'l he Narth aaralilu Celrrdrhe^ lla/xrle I i, 2A06, fr. l. t 9 1 Mapping Flood Extent Using a Simple DEM-Inundation Model Tao Zhengt and Yong Vhngl Central trfi chigan Univcrsih,l l:,asr Catolina Universi6l A gtid bascd one-dimensional digitai eler. ation modcl (DE\I) -inundation model has been dcvclopcd as a tool for flood extent mapping on floodplains. l hc validitv anrl ac_ curacv of the rrodel hale been assessed through compadson of modclcd rcsulrs rvrrh those dcrir.ed ftorr the widely used standard and complcx 1 D Hvdrologic Enginecring Centct Rir.cr,\nalvsis Svstem (HEC RnS) modei and vcrification against the Septcm_ bcr 1999 tlood on the lo*er'Iar Rirei floodplain, Notth Catoiina. 1hc two models are comparable in accuracv. lfith its simple implementation and ease of parametcrization, the DEll-inundatioo model is a potential altetnatir-e to the FIEC-RIS model. Inffoduction Floods ate onc ofthe most significant natutal hazards, costing lir.es, serious d^magc to ptopert\,, and di._uprions ro..c al and rconomic ccrrvrties. l'he abilitv to map the flood cxtent accurately and timelv can pror.ide ctitical infotma- tion fot immediate flood relief acrir.ities, and and post flood mitigation effotrs (\{i1eti 1999, Colby et al. 2000, Yang and Tsai 2000, AI-Sabhan et a1.2003). To this end, hydraulic models har-e been developcd and used for map pre ping flood extenr (Hydtaulic Fingineeing Center 1997, Cotreia et al. 1998,,\ckerman er xl. 2000, Chang ct aI.2000, l)obson and Li 2000, -\l Sabhan ct al.2003, Hunrer ct aI.2005, Bates et al.2006). Over the vears, both trvo and one dimensional hvdraulic models hale been dcr-el oped. The 2 D models include rhose that emplov sophisticated full finite,elcmcnt approaches or that t^he grid-based apptoachcs. For instance, Galland et al. (1991) der.elopcd a 2 D finitc element numerical model, thc TELEXIT\C-2D. Nicholas and Nlitchell (2003) also der-cloped a finitc-element 2-D model that solves the depthavcraged shallou. *'ater foim of rhe \aviet Stokcs equations. The 2-D modcls are genetallv capable of achieving high mapping accuracl-, cspcciallv fot hldraulic processes at fioe spatial tcsoluiion, b,-:r the1. tequire digical elevarion modeis (DEi\{s) of high rcsolution and accu tacl', as v'eil as othcr geophvsical model inputs. Thev all atc computatioflailv intensir'e. To ar.oid the dtawbacks of the finite element models, Batcs and Dc Roo (2000) der-elopcd a tastet bascd model, thc I-ISFLOOD,FP, rvhich takes a stora€le cell approach to simulate flood hydtologic aod hl'dtaulic process. Thc LISFI-OOD FP has been subsequentlt improred and 1'alidated for the Januatv 199 5 flooding on the Rir-et N{cuse, the Nethetlands (Hunter et al. 2005, Bates et al. 2006). Unlike 2 D modeis, 1-D hydraulic models are r,vpicallv chatacterized bv a serics of ciosssections of channcl and floodplain topogtapht. \ralidation resrs ha\.e tepotred that 1-D mod els, such as the Hl.dtaulic Engineeting Center Rir-et r\nalvsis System (HEC R.\S), atc capable uf -caching hrgh nccurdc\ rn foocl cxLcr-,napping (Flortitt and Bates 2002). Inlcstigations have bccn also conducted on ho!"'the accuracv of the model can be affected b). \'arious factots, :ucl_ as mr.h -esolutior. topoglnpnic rcfresen tation, and spatial resolr.rtion (Horritt and IJates Horritr et al.2006). In shorr, rhc existing 1 D an<l 2,D modcls can map a flood extent accurafell., buf thcr- are Jif[icu]r 'o bu parancrerrzeJ. \m,,nq L,illcr-. 2001, TL? the cstimatiol of Manning's coefficient of ftiction as input to the modcls, which is also rcferled to as N{anning's z (Chov 1959), is highly un- certain and unreliable. For instance, laboratorv cxperiments har.e reported higher r.alues fot \Ianning's r thao those recommended in the rvell-established tables by \l T. Chorv in 1959 (\\rilson and Hottitt 2002). ..\lthough different \.alucs halc been recommended (Acrement and Schneider 1989) and cxtersive studies have been conducted to derive the coefficients (Werncr et al. 2005, Wilson et al. 2006), thcte is still no Pfo\ren wav to estimate the r \\'ith a high levcl of confidence and accuracy. Additionalll', the implementation of the existiflg models tequitcs advanced ler.els of hydtologic and hydtaulic knowledge and expertise, rvhich is often lacking among prospective users, therefore hindcring the use of the models. Thus, thete are clear needs for a hvdrar.rLic model that is simplc in DENI-inundatioo modeL accounts for these factors in the."vater sutface height interpolation. In summary, the objectives of rhis paper are to detail the deleiopment of a DEN{ inuoda tion model, to compare the model wirh rhe FILC RAS model to assess thcir accutac,v in flood extent mapping, and to validate rhe DEM in undation model against a real flood event. Methodogy: HEC-RAS Model To meet thc needs for flood extent map ping, thc Hvdraulic Engineeting Centcr (HEC) of the US Armv Cotps of Engineers dcveloped a scrics of GlS-based hydraulic models, from the Atc/FlEC2 to HEC RAS (Hi'dtologic En gineeting Center 1997, I{taus 2000, Acketman ct al. 2000, USACE 2007). HEC RAS is one of the most popular 1 D hvdraulic models. Compalcd rvith its ptedecessors, HEC-RAS Xar. C,7.:ar G locaooos s:e- d and rou?:-::e:; o bett-eec:ie--rp Iength re:e:: :o d cross sec--:o::- r o\'eibark- ::.:iE ( quired- Tc' e:;dc uses eoe:z; lo$ \{anoioe'. z rrlrr b) cotltrac:os ri eraluate i:.2::sioo vert loss coe:-dci to s'eir s55e- F and enrralca rnd lic inputs =clu& informaior- rod clude kaor:: srr depth. no:=rJ & Methodologr: can providc ioitial aod pteliminarl' analysis, and comes vith some majot improvements. It facilitates the usc of digital datasets such as DEN.{ and TIN (ttiangular irre€tular netwoik) (Correia et al. 1998, Dobson and Li 2000, Yang and Tsai 2000), and featurcs an enhanced gtaphical uscr the result cao help the complex model fot in depth study. To meet the necds for simpie flood-extent mapping models, \Vang et al. (2002) dcveloped a model that maps flood extent by lincarl,v inrcrpo alng rhe :urFace ua cr hcighr of a rirer between cwo reighborirg g..rugirg clarion5 Lsiog the heights measuted ar the srations. In this intetface that simplifies the flood extent modeling ptocesses. The HEC-RAS model is designed to perfotm 1-D hydraulic caLculation for a fulL net work of natural or constructed \,'atet channelsIn the model, surface profiies of a stead,v florv in rvhich changcs in flos' depth and r.elocity occur graduallv ovet a considerable length of of article, an imptoved r-etsion of Wang et al. (2002)! modeL is developed. The nervly developed model is a 1 D DENI inundation model that features thtcc majot improvemeots. Fitst, Wang et al. (2002)'s rnodcl did not identifv the central channel of the tivet; this model does. Second, Wang et channel arc solved bv using a 1-D energ'' equation parametetization and implementation. Such a simple model, if capable of teaching compa- of the complex modei, can serve as an altetnative. In addition, a simple model tabLe accutacy water surface height at different florv condi- and cnergt'head loss equation (Hvdraulic Engineering Center 1997). The steady fLow's rvatet surface profiles are computed from dorvn stream to upstream at cross sections for a given dischalge fate at upstream and \r.atet sutface height value at downstream. In the soh.ing of the vater sutfacc profile along a dvet channeL, HEC R-\S requites geomettic and hydraulic input parameters. The geometric parametets include the river system schcmatics, cross section profile, reach length, energv loss coefficient, and stteam junction information. The schematic patametets dcfinc how river reaches are con- tions, weie not modeled (\Vang et al.2002). This ncctcd. Ctoss sectioo profiles ate tequited at al. (2002) represented distance betrveen gauging stations with a sttaight line, whereas this model traces the distance aLong the central channel line betrveen tlvo ncighboriog gauging stations. l,astlv. the changes in elevation of a tivcr channel and banks along a tivet, s'hich are important geometdc factois affectrng a tir'-et's DEM-inundd Compr-ec the sd l-D DE\I-ran 6cial g-ater ieizhr s a stream gi cc the DE\f to ie:erm non flooded a:er; suf€menrs are aFril the disrance bersq c.st fl! attificial \zre: @ tions mar be tlveen s laEioas in four majo: :;eF ceotetline, deireri: the centerline- esi centellrne's s rt:aa! r cented.ine. arci iod herght grid ior difr eate the cenieiic ( a) Or-eriar co-I motelv ser.ed the strean i€c trr'e cenrerii:re : positionei ap the leli ard :!! b) Ideodi- &e i The Narth Carolina Geoparher Iocations rvhete changes in discharge, slope, shape, and roughness occur along the rivei channel between the up- and dowo stleam. The reach length tefers to the measured distance betrveefl cross sections. Thc reach icngths fot the left overbank, right overbank, and chanoel are requited. To er-aluate energy losses, HEC-RAS uses enetgy loss coefficients inciuding a) N{anning's z value fot ftiction loss (Chow 1959), b) conttaction and expansion coefficients to evaluate tiansitioo loss, and c) btidge and culvert loss coefficients to evaluate losses related to weii shape, pier configuration, pressure flow, arrd enttance and exit conditions. The hvdtaulic trptr's include flo,.r regine. pcaL discharge infoimatiorr, and boundaty conditions that include known watet surface eler.ation, ctitical depth, normal depth, and ratiflg curve. J of the tentative centetline, and use it as the center for seatching the pixel witl.r lowest eleva tion r-alue within a certain tadius. Out expedend ment indicated that a radius of 300 m is suffi cient for most cases, rvhich is equivalent of 10 plxels on 30 x 30 m USGS DENL This pixel *ith the lorvest elevation is then the actual location of the delineated centedine. N{ove one pirel do$'ristream aloflg the tentative centetline, and perfotm the similar searching until the dorvnstream cnd of the tentative centedine is rcached. Thus a lowest-elevation plxel is identiEed fot each cottesponding pixei od the ceoterline. c) N{anually draw a new centerlifle by tracing through all the pixels with tle Lowest-elevation values from the upstream to downsffeam ends. d) Veri!' the centedinc created in tle step 3 with dre DENI, aerial Methodology: One Dimensional DEM-inundation Model Compared with the complex HEC-R-\S model, the 1-D DE}I-inundation model calculates an arti 6cial water heighr surface using sutface water height of a stream and compates the artiEcial sutface with the DEM to determine watet/noo-watet ot flooded/ non flooded ateas. The sutface \vater height measutements ate available at gaugng statioos. Because the distancc between two neighboring gauging sta tions may be quite latgc, sutface water heights between stations must be interpolated to create tl1e attificial watet height sutface. This is accomplished in fout major steps, tie delioeation of the stream centetline, derivation of surface water herght aloog of the reach of the ceotetlinet surface watet herght forlocations off dre centedne, and finaliy creation of the surface vater height grid for different flow conditions. To delineate the centefline of a dvet section, we a) Or,.erlay coJocated aetial photogtaphs ot temotely sensed images or.et the DEN{ coveting the sfteam section in question. Then, a tentathe centedine, estimation trve centedine is drawn in such a wav that it is positioned approximately equidistant between tle left and right banks. b) Identif the first DEM pixel on the upstream photognphs ot satellite irnages. If needcd, iepear steps a), b), and c) until a sat isfactory tesult is achieved. A satisfactoty centetline should be continuous with each pirel positiooed at the lowestpointof its cottespond1ng cross secuofi. T1picall1., dre delineated centedine is a curr.ed line compo.ed oI rhe deepesr pirels along a rir cr sucarr. Second, the water suface heght at each Location or pirel along the centetline is calculated. An assumption used for this calculation is that vrater sudace height decreases from upstream to dorvnstream and that the decrease depends on the changes of location and elevation long *re centedile. Figure 1 illusttates the calculation. Let I be the upstream end and B the downsfteam end where the channelt clerations (E,, and E) on rir.er's centedire and surface water height (H, and H) ate knorvn. kt Xbe a location between A and B. At X, elevation (E_) is deril.ed from DElvl and watct surface herght (H*) is computed using If M LD=o andEx+Dr=o I E^D, r-L,-l lu.=u. -ut.E\+D\ M AD It" IfM.AD+0 orE,+D,+0 ll , (1, 'l he rr,here H, = rr,atcr surfacc hcight at Lrcation,,1, AFI = *ater surface herghr diffetence benvccn gaugrng stauons I and 13, AE = clcvation dift-ctcncc betNeen:'1 end -\l) Lli.rrncc ber.eerr L.na B al,'n$ fl'L .-rrin centefltne, F/,: rvatct slLrface height at poinr X, E., = stream channcli clcvation at location X, end /). - distance bet*,een ,-1 and X along thc strean centeilifle. ln gcncral, onc should select k>cations.1and B *here gaugng -stati1115 etc locatcd. Thr,rs, H, and H,, as rvell as E , and L-,, atc knor," n. 't hird, with \\'xter surfacc heighrs at cach pircl alon!! the ceotedine c^lculatcd, one is read! to com putc \r^tci hcights at pixcls off the celtetLine. With lhc,..LrmlEnn rha is lclel, calculation of ,,r , r 'e r ' u r l ,r r centerlioc pi\cl accordiog to r1.re assumpdon of lerel ',. r c' r_'1, ( r, -u-(1]'n (Lojj-rccLronr. c) Rcpcat the proccss for al1 oft- ceotedine oirrl. \orc: \{ lrogrrm r.l. \n cn o!c B, , rr ,, I ro.. 'cc:or' : ' jlan t,;- Crr.ss S: con.rplish this stcp.) Lasd1, oncc all of the surfacc rvarcr hcights arc T--_ calculatcd, both on centetlinc and ofl--cerrter]ire, a suifecc \\?tcrhcl+t lai'er (a grid) is cteatecl \.ith same spatial (esolution as undcrhing DEtrl. It shouid be pointecl out drat if the size of thc scarch radius iot thc ncetes! on ccntcrline pirel to oftline pi\el is dif ficult to determinc, scarching the entite studv erca H. e: r , a'r. 1r,,,.'.cr.r.bc\(f\ Lime consuming. 'l hete are othel mcdrods ar ailablc fot c,,.rld Lr rr r-rcrr. nr a- <. intetpolating rvatcr surfacc heighr based orr kno\rn \\'atcr surfxce height at nearcst points. lor cxamplc, Ltscd the inlersc distance s'ciglrtccl iotc4roLatioo. VV \\'ernet (2001) t- the sufacc *'ater height at en off-centerLine pisel is boiled dorvn ro tinding thc on- ccnretline pirel to s'hich thc off-ccnterline pirel shates a senrc ctl)ss section- This is achicycd bv ftnd- ing thc on-ccntctlinc pircl r,, ith shortest straight lioc distancc to dre off centctlinc pixcl in qucstion.'l here arc four steps inroivcd in tl,is proccdurc: a) ldcntifi all the oo centc inc pi\cls th^t ^rc *'ithir-r a specrtied radius of the off centcrlinc pL\c1_ b) Calcuiate the strejght Lire distance bc'nvecn the off-ceotedifle pixel and each on-ccntcrlinc pixcl using rhe cquation: D= (\ \r, + (r ,.r)1, , - r \ Q) of pirel (-,:,.. and.l are x and l coordinates of pirel 1). I = 1, 2, 3, ..., n and feptesents the series of oncentetline pixcls that lie s'ithln thc scarch fll wheie afld-1' are and t- coordioetcs dius (Ftgure 2). The x an<ly cootclinates ate tcla tivc to thc origro located at the iorvcr lcft cornet the studl arca as co\-efed bv the Dlill, assuming thc DEII u-sed is of a squarc or rectan gular shape. . Ir..:r\ rl .,,n renre-l crn\c L Lr.,rhai of the shottcst straight lir-ic distance to the ot| cefltetlire Pi\el. d \.irqr -l c.urface r,are- [eiqhr ',' rJrr ',ft Methodology: Modeling Flood Extents by Using DEM-inundation and HEC-RAS Models In dre delincation oF the flood e\rent, the 1-D DE\I inundatioo mrdci supcrimposcs thc calculated surf^ce \\,atet height lar.cr or-cr thc DEII lar,er. llecxusc both la\ers grids wirh the samc spatial rcso ^re lutron, the lalues oi surfacc \\.ater he+lht and llround clc\-ation at cxch pi\el are knorvn. To delincate rva ter/non-rlater (regulat tlot) or fl oodcd/non-floodcd (tl'"'d i"rv rrr.r.. o'te rtecd. ,, h. rc rl ^ .cr, "t r str:eam\ surtacc \vzter hcights (a regular oie end a flood o[e).'Ihus, nvo surtxccs of the warcr herghrs arc calculatcd. .\t e locarion X,lct lJ,._,,,,. b c thc regu lar herght and / /,-,.- bc the floodcd hcighr on thc t\vo surfaccs, rcspecti|eh,- Thcn, ' if a locationt clcr.ation (on rhe DE\I data)is S H.,,.,,,,,, t1-ren the location is classiticd as regulat sttcam area, . if its clcr-atiou is > 1 i,.,._,. and :l H,.,, thcn ". thc Iocatior-r is a tloodcd atca, or' ' if the elcr arion is > 1 1.,,,"., thcn the Location is non floodcd or dn-- The Ill:,C R.\S modcl sirulatioo with re!!u lar and I'loocl rivel surfece height sill also classif,r' -:::-' r. ::-. Figure 2. and 1) atc o. calc-r::: > Cross Section,zl Waier Strface lrc ss Sectron -tr Bottom H "l 1 Cros s Section I D lo_ l"x Datum IF A-D Figure 1' Illustration of thc inrerpolation of water surface height, H. at a gi'en location X on the stteam centcrline. A location alvay ft'om the centerline of a stretm Pq Figure 2. Interpolation of rvater surface herght fot an off ccntetrine location, c Fire poirts. p. p- p. 2 and P-, on the central channel ate shown as an exampre. nistances between c and aLr point" on i" .n^n*t arc calculated, so that the point(s) on the centetLine haring the shortest distance to Cwill be idcntified. Z/ten.! t each pixel lithin llooclecl, ol rhc stud\ area as rcgulat srrcam, dn (non f-kroclcd). accufacv assessment of flood extents derived by both models inundatio[ crtcnts from bodr moclels, first rve summanzc dcscnptir c statisdcs l3 Scptcmber 1999 agair-rst reo-roteh seosecl clarx inr mclrsurelnents obraincd at sc\crai sitcs. 'lhc clrte ancl si.c sclccdon lrc bxsed o[ a\-aihb]c ancillar\ datascts deteilccl rn the ncrt sccdolr. ljlror m.i ni e! .rr( 1r.. Irnqr .rr, \ r'r(.-i rpp rre .., ( r. rc. on .r,td in Methodology: Comparison and '1b compare the of thc rcgulxr strcem area. fhoded arca, and non- Methodology: Study Area and Datasets 'l'he snrch erca is on thc lorver floodplain of lloodcd atca. Ncsr, spetial comparison anallsis of thc e\tcflts at thc same ilos conditioo is cardc'd out t() quantif| the amount of agrccment bcn\,ccn thc nro moclels on a pirei bv plrel basis. If l pirel is classit-rcd as samc catcgotv (tcgulat sttcam, lloodcd, or nor-r'i1c.,oclcd arca) br thc nlo modcls, thcre is an rlgl'ccmcnt; othc^\,isc, thcre is a disagreemeflt. 'I hc s atcL/nrn rvetct or floodcd/non-floodcd boundeties deLineeted bl thc l)l-lll-inuncletlotr ancl FIII(I R.\S moclels can clifier so to understand thc r ariatiot of rhc boLrndarics sratisricallr', rvc ,,rscc1 thc marchcdl)xir / tcst of thc bor-rndancs on borh sidcs of thc rilcr channci. Irigurc 3 shous nvo scts of bo,,Lnclatics, onc ccnterlifle, and dne chrnnel ctoss sections. fl he centerlinc is dcpicted as straighr iitc t,,r' rrnl.i. r' - Vlnl \, y1'",'r'rn" cfu.:-.u, Lur. n ( rhc Tar/PamLi.co Rilcr (drair.regc area - l5l limr), (latoline. 1r cr)rers per of Pitt (iounn'on rhc \orth rr-est and Berutirrt CountL on thc east (FigrLtc'1). 'fhc Tar Rircr flots into Pirr from thc north\\'csf rnd c\its t() Rcaufi)rt ft) rhc casr. .\ficr passing thc bridgc of Flighr,ar lrr.rr.-e 11, it is callccl thc lil ico lUr er r" lit'i Grccm ille aod thc othcr at \\hshiogton (Figulc -l). Crccnr ilLc is thc Iergcst cin io Pitr (i)unh'. antl \\'rLshir-rgtoo is thc llrgest ciB in Beaufort Coultr'. 'l here xfc ahrcc nlej()r rcasons for choosing this perticular ireight and daih meao discharqe;afld ofl going ilood lcscarch in this erea has rcsultccl in se|cral in-housc gco-spatjal and rcmotc scnsing datascts ((loib\' ct 41. ccntcillnc (c.9., lriglrrc 3). rLtc .100 lrinallv, to laLiciatc thc I D Dl-i\l inundation mocle1 as rvell ',ls HF,C-R.\S modc1, rve elaluated rnodclcd llood c\tenfs aa a rccord lrrgir flood llov 3. .-u.l\ .rrc.r. f^,'J rr:,. .J Lr 1,.. r p c. ,r''r,. 'r"l'i,.,- l',rrn. . rr,l h'.r'i,,rr- ',ctr_ rrr\ "r r, rrrlr ir '1 , .qrl' ,,r,r: )r'1\'' r'' (- .Lll'l ,ll 'l. lrl ,r t.lr t'r. crl L rL r'1c. ir, irn("1t5 li,r \..t(' : rra,, r' m apatt ir rhis snrclr'. -\long each cross scction. t\\o infcrccpt1on pornts Nith rhc boundarics rrc obtaincd; thc distaflccs bcnr,ecn the nvo poiots and thc ccnrcrLinc arc calcularccl. (lncc Lhc clisrancc nrcilsLrl('1r_lcnts frrr all closs scctions afe conrputccl, thcfc afc f,vo scts of distrncc'lllcasLrrcilclrrs: ()rrc liom thc l) I,l\l inLLndation modcl and thr: othcr tr-on thc FIEC-R \S moCc1. 'l hc null hlpothesis (H,) for ftc / tcst is that thcrc is no diffcrcncc benr-cen thc distenccs from the nro moclels (i.c.. thc bor-rndarics etc stadsdcalh idcnocai), and thc altcrnatir-e l-upoth csis i[l ) is tlut a significanr differcnce erists bc t\\'ccn rhc t\!o moclcls. \ sigriiicant 1cr cl o f tI=(1.(15 is cltosen to test rr hedrer I /,, should bc rcjcctcd. Similatlt', disrance measurcrncnts and /-tcst \as bc cx1' ried oua for thc b()urrderies on rhe ofher side of thc roughh Figure 2l.X)1.). \\rang d aL. 20U2. \\ana 2001, \\ang a0c1 Zhcng 2(Xl5). llasccl on thc sfxrc\\i(lc lend use arrd land cor-cr 1ar cr crcatcd by thc \otth (lalrLina (lcntcr tirr (lco- llrxphc Iflfotnudon and .\nali sis. thcrc arc tlficcn land usc and land co\ cl tlpes rr itlifl the snrdl arei $ st an-rps cultilatctl arcas tcrc clomirrant land ctx'ct npcs ang 200.1). l3ottr>mland lblcsts/har-dr,,ood ancl the stuclr-etea). 1'hc botton-rland bt oi clecicluous and \oodr r cgctation tlllcr thafl 3 m, \'hcre croNn den sin is ar lcasr 25" u. -frLpclo (,\. allahrf and c\T)rcss (Cr4r.rvr') atc rhc major spccics. l he cultir atccl lands (about l3{"o oi ests/iratds'rrrd sr,,amps dre areas occupied b\' croPs of cortorrr corn, tol)acco, ^r-cas and sorbc^ns. L-r additron, thcrc arc dcrclopcd ar ees, rvhich count firr abour 30 o of thc studt atea and a,:c meioh' concentratccl in tirc r icinitr of thc ciries Figutc 1. i ::. Ttu Xar/h Cnnlitu Ceo!tublto. Brrundades deli\red fiom HEC-RAS mrrdel I '-< CenterLine --+- --J/ I Fltrodsd 5.t ="<\--l>\=__ / ,/ --j/ fri)ln l)Fl\t rlunorltlon mrrdel Figure 3. Lllpothetical boundarics der.ir.ed from the FIlic-R-\S an<l Dl.]rI in'ndarjon models. Ninc cross-sections erc plottcd at c\'coh, distributcd intet\ a1s. Figure 4. Landsat I E'tlIi dara of ba'd 8 on 23 Septen.rbcr 1999 (path/ro*, 1.+/35). The studv oudi.ed b' the dotted lines cor-crs thc l ar/Pamlico Rir.er floodplain, North car.lina arca of Green\-ille end \\rashington (\\irng 200.1). Thc statc$'idc land u-se a[d land cover date tre uscd to estima!c fhc trlanningi z coefficient of roughncss at each cross section elong thc nr cr channei, u-hich is onc of the most importaot ioput paramctcrs to dte HEC R\S model. Tirc sr,rrfacc \\'ater heighr and dischargc data collccted at thc gaugrng stations of (]teenlrlle and Washington are gilen in lhble 1. Thc Dl-lll-inunda tion model onlr uscs tlte \\,atei sutfacc hcight, rvhcrcas HF,C-R.\S model requites borh heiglrt and dischargc akrng s'ith othcl pre\.iouslv discussed in puts. Tn,"rndation extents are modclcd at n\''o rcprc sentad\-e tloodinE florv conditions: a flood-srr+ie flo\r on 28 ltbrual 2003 and a record hgh flood florv on 23 Scptember 1999. Ihe florv condition on 28 Julv 1999 is used as the tegular florv (because of thc ar ailabilin'of Landset I-TN{+ data). l hus, con-rpari -sons oi thc nodcled results undct drc trvo disdnct flood florv sinrations as tefcrcnccd to the regular flow condition cao be petformed. Dlitrl data tbr drc studr arca werc obtxined tr<rm rhc USGS Nanonal l-lelation Datasct. Thc l)ll\l has a 30 m br. 30 m horizontal spatiel resolu- tion ancl a rertical eccutacv of !1m (USGS \l-D Thc tcttain rvithin the studv area is flet $ifh x minimurn clctetion of (J.0 n-r, a mcdian of 3.5 m. a r'r.r\''1 urr of 2).i rr. ,r tr.rrr ot 'l - rn, ..nd r . rn datd dcviation of 4.{l m- Thus, anl silyrificant in crcasc of ri\'cr'.s surfacc vater height could inund^tc a Largc arcr- .\lso, it shoulcl be ioted dlat the south sidc of thc'lat Rilet has considcraLrl)- morc relief than fhc north sidc. Conccir_abh', thc eccuracl ol 2007). l)EXI has a signiticant impact on floocl mapping and thc arailabilin oi hghct accutacv l)1,,\l u'illimprole floodpiailr modeling asscssmcnt. C)fl thc othcr hancl, because D!,\I accutacl should similarh irnpact both models, thc NED Dlr\l is consiclerecl as a telson ablc choicc fot con-rparing dre trvo models undcr the same set of conditionsRemoreh' scnscd imagett arcl aerial phorog taphvrrere used to idcndfv floodecl a11cloon {loodcd arcas to aid the lalidation of inundatir;rr cxrcrrrs rcsultiag tiom thc nvo n-rodels. Thesc datasets include Landsat I F,'l trIi clata accpired on 2l3Juh-1999 and 23 Septembet 1999, and obiiquc acrial photographs takcn orr 23 September 1999. lhcsc datascrs, con bincd s'idr fu.'rZ obseh'ations mxde in Ocrobct 1999, Tab-e rvete used to idcnriil tveiry fl\-e lloodcd sites, twentr-n\-c reguler nr-cr sitcs, and t\vent\-6\.e oooflooded sites. Thus. the accuraci'oi dre modclcd tlood cxtcnrs at the record high flood flo*'can bc e\ aluated at thc sc\_cnt\ -tir_c sitcs. The erees coteted bl thc sitcs aocl br catcgorics arc: Legular flrer atca of ,+.61 kmr, floodcd area of -1..11 kmr, and non floodcd arca of 5.32 lim:, for a rotal oi 1-1.36 kmr or 9.101, ofthe entire studv atca. Sincc rhclc is no othcr temorelv senscd at in .ti/ ddt^scts on 28 lrcbruair 2003, no lcriflcation of the modeled resuifs were pertoimcd. lt should bc notcd that d.re USGS Digi tal Otthophoto QrLartet (I)OQQs) accpitecl io 1998 1 l)atc Disch:::: : \\hrer::;:-- s'ctc uscd to aid the initiaL idcntihcation of thc srcam ((n(rlinr/'rthcl)l \l rnd irrJr,r.t ca cgorc. r,'sites \\,hete gtound acccss rs impossiblc. Iiinalh, aLI digital datascts used for tJris studv hale bceo rc projccted into ttrc U nircrsal Tran sr ctse \{etcator (tr 1 \ D coordinatc srstem using thc \\bdd'Geodctic Sl'srcm 19{3-1 Rcgula:: Flood s:-'---;:l (\\GS8.1) models iot the sphcroid and datunr. llc c. rci Results and Discussion Tt'o la-i'crs consisring of \\:ater and iofl \\:ater catcsorics co1€ring the cntife stud! arca \\'efe fttst crcarcd nr thc rcgular dr-er f1os. condition (lable 1) using DFitr{-inuodation and LlliC-R\S modcls, rc floodcd r:.- -1r and or_crl:..::-: mal)S rcP:aira .jRirrt s-a: :: . :l c.... In tl r jgu'r. .hr $1rrr.rfc.. shot'n in black and non watcr atca in rvhite. lhc 'l(.o\(1. Fid,n main chaoflel oi rhc 'far Rilcr is clearlv dclincatcd, and nvo ffibularies (Chicod and Tiantcrs clccks,Irig ulc 5b) arc idenrified. \'isual examinetioo of thc mocleled tcsuits indicatcs that drc n'arcl ar-cas rlav bc sin-rilar. Hcxvelet, in thc upstream secti{)n rherc is mote atca classihcd as rvatet bv rhc FIIIC ]L\S modcl than bv thc DEtrI inundarjon model {Figure 5). .\t the regulat flor'', s'atcr areas ate 1 1.85 Lm3 and 1(r.95 Ln', according tcr thc DIrIl-inuodarlor model dfld HFIC-R.\S model, re-<pcctir eh' (l'able 2). Four additional lavers \\ ere modeled for \\'eter and nofl-$ater categolics iof thc flood-stagc flo\' and recotd high flood f1os.conditions using the nvcr modcls. Thc \\:atet aLeas on ell thcsc tbur ialcrs in clude the regulat tivcr suriacc atea (e.g., Figute 5). :. \\,hich sh.,i .: tccord lig:r :. .": In drcsc ilE:::.t.t..L .t.- -- : -: i in rvhirc. :i :.: (lrigurc 6.. :::r:r .: \t tl: e rh.r.orTr'-.--north oi ria :: :: lief on drc :: :: contdbudc-a : ,--: iltundation ::_--,:: : occlrf\\-ithl.:l_. -: p.rr .h". ' r" oofl tloodcc ri,:: i rr,r .\ rnnt. -: -. Table 1. Rivet data measured at the Greeflyille aod \I'ashington gauging statrons. Darc 28/07 / 1999 28 / 02/2003 Dischatge (mr,/s) .r.39 / N.\ 302.99 Water height (m) 0.37 / 0.27 3.30 Table 2. \{odeled Recotd-high flow 311.19 0.31 1816.26 ,1.96 / / 2152.08 1.60 extents of regular rivet, flooded, and non-flooded atcas (lirn:) at three flow conditions. Regular Flo oded Non- floodcd HE(]-R.\S DIr\{ inundation 11.85 16.95 11.85 xxx + 1,10.02 HEC-R-\S 16.9 5 DEtrI inundation 11.i15 FII,C-R-\S 16.95 lJ Flood stage / / 23/09/1999 tl i\l- 1flU n w'hich should be cxcluded in otder to map the flooded area. This cxclusion is donc through tecoding and ovedaving opetations. Thus four inundation maps reptcsenting thc flood extents s,hen the Tat R-ir'-er rvas at a flood stage florv (Frgure 6) and at a recotd-high flood florv fgure 7) rvere generared. In tlese f,gures, rhe regular river areas aie sho$rn in black, the flooded arca in gray and flon flooded atca rn rvirite. At the flood-stage flou, on 28 Februatv 2003 (Tigurc 6), there ate latge flooded arcas surrounding the tegular tir.et atea, and r-trote flooded areas to the north of rhc riYet thao to the south. l'he io\r€i rclief on the nordl badli than south bank is a factot contibuting to this difference. Comparison of both rnundation maps indicates that morc disagreements occur within the upperhalf of the snrcl,v arca lnorur- \'-est) than at the lo$'er half (sou*reast). fhcrc are non-flooded islands (surtounded br' flooded area). Fot cxample, thete is an island in the middle of the 56.10 45.61 7 7.80 74.35 89.02 4.35 61.31 65.67 9 flooded arca on rhe inuodation map dedr.cd from the DFl\I-inundation model figure 6a), aod an island of much larget size exists at thc cortesponding location on the flood map detivcd ftom the Hll,CR.\S model (lrigure 6b). Bodr islands ate identified using black attou's in thc ftgutcs. Frgure 7 shorvs thc modeled inundation ex tents at a rccord high flood florv condition. Thc cx, te11ts are risually similar The aerial digrral photo_ gtaphs acquted on 23 Septcmber 1999 and ground truth collccted in October 1999 indicate that thc majoritv of flooding occutred on the north side of tl-re rivet, rvl, ere thc eler-ation is much los,-et thao the cotresponding part on the south side. The sLighrly higher eler-ation on the south bank is onc major factor to its smaller floodcd area as compated to the notth side. Trvo noticeablc disagtcements of thc e\tents as pointed bI t\\,o paifs of black arto$,s are obserl.ed: onc occuts neat the notthvest cotncL and tie other ncar the castctnmost location. In addition, Zt '., l' ll t. DElr,'tl-Inundatlon Tlte \o,; a;- :r: N,Ioc1€1 (a IIEC-RAS h4i:de1 (tr Figure 5. Regular rir cr atea modcl (irr blacl<) and ar.rd b) non rvaaet atea (in rvhite) dclivcd ftom a) DIi\I inundation at a repplar flo*, on 28Juh'1999. HllC-R\S model Figure 6, h:=c-: a flood stage:i:: iloocei Figure 6. Inundarion ertenrs deri'cd from a) the DEr{ inundation model a ancl b) rhe HLC-R\S model at flood stage florv on 28 Febtuarr' 2003. 'l he rcgular dr.cr- area is in blacli, floodcd arca in qra\.. ancl non flooded area i. rvhite. Unflooded islaods cxist, as pointecl b1. biacli arro* :. .,, .rn-pler. 12 Zheng d' lvn,g The l',axlt Ca,':s ( in compatison sr ir f flooded area sffeam sectroa si islands ate noq- d eas are anribured ard discha4r rol pated to 28 Feb,rr marizes the a:ra r tion exrenr rup- -l stage ro recor&b flom crease 56-10 undation moddi modei). r The ganel eled inundaooo r quanrified -&e dq pixel basrs, The r by the nro oodel km: or 9t1.5' ' o 140.34 km: or 89. j Cfable 3,. Tail€ disagreemelr. br The res:Its, from the srream r or flooded noGd of the inunc;io July 1999. .'and.l boundaries :r'e li and 1.6ji aod O tir-elr: The : rz.Lr null hrpotheses a vatef/norr F4tct DENl-inunciaoo r tisticallr ditr'er€c- fot ahc ilood .4r dr are all gteater conclude riat de on thc nor,h rod same. Thu-. a: Figure 7, Ioundation extents derived from a) DENrt-inundation model and b) tecotd-high flood flow on 23 Septembet 1999. FIEC lL,\S model at a rlr models hare co,q The frndrngs ?f,e rt inundation aodd put to the i':el e flood erceor,. at d 'l/t( \atth Coi)/lnd Geou..dret. io cornparison tith Fisurc 6, there is mucl.r grcatet iloodcd area in |igute 7, cspecrallv rvithio thc upsltcam sccdon rvhete fhc maioflt\ of non-floocled islands are nos' flooded. Thc incteascd flooded areas ate attributcd to thc highcr \\.atei surfacc lerel and dischargc volumc on 23 Seprembcr 1999 compared to 28 |cbtuarr' 2003 (fablc 1). 'thble 2 sum marizcs the arca oi cach catcgorv on cach inundahon extent map. -\s thc dr-cr changcs tiom its flood sfegc to record high flood florvs, iloodcd ereas increasc fton.r 56.10 to 77.80 kmr(>ascd on DljtrI inundation modcl) and -{5.61 ro 7-1.35 Lrnr (HE(l,R\S model). 'L he spatial colr-rpatison alxlvsis of thc mod clcd inundador-r maps at thc samc tlos' condition cu:n:lcJ . drc J,cn, 'r .tr'c.m.n ,,n.r pir..l,r' c...:t.d.,s - rr :,rnrrc.,rcqu"1(... bl the nvo modcls is 150.18r km (oi a rotal arca of 157 km) or 95.1'li, on 28 Jul.r' 1999 naps, 1.12.08 lim'or 90.59/o on rhc 28 f-ebrual 2003 maps, and r<lbasr.. I I c..r'er kmr or 89..10i, on 23 Scptcmbcr 1999 maps 3). thblc i also detalls rhc a€yccmcnrs and fl'ablc drsagtccments bv rhe cafcgories. 'l he rcsults oi thc / tcsts on the mean distance ftom drc stream cefltcdinc to thc rvatcr/non satet ot flooded/non tlooclccl bour-rclaries on both banks of thc inundation maps arc shotn in 'l able.1. On 28 Jolr' 1999, I andp r-alues for thc \\-arer/noo-wxrei bounclancs atc 1.573 and 0.122 lot the notrh banli ancl 1.63i and 0.109 iot rhe soutl, banli, rcspec tl\'ch. 1he I ralucs at borh banks suggest that rhc r-rull hrlorl,cscs arc nof rcjecrecl, indicarirg rhat the \\1lrrr' .lor rr,. rr lrnurJ fl(.r r(.u :.1! t.uri th(. DI-\l-ir-runclation and HIiC R-\S models are ior sre 1.10.34 ustlcallt dtittircnt. Bccause rhcf r.alucs of the I tests fot thc flood srige flo\v and rccord-l.righ flood flosale all gtcatcr than or cqual to 0.103 (fable:l), rvc conclude that rhc flooded/non floodcd boundanes oo the nofth and south banks are statistrcalfi.rhe saffrc, t hr"rs tir, the l)FiNl-inundaoon and HFIC,R-\S moclcls hare comparabLc resulfs in this srudr area. Thc finclings atc lcrl cncour^gjeg. Ne\r, thc DEII iltundation moclcl and thc FIIIC R \S modcl rvcr.c puf to the hnal tcst. lhe accutac\.of rhc modeled tloocl crrcr-rts at the sercntr-fi\.e sclcctcd sites werc Lj r.alidated against the ancillal- datasets collcctcd dut ing ar-rd aftet thc 1999 llood, as dcscribed in the prcrous sections.'l he rcsults indicatecl thar bod.r rnod els tcachcd high accuracv (Iable 5). Based on thc DEII inundation modcl, the producer's accuracics ate betrvcen 88.31'o and 99.3r.i, en<l uscr's accuracies 93.1u'o and f.i.6'1i,.'t hc or-etaLl accuraq' is 9-5.1oi,. r :tjl rl l rh .r(\ rr.ic e. ,rr, .lso ror.l 1ql thc Fll-(l-R.\S modcl (l rbic $1 1..in,, -5). Conclusion ,\ hldraulic 1D Dl:r.Il inr,rndation modcl, rvhich is simpler lhan rhe srandarcl complc\ l,D HEC-lL\S trodcl, has bccn clerelopcd. Comprrcd with thc HE(l R\S modcl, rhe Dliil-inuoclarion modcl requires t1\\,c1 tnpur parametets thatare reacliL| alailable. Thc DF,trl inrLndation modcl is also casicr to implcmcnr than rhc HLC R-\S modei. l.utthcr.n_orc.cu np..ri.,,n.L(n\rc.l t1 r-rJ. i.,ncrrcn. lr,,m the models and accuracl elaluatioo tbr a iloo<l cr ent on thc floodplain of thc'lar/Pamlico Rir.er, Nonh (iarolina hale sho\\'n that fhc results fron drc nlo moclcls arc r crv similer and bodr tcachcd or.ctali ac clLracl greater than 93'r',,.'1hus, the DIIII inundation model can bc an efitctir.c alternatir-e to d1e morc cotrplcr FII-C-R\S model. Refote concluding, t e \r)uld likc ro mcfldon drrce tecent dcrclopmcnts: the crcatio[ of the DIIII fot thc statc oi \orth (iatoLina, implemcntation of lnote fl\_cl Eaugl'1g staUons bt' thc USGS, and ar-ai1 abiJln of rcal-time gaugc data. ,\li of rhcsc dc\alop mcnts posrti\-ch impacr r|c appJicarior of rhc DFI\tinunclation modcl. -\itet thc 1999 tlood in easrcrn North (iarolina, thc srafc of North Carolina initi atcd a starewidc flood mappingprogram (\iC Fitxrd plain i\lapping Progtam 2007). Onc of rhc producrs dorvnloedablc for free from thc progtan is the starc s'ide Lieht cletection and rangrng Q,ID,\R) dcrir-ed DIllI. 'l he DI-trI is oi 15 r 15 m (5iJ s 50 ft.) rcscr iuoon, and ITas a r_ertical accuracr.of apporir-ratch' 0.2 n-r. One distinct fearulc oF the nc*, LID:\R dcrn'ed Dlrll. as colrrpared rvith othet DtIIs (e.g., \.-LD D1-110, is rhat the LID-\R deriled DEtrI has been hlclro corrcctcd, i.e.! allthe channcls of stteams l4 Zhpr, !" Table 3. Spatial compadson of the inundation extent maps detir,ed ftom both modeLs at fiiee flow The arca is in Lm'?, and the petcentage within the [l is computed out of the tota] studr.area. (a) A regular flow (07 ly Lt Tbe llartl Ctn:z; ( stages. Table {, \f: I /28/1999) HEC DEI,I-inundation model Non-u'atet atea Water R-AS model Nnn-s7,tPr erc, \Y/"i.' \\iter rm- 139.17 [88.7o;] s.es [3.8%] 0.8.1 [0.svo] 11.01 17.0%l On nor,! On so:& h h (a) A flood-stagc flow (02/28/2003) HtC Non-flooded atea R \S moJel 1.07 Flooded atea 87.94 [56.0%] 1 .02 [4.sv"] Rcgular rir.er atea 0.04 [0.0%] 0.i10 Flooied oc [0.7%] 43.L3 p1.s9rl ls.l%l On nord. On so::fr 0.00 [0.070] s.94 [3.87C 11.01 [7.0%] have been h h l:r:uel aod po-in: been temor ei io l,vsts, (a) A record-high flood flow (09 /23 /1999) HEC-lL\S model Non-flooded area 13.0%1 62.67 p9.9ah) 4.64 1.3412..84^l Regular river area 0.00 [0.0%] 67.09 1.21 sreaal & flrgule 8a, are bar (Figure 8b . S-rce sufes su{face elesll Flooded area Non-flooded arca l.looded area ample, and or-erpassc 112.7y"1 0 10.0%l 6.37 l1.1y"l 10.8%1 10.58 16.7%l sfr instead oi tha: oi dto correcrioq i r nuiq of rzter ic s surfaces befleeir ( hydro-correcdocof a steam 'cecoo r this simpliEes fre undation modei r lina). The DE\t-n used oo a sre3:E s The Narth Camlina Ceorrarher) I/olane | 4, 2006, pp. I j- 28 15 Table 4. N{atched-pairs l tests on the $'ater/''on-water or flooded/non-flooded bouodaties derived ftom DEM-inundation and HEC-R-\S models at thtee flow conditions. (a) A regular flow condtnon p7 /28/ 1,999) Watef /riori-water boundary On notth bank On south bank @) At tle 1.573 12.20k 1..633 10.9% flood-stage and record-high flood conditions. flow (02/28/2003) r p The flood stage Flooded/non-flooded On north bank On south bant boundary The record high flood flow (09 t /23/1ee9) p 1.620 11.1"/,' 1.612 10.3"k 1..490 14.30k 1.648 1.0.6% have been manually and cleatly delineated by analysts, and portioris of bddges arrd ovetpasses have been removed from *re DENI (Frgure 8). For ex tions where prefetably no major inflow from tribu taries erists. With rhe inflow (ftom the ttibutaries into the main steam) the sutface watet height at the ample, stteams cleady depicted by the LIDAR DEM (Figure 8a) ate barely noticeable in the NED DEM (figute 8b). Since the airbome LIDAR sensor measutes surface elevation, surface elevations of btidges dowostream gauging starion will be aqmented. Thus, and or.erpasses will appear on the uncorrected DENI instead of that of dre undedying surfaces. The hydro cotrection is necessary to ensute the florv continuity of watef ifl stieams uoder bridges and oo toad surfaces beneatl overpasses. Thus, because of the hydto-cottection, the delineatioo of the centet the of a steam becomes easy or may already be done; *ris simpJifies tlre implementation of fie DEN{-in undation model (within the srare of Nortl CaroIna) The DENI-inundation modelis designed to be used on a ctream .ecrion berween rwo gauglng sfa- the inflow can affect the model output. Although this may limit the applicabiliq' of the model, the er,rcr increasing numbcr of gaugrag stations in the United of a ptoblem. Fot example, atea, thrce additional gauging stations States is making this less in the study (between the Gteenville and Sfashington stations) har.e been recently added (LISGS NXriS 2007). The surface water heights measured at the new Chicod Cteek and Ttanters Creek stations (e.g, Figures 4 and 5) will help address the influeoce of the tribu taty inflows (to the Tat River) and estimation of the sufface water heights at the meeting poilts of the Tat River/Chicod Creek and Tat River/Ttantets Cteek. The new gauging station (SR1565) near Grimesland at Tar Rivef flot only divides the stream ro Zhenlr d-7 Wa ! The North Cz-,is t Table 5. Error matrix and classiFcation accuracy derir'-ed from bo*r modeis at sitcs of tegular water, flooded, and non-flooded areas. The date is 23 Seprembct 1999. The atea is in kmr. (a) DENI inundarion model Reference data Model o.otqwt l.koded area Non-/looded Flooded area 1.17 0.31 Nan llooded ared Rega/ar water area 0.24 0.00 4.98 Tatal 1.41 area Repalar uatet area 0.24 0.08 0.03 ProCucer! acctracr :o ,, Total 4.72 +.31 5.30 4.34 4 6i 1416 rccJraclo ,r L ser.q Flooded area 88.3 94.6 Non-flooded area 93.9 93.6 Regulal rvatet area 99.3 93.1 Overall Accuracy 95.17o Figure 8- Sll comP2rE (b) HEC RAS N{odeL segment betsreed Model F-/ooded output area arca 4.22 Floaded Non,fl.aaded area trt tlvo segmeolr- Refetence data BJgklar water dtea Tltal 0.23 0.18 4"63 Nonflooded area 0.19 5.04 0.06 5.29 Rega/ar water area 0.00 1.44 4.41 0.05 5.32 1.39 7bt. 4.63 t4.36 dent measurem tictmore. dre of nearlr L9 18.[f,8 more lil<elr 6ar d two statiofls. Fro herght meaolu one can use the s extent scenrlos tt Producet's Flooded area 91.1 95.7 Non-flooded area Regular vater alea 95.2 94.7 98.9 94.8 Overall Accuracy 1 tty The h{tr deai the DE}I-inuodai ing the need. tor situations br 6e d agencies at ditt-erer situations $ here d 9 3.19ro logrc/ hr-draulic hc implement dre R'\S, u TELE\L\C- Figure 8. Strean-rs or centerlines are cleady delineated in the hydro-corrected LIDAR DEX{ (a) compared to the USGS DEN.I (b). The DIIN{s cor.et ateas near (east ot) Greenr_ille. NC. scgment betwcen Gteeor.ille and Washington into two segments, but also providcs another indepefl_ dent measutement of the sutface rvater height. Furthcrmore, the USGS cutrendv maintains a network of nearlr. 18,000 gaugng stations across the country. The high densirr of gaugillg stations has made it more likelv that thete is no major tributary between two starioos. Finallv. using feal-time surface water heighr reasuremerrs ar.alal_r c ar gruging .rauq-,t., ooe can usc the model to simulate a range of flood_ cxtefll scenaflos 1n an e1rcnt of a flood. Thereforc, tie DEN{-inundation model wi11be capable of meet ing t-hc needs for quick imolemerradon in wgenr situations by the flood management and mitigation diffetent gor:€inmeflt levelsJ especially in situations rvhere thete is a lach of suf{icient hydroagencies at logic/hldrauli, knorr ledg< and ti-ru,rd,esource\ ro irnplemeot tJre mote complex modeis (e.g, HEC_ RAS, TELENL{C 2D, and LISFLOOD },p). as Refetences Ackerman, C, T., Evans, T. A., and Brunnei, G. W., 2000. HEC GeolL,\S: Linking GIS to hv dtaulic analysis using ARC,/INFO and HECIL{S. In: N{aidment, D., Djokie, D. @ds.), H-1 drokgi and H-1dua/i. Madeli g Srlj)part uith Gea Infomatitt Sylnnt. ESRI press, New york. Acrement,Jr., G., and Schneider, V, 19g9. Guide graplic for selecting tr{anning,s roughness coefhcients for narural channel and floodplains. LlSG.l1.eth ica/ Rqon W.\'P2ii9. Al-Sabhan, W., Mulligan, M., and Blackburn, G. A., 2003. A real-rime hv&ologrc modei for flood prcdiction using GIS ar'd r\eW\l' Conpaten, Enirannent, and IJ tb.ta SJnenr.2j,9-32. Bates, P. D., and De Roo, A. p. J., 2000. A simple raster based modcl for flood inundation simu_ lattor. .loarna/ o.f Hldro/agy,236, pp. 5l_17 . Bates, P. D., Wilson, M. D., Hordtt, M. S., 2006. .y'. floodplain inundation dynamics ob_ scr\-ed using airborne synthetic apeihrre radar Reacl-r scale Zhcn.g imagcrr: l)ata analvsis aod modeling. /azrzzl [ | fi nkgl, af' 328 (1 -2): 306-318 Chang, T. J., Hsu, M-H., Teng, W-H., and Huang, C-J., 2000. -\ (lIS disributcd t'etcrshcd modcl tbr simulanog flooding and assisted tnur,rlatiotr. .larrnal of tc .lmeictt V'aterRetoutut .,7!!aid/irt. 3 6(5): 9r5 988. Chow, V T.,1959, Orlen (banw/ I lldnulit.\IcGrawILIL Book Companr, Ncrv \brk. Colbn J. D., Mulcahy, K, and Wang, Y, 2000. lodeling floocling extcnt from Hurricane Fiold h tlre coastal plains of North C,atdtra. Enitr ronlrrn/dl I Ididrd!.2: 157- 1611. Correia, F. N., Rego, F. C., Saraiva, M. D., and Ramos,I.,1998. CoupLing GIS rvirh hldtologLc and lrldtaulic flood n-rodeling. V'dler Ra;our*t llanagentn/. 12: 229 219. Dobson, R., and Li, X., 2000. il hc accuracv ard cfltcicncl of CIS-based floodplain detetmioa tions. In: \Iaidment, D., Djokic, D. (L.ds.), lJ1rlralogtt and Illdna/i t\Iodling.f gfoi tuilL Gea gtphi tnlirnatian Jj'tla,,z.r l.)Sl{l Prcss, Ncrv\.orL. Galland,J. C., Goutal, N., Hervouet, J . M.,1991. 'i FlLEltrl.\C-a ncu, numcdcal-modcl for soh'ing shxlk)\ \\.efct eg.r tlons. .,1ird te! in ll'aler I\t razr,n. 1.1 (3): 13ii i.lli. Horitt, M. S., and Bates, P. D., 2001. Effccts of spatirl rcsolution on a iasrer based model of flood flors. Jarrn al al lJldrolagl. 253, pp. 239 -249. Horritt, M, S,, and Bates, P. D., 2002. Evaluarion of 1D aod 2D numcricel tnodels fot ptedicting nlcr flood inr,rndation. H y/ro laELu / P 'aallrt. 268 (1,:l): 87-99 Horritt, M. \O\' S., Bates, P. D., and Mattinson, M. J., 2006. I,iffects of mcsh tcsolr,rtion en<l topo graphic rcprcscntation io 2I) finite |olumc models oi shallow \\'afet flit'ral t1o*-. .launal ol t lydn log. 329 (.1 -2): 306 311. Hunter, N. M., Bates, P. D., Horritr, M. S,, De Roo, P. J., and Werner, M. G. F., 2005. Utilin' looJ r"f Ctffcrcrrr .r..r.. r1e. t'- c.,Ll'-r'rnq undatioo models s'ithin a (]LUE liames.ork. I lydnkgy and Eartl Jjrnn Sienn.9 (1):172-130. Hydraulic Engineering Center, 1997. I IE(. P-/LJ' Riler .1nahri.r ,ten: Hlrurli Relirenu llanta/. Hlclraulic Enginccring (ienter, Dar-is, C,\. V att.2 The Kraus, R., 2000. Floodplain determinrrion using --\rcYicrv CIS and FILC-R \S.ln: trlaidment, D., D (I:ds.), lfinln$t atd H-lduttli Nodelin! J ffarl witL Cuograpl:it Inlilrndti)n .1.)ltenr. ESRJ Ptess, Nerv \blk. Mileti, D. 5.,1999. Dltat/,::r: $ dcsign: d rctrlcvenl r.tl nutunt I ha:.a r* h the L- nttel S tu tr.t. John Hcnr\ Djoloe. 1)ress, \\iashilgton D.C. NC Floodplain Mapping Program, http:// *:ncfl oodmaps.cotl. I-ast acccssed: Dcccm ber 2001. Nicholas, A.P., and Mitchell, C.A., 2003. Nun.reri r,,'s cJ 'lmLrrrno" oiu.erL,tr'. 1'1 '6s q:. 1ep6 3-"phi,,'11. (umfl( \ flaodnl3 - rn\'rr"rIn(n, Hylrologttal Pro,v.r.,er. 1l (:l): 72i-l+0. . USACE, http:/ /*.sthec-usace.atmt.mil/sofnvate/ hcc tas/. 1,asr \ccessed: Decembet 2i)0f. USGS NED, http://ncd.uses.gor-/. Last acccsscd: Decetnbet 2001. USGS NWIS, http://\\.aterdata.usgs.gor./nc/nrvis. l-ast accesscd: Dcccmbet 20lll. Wang, Y., Colby, J., and Mulcahl', K., 2002. -\r.r efftcietrt n-rethod fot mapping flood exteit in a coastal tloodplain b\' intcgtaflng Lendsat '1trl ,rnd Dl-l\t drt^. I tendlil dl .la tr// .l Renate S e ni tg. 23 (18) 3681 3696. Wang, Y., 2004. Using l,andsat 7 t ll dara acq,,rited dals aftcl a throd e|cnt to delinearc thc maxi rnnm flood on a coastal floodplain. l landhzl1Ltl .larrna/ al Reno/c Stn.ting.25(5) 959 9 Wang. Y t"1. and Zheng. -t., f,![$. t-,,np. rr* ,r ' 'i <ligital clcr ation models and undetstanding .hririrnp..'. r'n 1 2002. C; Ll coastal floodplain of (\,(rnr?ll .L4'r. oi Notth Carolina. laanrl c ionJ Hd prdt 1I iau, 6(.1). 34-4A. Werner, M. G. F., 2001. Impact of gnd sizc in GIS besed floocl cxtenf mappine using r 1-D {1orv nodel. Phy.tit.t ani CLutit/ry' al t/tt Earlh Part B Hlln/o3y Orat and .1tnn.tphere.26, pp. 51 ,. -522. Wernei M. G. F., Hunter, N. M., Bates, P D., 2005. ldentitrabilin of drstriburcd floodplain toughness \.alues in llood extert estinraooa.Jrl1/- ,.t q [[ ,.0,. 11.] I t:1.11 tWilson, C., and Ho(ritt, M., 2002. Ileasuring thc florv rcsistance of submerged gtass. I lldtoiogiLa/ l)lz.r.rra. 16, 2589 259f1. Xorii C ;' ': Wilson, C. {'. \ and Bare=, "C)pc::::-.:of sub:-r:--rtlratr,;.:,':;:-, Yang, C., ani T GIS br.r::l pLain Al lla r :e --: "::': 56-,5- -. r Wilson, C. A., M. E., Stoesser, T., Bates, p. D., and Batemann Pinzen, A., 2006. Ciosute ro "Opcn channel flos, thtough diffitcnt forms of submergcd flexibl e rcgetaioi' . loarnul al' I 11, drarll Eu,ginuing ASCE. 132 (7):750-150. Yang, C., and Tsai, C., 2000. Der.elopmcnt of a GIS-bascd flood infotmation si,stem for flood plarn modcling and datnage calcultiot. Joarnal aJ theAm idn ll'ahr 567 517. RatawLet lroiaiot.36 e.):