|
<br />
<br />~'
<br />-
<br />.~
<br />f,....;:,.
<br />'.~
<br />~--
<br />,-r~.-'
<br />
<br />~-
<br />
<br />:~
<br />
<br />M,..-.
<br />
<br />'":-r
<br />-,,,,;.
<br />,;~.;.
<br />
<br />;,:,~~.
<br />
<br />~t:
<br />,";;;;;
<br />
<br />ft-
<br />-,.~i.
<br />~-'
<br />~.-
<br />
<br />~''f-
<br />
<br />.....,..
<br />?~
<br />
<br />"~"~
<br />
<br />,.
<br />"'"
<br />-::!to.
<br />
<br />.~.'
<br />ii;;:
<br />
<br />COMPARISON OF NINE UNCALIBRATED RUNOFF MODELS TO
<br />OBSERVED FLOWS IN TWO SMALL URBAN WATERSHEDS
<br />
<br />By Phillip J. ZarrleUo, Hydrologist, U.S. Geological Survey, Ithaca, New York
<br />
<br />~hstract: Ninc uncalibratcd runaffmadel results-CASC2D, CUHP, CUHP/SWMM, DR3M, HEC.I, HSPF,
<br />PSRM, SWMM, and TR20 were campared to' abserved flaws in two' small urban watersheds with distinctly different
<br />climatic and physiagraphic settings; a 3.10 mi2 semiarid maderately-slaped watershed near Denver, CO, and a 0.14 '
<br />mi2 caastal watershed with steep slapes near Seattle, WA. All madels were run by experienced madelers using identi-
<br />cal data pravided an selected basin characteristics and rainfall that are readily available to' the engineering cammu-
<br />nity. Observed streamflaw was nat pravided, and thus the results are dependent an madelers' judgment and the ability
<br />to' canceptually and quantitatively represent the hydralagic system, Data far six starms in each watershed (antecedent
<br />canditians, rainfall valume, intensity, and duratian) were provided. Simulated peak flows differed from abserved
<br />peak flaws by as much as 260 percent and simulated stann valumes differed fram abserved starm valwhes by as
<br />much as 240 percent. The average raat mean square madel errar (RMS) was slightly larger far peak flaws in the
<br />caastal watershed (68) than in the semi-arid watershed (55), whereas the RMS far stann valume was abaut the same
<br />in the two' watersheds (54 and 56, respectively). In general, the madelsbased-on the SCS.curve numberbad the paar-
<br />est fit. Results indicate simulated flaws fram uncalibrated models have wide variability far bath types afwaiersheds;
<br />this cauld result in aver- ar underdesign af stannwater-management structures.
<br />
<br />INTRODUCTION
<br />Planning and design af stonnwater-drainage systems, culverts, detentian basins, and ather stannwater facilities
<br />requires infannatian an stann peak flaws and runaffvalumes. Rainfall-runaffmadels, aften utilized to' estimate this
<br />infannatian, vary in camplexity, functianality, and applicability to a given region ar stann type, but anly'through a
<br />process af calibratian and verificatian can reliable results be acquired. Uncalibrated madels are aften used. hawever,
<br />because either the infannatian needed far calibratian and verificatian are unavailable, ar the expense afthese prace-
<br />dures can nat be justified. Little infannatian is availahle an the reliability and error assaciated with the use af an
<br />uncalibrated madel.
<br />
<br />TO' detennine the variability afuncalibrated flaw simulatians across a range afmadel types, a si1U1'1o camparisan af
<br />nine selected, nanproprietary, runaff madels were made far six stanos in each af two small urban watersheds with
<br />distinctly different climatic and physiagraphic settings; Harvard Gulch, a semi-arid watershed near Denver, CO, and
<br />Surrey Dawns, a caastal watershed in the Pacific Narthwest near Seattle, WA. Each afthe madels were run by expe-
<br />rienced madelers using anly data pravided an selected basin characteristics, antecedent stann canditians, and stann
<br />rainfall that are typically available to' the engineering cammunity. Observed streamflaw were nat provided and the
<br />madels were run withaut being calibrated. Thus, !)ladel results are dependent an madelers' judgment and the ability
<br />to canceptually and quantitatively represent the hydralagic system. This paper describes the results af the uncali-
<br />brated made Is.
<br />
<br />Acknowledgments: Numeraus peaple pravided their time and expertise to' perf ann madel simulatians- Dr. Gret
<br />Aran, Penn State Univ. (pSRM); Darryl Davis, Arlen Feldman, Gary Brunner, Dr. David Galdman, and Stephen Brei-
<br />thaupt, U.S. Anny Carps afEngineers (HEC-I); Richard Dinicala, U.S. Geolagical Survey (HSPF); Dr. Pierre Julien,
<br />Scott Hagan, Aaron Egben, James White, and James Light, Calarado State Univ. (CASC2D); Tania McNutt, Cala-
<br />rada State Univ. (CUHP); Michael Schmidt, Camp, Dresser and McKee (SWMM). The ASCE Urban Water
<br />Resaurces Research Cauncil, particularly Ben Urhanas, provided the initial idea and suppart far this study,
<br />
<br />RAINFALL-RUNOFF MODEL CHARACTERISTICS
<br />
<br />Runoff models differ mainly in the methads used to generate runoff and to' route it through a basin; they also differ in
<br />the control options available, data handling, and user interface, but these differences generally have little ar nO' effect
<br />on haw the model computes runaff. The test madels (summarized in table I) calculate runaff(excess precipitatian)
<br />by ane afthe fallawing; (I) SCS curve numher, (2) Hartan's equatian, ar (3) cantinuaus sail maisture accaunting.
<br />The SCS curve numher is the mast widely used method because of its relative simplicity; it defines the watershed
<br />starage and is detennined for a watershed or sub.watershed predaminantly from the types af soils, vegetative cover,
<br />and land-use characteristics (Soil Conservation Service, 1986), Harton's equation assumes that the soil infiltration
<br />rate decreases expanentially as a functian aftime since the stonn began. Same madels accaunt far sail-maisture star-
<br />
<br />P.ECr.:!VED
<br />
<br />7-163 MAR 2:J 1999
<br />
|