You are on page 1of 5
UNCERTAINTY IN FLOOD EsTIMATES ASSOCIATED WITH ROUGHNESS COEFFICIENT By Ellen E. Wohl ‘AgsrRAct: Although paleoflod reconstructions ae frequently used to provide discharge estimates of extreme floods, a concem remains regarding uncertainty in Manning's n-values. An analysis ofthis uncertainty relative to a commonly used step-backwater model was performed for channel reaches in five canyon rivers. Values of rn were selected for each study reach using fout methods. Each set of n-values was then varied by 10% and 259, The percentage change in discharge associated with varying n-values i inversely proportional to channel ‘gradient and to widtlvdepth ratio, and proportional to roughness. For channel with a gradient less than or equal to approximately 0.01, varying m by *25% produces a maximum change of 20% in discharge. The greatest discrepancy among n-values estimated using the four methods occurred for channels with relatively low gradient and low roughness, but associated discharge varied fairly consistently between 13 and 27%. These results indicate that uncertainties in discharge estimation resulting from the roughness coefficient in step-backwater modeling ‘of paleofloods are comparable to or lower than uncertainties associated with other methods of indiecty est- ‘mating flood discharges. INTRODUCTION Paleolood reconstructions of ungauged flood discharges ae proving increasingly useful in developing food magnitude-e- ‘quency relations for basins throughout the United States and the world (Baker 1987, 1988; Ely et al. 1993; Enzel etal 1993, 1994; Salas et al. 1994; Wohl et al. 1994c). One rmensional,step-backwater hydraulic models are widely used, in conjunction with paleostage indicators, to estimate water surface profiles and discharges for these ungauged floods (O'Connor and Webb 1988). Paleostage indicators include de- bis accumulations, silt and scour lines, slackwater deposits, for any indicators of low stage (Baker 1989; Salas etal. 1994) that may be used to indirectly estimate water surface profiles After floods. In channels where the cross-sectional geometry Tikely to remain relaively table during and following a food, sep-backwater programs can be used to model a specified dis charge along the surveyed channel geometry to compute Wa- ter-surface profiles. These computed profiles are then com- pared to the profiles defined by paleostage indicators until & Meats” discharge is obtained. The HEC-2 model (HEC2 1990) is most commonly used for this procedure (Partridge and Baker 1987, Pickup 1989; Enzel etal. 1994; Wobl eta. 1994a,b), bu questions have been raised ast the uncertainties ‘of using a step-backwater approach under conditions for which the roughness coefficient cannot be directly measured. This paper evaluates uncerainties in estimating paleotlod dis- Charges as a result of uncertain mvalues for five field data ses ‘The HEC-2 model is used because this isthe method most Tikey to be employed in paleoflood studies, and because the ‘wate-surface profiles generated with various mvalues are Compared direlly to paleostage indicators to choose a best discharge foreach roughness coefiient estimation scenario, “The HIEC-2 model assumes fw conditions that are seady with time and gradually varied in space. To estimate water Surface profiles associated with gradually varied flows, the Slope ofthe energy grade line a a section is assumed to equal the energy slope for a uniform flow withthe velocity and hydraulic radius of the section (Chow 1959). This allows the Use of uniformflow formulas, such 85 the one-dimensional "Dept of Barth Resour, Colorado State Univ, Collis, CO 60523 Note. Discussion open unl July 1, 1998, To extend the closing date fone month, a ween request must e filed withthe ASCE Manager of “Journals. The manvscrp for this technical nove was submited for review ‘and possible publication on May 1, 1995. Ths tecnica note is pat of ‘he Journal of Hydraulic Engineering, Vo. 124, No.2, Febrtry, 1998. ‘GASCE, ISSN 0739-9499/90000 0819-022318400 4 $30 pet page. ‘Teemicat Note No, 10614, energy equation, for evaluating the energy slope at each cross fection in nonuniform channels. Rate of energy loss between ross sections is caused by frictional losses associated with boundary roughness, eddy losses from turbulence, and flow separation caused by channel constrictions and expansions (O'Connor and Webb 1988). Frictional losses are accounted for in the selection of Manning's n-values, while flow sepa- zation and eddy losses are accounted for in the selection of expansion and contraction coefficients. Inthe HEC-2 program, separate n-values may be specified for the right overbank, ‘main channel, and left overbank portions of each reach, or the cross section may be further subdivided. ‘When applying HEC-2 to paleohydrologic studies in canyon rivers, errors in discharge estimation may result from changes in channel configuration during the flood, and between the ‘occurrence of the flood and the date ofthe study; representa- tiveness of the paleostage indicators of actual flood stage; in- accuracy in defining the high-water matks; post-lood altera- ‘ion of paleostage indicators (eg. partial erosion of slackwater deposits); and inaccurate energy oss and roughness cocfi- cients. This paper addresses the latter source of error by eval- uating model sensitivity to n-values for various channel con- figurations. "The selection of n-values for flood flows, particularly in hhigh-gradient channels, may be subject to error (Dawdy and Mouyed 1979; Sauer et al. 1985; Jarrett 1987; Jarrett and Malde 1987; O'Connor and Web 1988; Burmham and Davis 1990). Although Dawdy and Motayed (1979) and O'Connor ‘and Webb (1988) concluded that the HEC-2 program was rel- atively insensitive to variations in Manning's n for deep, nar- row channels, there have been no sensitivity analyses for n covering the fange of canyon channel morphologies to which the HEC-2 model is commonly applied in paleohydrologic studies. This paper addresses the issue of sensitivity to m by using five field examples covering a range of canyon-river ‘morphologies. The results of these analyses in turn have im- plications for the accuracy of paleostage-based discharge es- limates relative to gauged discharges. (CHANNEL CHARACTERISTICS ‘Table 1 summarizes the relevant characteristics for the five channels considered in this analysis, and Fig. 1 shows repre- sentative channel cross sections plotted at the same scale. Figs. 2-6 illusrate the actual appearance of the channels. Together, these channels represent a Wide range of maximum discharge, ‘gradient, widih/depth ratio, channel substrate, climate, and ri- JOURNAL OF HYDRAULIC ENGINEERING / FEBRUARY 1998/ 219 TABLE 1. Channel Charactoriatices Peak Discharge | Predom- Drainage | Reach ‘ant wey ‘rea’ | tongth | Cross | Channel | Riparian | Estimated*| Gauges | channel | dap» | Climatic cannot | err) | (tm). | sections* | gradient | vegetation] (ms) | (mes?) | substrat rato" regime a) @ |@ | @ eo © o ® @ (10) fa Poadre River 400 | sie | 26 | 0003 | Moderate] S000 | 60) | Cobbies [1097181 = 61 | Semi ‘Colorado Fiteroy River, | 25.000 | 2620 | 6 | 0.001 | Moderste| 30,000 | 30,000 | Sona | 260.4723.4 = 1200| Seasonal ‘Ausualia. tropics pana iver, | 2950 | 5600 | 44 | 0.005 | arse | 2100 | 1.080 | cosbies, Josums=75 [Hyped israel boulders Anniver, | 27490 | soo} 7 | oor | tae | 4000 | 1,500 | soutdees |2ssn70= 14 |subwopical ‘Nepal ctaying Grek] 289 | 340 | os | oat | Thee 150 | — | powders |166%2=20 | sobxopica Nepal "Number of cross sections in sud reach, “Largest dacharge estimated from paleostage indcatos ‘Langest ischrge ever gauged ator near sud reach ‘For argent modeled discharge, sn er ag ont Me uw co = | «| «| y orovnee / aaa FIG. 3. Fitzroy River, Australia, Looking scross Vegetated High-Fiow Porton of Valley Botiom toward Low-Flow Channel FIG. 1. Sample Cross Sections of Channola; Stage ls Shown and Bedrock Valley Wall {or Largest Flood Modeled, along with Type of Palecetage indl- ator Used (SWD Is Slackwater Deposit, Scour le Scour Line) parian vegetation. Each channel has been described in greater etal elsewhere (Wohl et al. 1994a,b; Wohl 1995). SENSITIVITY ANALYSES FOR MANNING'S 1 Manning n-values for each study reach were estimated using four methods: 220] JOURNAL OF HYDRAULIC ENGINEERING / FEBRUARY 1908 1, Jarret’s (1984) equation for high-gradient channels n= 032sRo w where $= channel gradient; and R = hydraulic radius in meters. This equation was developed for natural main channels having stable bed and bank materials (boulders) without bedrock. Tt is intended for channel gradients from 0.002-0.04 and for hydraulic radii from 0.15~2.1 1m, although Jarrett (1984) noted that extrapolation to larger flows should not be too much in error as long as the channel substrate remains fairly stable. 2, Limerinos’s (1970) equation for natural alluvial channels FIG. 4. Paran River, lerael, Looking Upstream FIG. 6, Chayang Creek, Nepal, Looking Upstream; Scour Lines Appear slong Right Side of Photograph at Base of Vege tation 0.0926n°" 116 + 2 logan ° where R = hydraulic radius and dy = size of the inter= mediate particle diameter that equals or exceeds that of 84% of the steambed particles, with both variables in feet. Thig equation was developed for discharges from 6-430 m's-f, and 2/R®"” ratios upto 300 (although Lim- einos found that litle change occurs over values of 30). 3. Visual estimation of n-values was performed at each site using Bares (1967) a8 a guideline. acl 4. The Cowan (1956) method for estimation of n, as mod- FIG, 5._ Arun River, Nepal, Looking Upetream; Scour Lines Ap- ified by Arcement and Schneider (1988), is designed spe- pear at Base of Vegetation on Right Slae of Figure cifically to account for floodplain resistance: TABLE 2._ Results of Sensitivity Analyses: Varlous n-Values and Equivalent Discharges Pouare ‘ann Chayang zo Paran wenos[—n | @ | =» | a] | a | ant | om (im om ea [ca o}e@etlolw}olol|loa @ | oo | on | oa | os | a9 | a5 Tare | 024 | 3.000 | o0si | 4.500 | 0.096 | 1,300 31.000 | 0015 | 9.000 | a0s4 | 2.400 | 0.039 | 200 “rive | 002s | soo | 0.056 | 4300 | os | 100 430,000 | ois | S000 | o.037 0083 | 200 “10% | 22 | 000 | core | 4300 | 0.086 | 100 32000 | ois | 9.000 | O03: 0035 | 200 425% | 030 | $500 | 006s |

You might also like