. Jarrett, Coarse-sedimentdis- Univ., Fort Collins, 1986. tribution as evidenceof an elevationlimit for flash flooding, Bear Cohn,T. A., W. L. Lane,and W. G. Baier,An algorithmfor computing Creek, Colorado, Geomorphology,14, 199-210, 1995. moments-basedflood quantile estimateswhen historicalinforma- Hadley, R. F., and G. C. Lusby,Runoff and hillslopeerosion resulting tion is available,Water Resour. Res., 33(9), 2089-2096, 1997. from a high-intensitythunderstorm near Mack, western Colorado, Collins, D. L., N.J. Doesken, and W. P. Stanton, Colorado State WaterResour. Res., 3(1), 139-143, 1967. summary--Floodsand droughts,in National WaterSummary 1988- Hansen,E. M., andF. K. Schwarz,Meteorology of importantstorms in 89, •S. Geol. Surv. i•ter SupplyPap., 2375, 207-214, 1991. the ColoradoRiver and Great Basindrainages, Hydrometeorol. Rep. Colman, S. M., and K. L. Pierce, Guidebook, 1983 Friends of the 50, 167 pp., Natl. Weather Serv.,Natl. Ocean. and Atmos. Admin., Pleistocenefield trip, glacial sequencenear McCall, Idaho, U.S. U.S. Dep. of Commer., Silver Springs,Md., 1981. Geol. Surv.Open File Rep., 83-724, 29 pp., 1983. Hansen, E. M., F. K. Schwarz, and J. T. Riedel, Probable maximum Colman, S. M., and K. L. Pierce, Summaryof Quaternary dating precipitationestimates, Colorado River and Great Basin drainages, methods,in QuaternaryNonglacial Geology, vol. K-2, Conterminous Hydrometeorol.Rep. 49, 161 pp., Natl. Weather Surv.,Natl. Ocean. U.S., edited by R. B. Morrison,pp. 45-74, Geol. Soc.of Am., Boul- and Atmos. Admin., U.S. Dep. of Commer., Silver Spring, Md., der, Colo., 1991. 1977. Costa,J. E., Paleohydraulicreconstruction of flash flood peaksfrom Hansen, E. M., D. D. Fenn, L. C. Schreiner, R. W. Stodt, and J. F. boulderdeposits in the ColoradoFront Range,Geol. Soc. Am. Bull., Miller, Probable maximumprecipitation estimates--United States 94, 986-1004, 1983. between the Continental Divide and the 103rd meridian, Hydrom- Costa,J. E., A comparisonof the largestrainfall-runoff floods in the eteorol.Rep. 55A, 242 pp., Natl. Weather Serv., Natl. Ocean. and United Stateswith thoseof the People'sRepublic of China and the Atmos. Admin., U.S. Dep. of Commer.,Silver Springs,Md., 1988. world,J. Hydrol.,96(1-4), 101-115, 1987a. Hansen,W. R., J. Chronic,and J. Matelock, Climatologyof the Front Costa, J. E., Hydraulicsand basin morphometryof the largestflash Range urban corridorand vicinity,Colorado, U.S. Geol. Surv.Prof. floods in the conterminousUnited States,J. Hydrol., 93, 313-338, Pap., 1019, 59 pp., 1978. 1987b. Harden, J. W., A quantitativeindex of soil developmentfrom field Costa, J. E., A historyof paleofloodhydrology in the United States, descriptions:Examples from a chronosequencein centralCalifornia, 1800-1970, in The Historyof Hydrology,edited by E. R. Landa and Geoderma, 28, 37-70, 1982. S. Ince, pp. 49-53, AGU, Washington,D.C., 1987c. Harden, J. W., Field methods,in Field and LaboratoryProcedures Used Crippen,J. R., and C. D. Bue, Maximum floodflowsin the contermi- in a Soil ChronosequenceStudy, edited by M. J. Singer, and P. nousUnited States,U.S. Geol.Surv. Water Supply Pap., 1887,52 pp., Janitzky,U.S. Geol. Surv.Prof. Pap., 1648, 3-8, 1986. 1977. Harden, J. W., Soil developmenton stablelandforms and implications Cudworth,A. G., Jr., Flood HydrologyManual, Water ResourceTech- for landscapestudies, Geomorphology, 3, 391-398, 1990. nology Publication,243 pp., U.S. Bur. of Reclam., Denver, Colo., Harrison, S.S., and J. R. Reid, A flood-frequencygraph basedon 1989. tree-scar data, N. D. Acad. Sci. Proc., 21, 23-33, 1967. Doesken,N.J., T. B. McKee, and B. D. Richter, Analysisof Colorado Henz, J. F., The Buffalo Creek flash flood of July 12, 1996, a recon- averageannual precipitation for the 1951-1980period, Fort Collins, structionof rainfall and meteorology,report, 28 pp., 1 append., Clim. Rep.84-4, 53 pp., 1 map, scale1:500,000, Dep. of Atmos.Sci., Henz Meteorol. Serv., Denver, Colo., 1998. Colo. State Univ., Fort Collins, 1984. Hosking,J. R. M., and J. R. Wallis, RegionalFrequency Analysis: An Eastwood,D., Statisticalanalysis of extremeprecipitation in Wyoming, Approach Based on L-Moments, 224 pp., Cambridge Univ. Press, M.S. thesis,119 pp., Univ. of Wyo., Laramie, 1995. New York, 1997. Elias, S. A., Late Pleistoceneand Holocene seasonaltemperature Hupp, C. R., Plant ecologicalaspects of flood geomorphologyand reconstructedfrom beetle assemblagesin the Rocky Mountains, paleofioodhistory, in Flood Geomorphology,edited by V. R. Baker, Quat. Res., 46, 311-318, 1996. R. C. Kochel,and P. C. Patton,pp. 335-356, JohnWiley, New York, Ely, L. L., Responseof extreme floods in the southwesternUnited 1988. Statesto climatevariations in the late Holocene,Geomorphology, 19, InteragencyAdvisory Committee on Water Data, Guidelinesfor de- 175-201, 1997. terminingflood-flow frequency, 2nd ed., report, U.S. Geol. Surv. England,J. F., Jr., Assessmentof historicaland paleohydrologicinfor- Off. of Water Data Coordination, Reston, Va., 1981. mationin flood frequencyanalysis, M.S. thesis,292 pp., Colo. State InteragencyAdvisory Committee on Water Data, Feasibilityof assign- Univ., Fort Collins, 1998. ing a probabilityto the probablemaximum flood, report, 79 pp., U.S. Enzel, Y., L. L. Ely, P. K. House, V. R. Baker, and R. H. Webb, Geol. Surv. Off. of Water Data Coordination, Reston, Va., 1986. Paleofloodevidence for a naturalupper bound to flood magnitudes Jarrett, R. D., Hydraulicsof high-gradientstreams, J. Hydraul.Eng., in the ColoradoRiver Basin,Water Resour. Res., 29(7), 2287-2297, 110(11), 1519-1539, 1984. 1993. Jarrett, R. D., Determinationof roughnesscoefficients for streamsin Federal EmergencyManagement Agency (FEMA), Flood insurance Colorado, U.S. Geol. Surv. WaterResour. Invest. Rep., 85-4004, 54 study,town of SteamboatSprings, Routt Co., Colo., report (rev. pp., 1985. 1983), 40 pp., Washington,D.C., 1976. Jarrett, R. D., Evaluationof the slope-areamethod of computingpeak FederalEmergency Management Agency (FEMA), Multihazardiden- discharge,in SelectedPapers in the HydrologicSciences 1986, edited tification and risk assessment: A cornerstone of the National Miti- by S. Subitzky,U.S. Geol. Surv.Water Supply Pap., 2310, 13-26, 1986. gation Strategy, subpart C hydrologic hazards, report, 369 pp., Jarrett, R. D., Flood hydrologyof foothill and mountain streamsin Washington,D.C., 1997. Colorado,Ph.D. dissertation,238 pp., Colo. State Univ., Fort Col- Foster, J., Improvingthe state of the art in risk analysis,Hydro Rev., lins, 1987. 38(6), 12-14, 1999. Jarrett, R. D., Hydrologicand hydraulicresearch in mountainrivers, Fredlund, G. G., and L. L. Tieszen, Phytolith and carbon isotope WaterResour. Bull., 26(3), 419-429, 1990a. evidencefor Late Quaternaryvegetation and climatechange in the Jarrett, R. D., Paleohydrologictechniques used to define the spatial southern Black Hills, South Dakota, Quat. Res., 47, 206-217, 1997. occurrenceof floods,Geomorphology, 3, 181-195, 1990b. JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD 2983
Jarrett, R. D., Paleohydrologyand its value in analyzingfloods and Matthai, H. F., Floods in June 1965 in the South Platte River Basin, droughts--Floodsand droughts,in National WaterSummary 1988- U.S. Geol. Surv.Water Supply Pap., 1850-B,64 pp., 1969. 89, U.S. Geol. Surv.Water Supply Pap., 2375, 105-116, 1991. McCain, J. F., L. R. Hoxit, C. F. Chappell,and F. Caracena,Meteo- Jarrett, R. D., Flood elevation limits in the Rocky Mountains, in rology and hydrologyin the Big Thompson River and Cache la EngineeringHydrology: Proceedings of the SymposiumSponsored by Poudre River basins,Larimer and Weld Counties, Colorado, Part A theHydraulics Division of theAmerican Society of CivilEngineers, San of stormand flood of July31-August 1, 1976, U.S. Geol. Surv.Prof. Francisco,Calif.,July 25-30, 1993,edited by C. Y. Kuo, pp. 180-185, Pap., 1115-A, 1-85, 1979. Am. Soc. of Civ. Eng., Reston, Va., 1993. McKee, T. B., and N.J. Doesken,Colorado extreme storm precipita- Jarrett, R. D., Historic-flood evaluation and research needs in moun- tion data study,Clim. Rep. 97-1, 109 pp., Dep. of Atmos. Sci., Colo. tainousareas, in HydraulicEngineering--Proceedings of the Sympo- State Univ., Fort Collins, 1997. sium Sponsoredby theAmerican Society of Civil Engineers,Buffalo, Meixner, R. E., and M. J. Singer,Use of a field morphologyrating New York,August 1-5, 1994, edited by G. V. Cotroneo and R. R. system to evaluate soil formation and discontinuities,Soil Sci., Rumer, pp. 875-879, Am. Soc. of Civ. Eng., Reston,Va., 1994. 131(2), 114-123, 1981. Jarrett, R. D., and T. W. Browning,Geomorphic estimates of rainfall, Menounos,B., and M. A. Reasoner,Evidence for cirqueglaciation in floods,and sedimentrunoff: Applied to the 1996 wildfire, Buffalo the Colorado Front Range during the Younger Dryas chronozone, Creek, Colorado,Geol. Soc. Am. Abstr.Programs, 31(7), 313, 1999. Quat. Geol., 48, 38-47, 1997. Jarrett,R. D., and J. E. Costa,Multidisciplinary approach to the flood Miller, J. F., R. H. Frederick,and R. J. Tracy,Precipitation Frequency hydrologyof foothill streamsin Colorado, in AWRA, International Atlas of the WesternUnited States, vol. III, Colorado,67 pp., U.S. Symposiumon Hydrometeorology,edited by A. I. Johnsonand R. A. Gov. Print. Off., Washington,D.C., 1973. Clark, pp. 565-568, Am. Water Resour.Assoc., Middleburg, Va., Mills, H. H., and J. B. Allison, Weatheringand soil developmenton 1983. fan surfacesas a function of height above modern drainageways, Jarrett, R. D., and J. E. Costa,Hydrology, geomorphology, and dam- Roan Mountain, North Carolina, Geomorphology,14, 1-17, 1995. break modelingof the Lawn Lake Dam and CascadeLake Dam NationalResearch Council (NRC), EstimatingProbabilities of Extreme failures, Larimer County, Colorado, U.S. Geol. Surv. Prof. Pap., Floods,Methods and RecommendedResearch, 141 pp., Natl. Acad., 1369, 78 pp., 1986. Washington,D.C., 1988. Jarrett,R. D., and J. E. Costa,Evaluation of the flood hydrologyin the National ResearchCouncil (NRC), EstimatingBounds on Extreme ColoradoFront Range usingprecipitation, streamflow, and paleo- PrecipitationEvents, 25 pp., Natl. Acad., Washington,D.C., 1994. flood data, U.S. Geol. Surv. WaterResour. Invest. Rep., 87-4117,37 National ResearchCouncil (NRC), ImprovingAmerican river flood pp., 1988. frequencyanalyses, Comm. on Am. River Flood Frequencies,Water Jarrett, R. D., and H. E. Malde, Paleodischargeof the late Pleistocene Sci. and Technol.Board, Natl. Acad., Washington,D.C., 120 pp., Bonnevilleflood, Snake River, Idaho, computedfrom new evidence, 1999. Geol. Soc.Am. Bull., 99, 127-134, 1987. Natural ResourcesConservation Service, Soil surveyof Routt County Jarrett, R. D., J.P. Capesius,D. Jarrett, and J. F. England,Jr., 1995: area, Colorado, report, U.S. Dep. of Agric., Lakewood, Colo., in Where the past (paleofloodhydrology) meets the present,under- press,2000. standingmaximum flooding, Geol. Soc. Am. Abstr.Programs, 28(7), Ostenaa,D. A., andD. R. Levish,The applicationof a non-inundation 110, 1996. approachof paleoflood hydrologyfor the assessmentof extreme Jensen,D. T., Probable maximumprecipitation estimates for short- flood hazardson the Ogden River, Utah in Environmentaland En- duration,small-area storms in Utah, final report, 75 pp., Utah Clim. gineeringGeology of the WasatchFront Region, edited by W. R. Lund, Cent., Utah State Univ., Logan, 1995. Utah Geol. Surv. Publ. 24, 495-508, 1995. Kircher, J. E., A. F. Choquette, and B. D. Richter, Estimation of natural streamflow characteristics in western Colorado, U.S. Geol. Parrett, C., Regionalanalysis of annualprecipitation maxima in Mon- tana, U.S. Geol. Surv. WaterResour. Invest. Rep., 97-4004, 51 pp., Surv. WaterResour. Invest. Rep., 85-4086, 28 pp., 1985. 1997. Knox, J. C., Large increasesin flood magnitudein responseto modest Parrett, C., Characteristicsof extreme storms in Montana and methods changesin climate,Nature, 361,430-432, 1993. Kochel, R. C., and V. R. Baker, Paleofloodhydrology, Science, 215, for constructingsynthetic storm hyetagraphs, U.S. Geol. Surv.Water 353-361, 1982. Resour.Invest. Rep., 98-4100, 55 pp., 1998. Komar,P. D., Selectivegrain entrainment and the empiricalevaluation Patton, P. C., and S. A. Schumm,Gully erosion,northwestern Colo- of flow competence,Sedimentology, 34, 1165-1176, 1987. rado, Geology,3, 88-90, 1975. Levish,D., D. Ostenaa,and D. O. O'Connell, A non-inundationap- Patton, P. C., and S. A. Schumm,Ephemeral-stream processes: Impli- proach to paleofloodhydrology for the event-basedassessment of cationsfor studiesof Quaternaryvalley fills, Quat. Res., 15, 24-43, 1981. extremeflood hazards,in 11thAnnual ConferenceProceedings, pp. 69-82, Assoc.of State Dam SafetyOff., Lexington,Ky., 1994. Phillips,J. V., andT. L. Ingersol,Verification of roughnesscoefficients Lewin, J., Floodplaingeomorphology, Prog. Phys. Geogr., 2, 408-437, for selectednatural and constructedchannels in Arizona, U.S. Geol. 1978. Surv.Prof. Pap., 1584, 77 pp., 1998. Lins, H. F., and J. R. Slack, Streamflow trends in the United States, Pitlick, J., Relation betweenpeak flows,precipitation, and physiogra- Geophys.Res. Lett., 26, 227-230, 1999. phy for five mountainousregions in the westernUnited States,J. Linsley, R. K., M. X. Kohler, and J. L. H. Paulhus,Hydrology for Hydrol., 158, 219-240, 1994. Engineers,3rd ed., 508 pp., McGraw-Hill, New York, 1982. Pruess, J. W., Paleoflood reconstructionswithin the Animas River Madole, R. F., Surficial geologicmap of the Craig 1/2 x 1 degree basinupstream from Durango,Colorado, M.S. thesis,192 pp., Colo. quadrangle,Moffat and Routt Counties,Colorado, U.S. Geol. Surv. State Univ., Fort Collins, 1996. Misc. Invest.Ser. Map, 1-1346, 1982. Pruess,J. W., E. E. Wohl, and R. D. Jarrett, Methodologyand impli- Madole, R. F., Lake Devlin and Pinedaleglacial history, Front Range, cationsof maximumpaleodischarge estimates for mountain chan- Colorado, Quat. Res., 25, 43-54, 1986. nels,Upper Animas River basin,Colorado, U.S.A., Arct. Alp. Res., Madole, R. F., Surficialgeologic map of the Meeker 30 x 60 minute 30(1), 40-50, 1998. quadrangle,Garfield, Moffat, Rio Blanco,and Routt Counties,Col- Schwarz,F. K., L. A. Hughes,E. M. Hansen,M. S. Petersen,and D. B. orado, U.S. Geol. Surv.Misc. Invest.Ser. Map, 1-1823,1989. Kelly, The Black Hills-Rapid City flood of June 9-10, 1972, A Madole, R. F., Quaternaryhistory of somesouthern and centralRocky descriptionof the storm and flood, U.S. Geol. Surv.Prof. Pap., 877, Mountain basins:Yampa River basin,in QuaternaryNonglacial Ge- 47 pp., 1975. ology,vol. K-2, ConterminousU.S., edited by R. B. Morrison, pp. Smith, F. A., and J. Betancourt,Response of bushy-tailedwoodrats 407-440, Geol. Soc. of Am., Boulder, Colo., 1991a. (Neotomacinerea) to late Quaternaryclimatic change in the Colo- Madole, R. F., Surficialgeologic map of the Walden 30 x 60 minute rado Plateau, Quat. Geol., 50, 1-11, 1998. quadrangle,Jackson, Larimer, and Routt Counties,Colorado, U.S. Snipes,R. J., et al., Floods of June 1965 in ArkansasRiver basin, Geol. Surv.Misc. Invest.Ser. Map, 1-1824,1991b. Colorado,Kansas, and New Mexico, U.S. Geol. Surv. WaterSupply Madole, R. F., Surficialgeologic map of the SteamboatSprings 30 x Pap., 1850-D, 97 pp., 1974. 60 minute quadrangle,Grand, Jackson,and Routt Counties,Colo- Soil ConservationService, Soil Surveyof Rio BlancoCounty Area Col- rado, U.S. Geol. Surv.Misc. Invest.Ser. Map, 1-1825, 1991c. orado, 219 pp., U.S. Dep. of Agric., Washington,D.C., 1982. 2984 JARRETT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD
Stedinger,J. R., and V. R. Baker, Surfacewater hydrology--Historical diatoms,and pollen at Moon Lake, North Dakota, Quat.Res., 48, and paleofloodinformation, Rev. Geophys.,25(2), 119-124, 1987. 359-369, 1997. Stedinger,J. R., and T. A. Cohn, Flood frequencyanalysis with his- Vierling, L. A., Palynologicalevidence for late- and postglacialenvi- toricaland paleofloodinformation, Water Resour. Res., 22(5), 785- ronmental change in central Colorado, Quat. Res., 49, 222-232, 793, 1986. 1998. Tomlinson,E. M., and M. E. Solak,Site-specific meteorological and Waythomas,C. F., and R. D. Jarrett,Flood geomorphologyof Arthurs hydrologicalstudy of the Williams Fork drainagebasin, N. Am. Rock Gulch,Colorado, paleoflood history, Geomorphology, 11, 15- WeatherConsult. Rep. AR 94-1, DenverWater Board,Denver, Colo., 40, 1994. 1994. White, S. E., Rock glacierstudies in the ColoradoFront Range, 1961 Tomlinson,E. M., and M. E. Solak,Site-specific probable maximum to 1968,Arct. Alp. Res.,3(1), 43-64, 1971. precipitation(PMP) studyof Elkheaddrainage basin, NAWC Rep. Wilcock,P. R., Flow competence,a criticismof a classicconcept, Earth WM 95-6, TRC N. Am. Weather Consult., Salt Lake City, Utah, Surf.Processes Landforms, 17, 289-298, 1992. 1997. Womack, W. R., and S. A. Schumm, Terraces of Douglas Creek, Trieste,D. J., and R. D. Jarrett,Roughness coefficients of largefloods, northwesternColorado, an exampleof episodicerosion, Geology, 5, in Irrigationand DrainageDivision SpecialtyConference "Irrigation 72-76, 1977. Systemsfor the21st Century", Portland, Oregon, 1987, edited by L. D. Jamesand M. J. English,pp. 32-40, Am. Soc.of Civ. Eng., Reston, R. D. Jarrett, U.S. GeologicalSurvey, P.O. Box 25046, MS 412, Va., 1987. Denver,CO 80225.([email protected]) Tweto, O., Geologic map of the Craig 1 x 2 degree quadrangle, E. M. Tomlinson,Applied Weather Associates,P.O. Box 680, Mon- northwestern Colorado, 1:250,000 scale, b:& Geol. Sutv. Misc. In- ument, CO 80132. vest.Ser. Map, 1-972, 1976. Valero-Garces, B. L., K. R. Laird, S.C. Fritz, K. Kelts, E. Ito, and E. C. Grimm, Holocene climate in the northern Great Plains inferred (ReceivedSeptember 9, 1997;revised April 3, 2000; from sedimentstratigraphy, stable isotopes, carbonate geochemistry, acceptedApril 7, 2000.)