|
<br />1.'0
<br />
<br />R.l. H"rJ,"~I"I.ll"""",I'ifH.,'JroI"t-,"116t1999) 11~_/J6
<br />
<br />
<br />T~"k ~
<br />n..lf"",,<>ir..,.'lulI"" <'nlhC"po>kQUlpw<ii..:harseCm'. ')
<br />E'<n' !'.t..u.OO'I',"..1i"':lI..r1:c,m'.-'lformc.h
<br />
<br />E.lr.,I }7.,>..l J~.S. ~O.}~ 5}.8~ "'" ~5.66 ~~.93
<br />E..r., ~ :.'.-J7 :6.;;5 .\1.90 )).93 >-I.OS 35.09 3HO
<br />E...,,-, 14.-0 17.6, :U.:8 ~1.13 21.20 21.75 21).76
<br />
<br />oJ. Results
<br />
<br />Th~ usuli5 :If( di,",u5~d in I~ order lisled before
<br />folIo.... (J by :l. synopsis or" !he generalln:nds observed
<br />r"r0mth..malysis.
<br />
<br />4.1. O:q!.r hyJm-graph
<br />
<br />\\ben bulk flow ch;lf;l.clcrislics. such as the peal:
<br />.:1i;.:ha.ru. the \olurnc of th~ outflow hydrogr.l.ph
<br />:l.llJ Ul~-r:J.e;3.D flow \\CTe analysed, a standard pallem
<br />.....:li idcntitied This is illuSlTated in Fig. 3. which
<br />sho\l.s !hc re;.ul~ from e\'..nl 2. As me spatial resolu-
<br />tion in,;uJ...~5. the bulk dow characteristics increase
<br />from mc;h I to 6. :t.I!hough !hey decrease for mesh 7.
<br />T1:e;e i, .uI in~,.e:L.~ in me peak discharge from 23.60
<br />to ::.1l9;n';.-: C-~Sl(l from mesh I 106. then a
<br />je.:Te~;b... :.I<C from me;;h 6to mesh 7.A similar
<br />:t;l::I:m ,,; ,irr.i1ar proportion is identified if other bulk
<br />jo.... ...;;,ri::ble5:lTe ;;.naly>ed.lfthc results are analysed
<br />:ror.l rr.c,h -: to mesh 7 there is minimal vMalion in
<br />?C:J;di;.:)arge.~'ith.peaJ,;discharlesincreasinl ~ro~
<br />33.9:to:SJ)9m:s-'.~ndecuasingt033AOm s
<br />:'or :::esh -. TIs indici4es. as previously suggested. that
<br />th..re is ;:lI oFtimum ~5h resolution beyond which
<br />re~lLi r:l.:.\' ~o[ ;;illniiic;m[I)' vary. Similar trends may
<br />~ icentit~d b th; bulk ~ow characteristics for event
<br />I :l..1d 3.;t.: Table::!.
<br />.-\>th:qr.l.da1u;olu(jonincreases.theele~ntsize
<br />
<br />T..h}
<br />TI:...:::.....,;~:~"""m:l>cp<rc~lItajCofthodom.>ill;nunda<ed
<br />
<br />h.,..6uoo~'='
<br />
<br />Pe~..n"leof~lI;nundolCdfOfmc.h
<br />
<br />"
<br />,
<br />
<br />~1.I1
<br />n.~~
<br />96.97
<br />
<br />IS.~
<br />7].~~
<br />97.17
<br />
<br />b.:comes smaller and from II!<:sh I 10 mesh 6 theu is
<br />an increase in Ihe: magnilUde: of the: hydrograph.
<br />However, Ihere is no difference in time [0 peak. For
<br />mesh 7 [here is a consisleTll decrease in peak
<br />discharge:.
<br />
<br />oJ.:? ImwJmimll!xte/ll
<br />
<br />It is neces~ary to understand Ihe effect of mesh
<br />resolution on inundation extent predictions so thai
<br />accurate risk as~ssmenls may be made. In this analy-
<br />sis inundation extent is expressed as a percentage of
<br />[he domain inundated (event 2 used as example). In
<br />Table 3 the percent of the domain inundated to 2S. 10
<br />and 5 cm is shown, while in Fig. 4 !he inundation to
<br />::!5cmis5hown.
<br />If the results in Fig. 4 are studied aconsistenl lrend
<br />can be identified. As the resolution increases. the
<br />extent of inundation decreases, This is shown most
<br />dramatically for the area inundated 10 2Scm.
<br />However th~ same trend may be identified in Table
<br />3 for inundation depths of 10 and 5 em, especially at
<br />the beginning and the end of the simulalion. This is a
<br />direct consequence of the e/fecl of channel size; the
<br />lower the mesh resolution, the wider the chll.T\nel. This
<br />i!i similar to the problem discussed earlier with the
<br />los!i of infonnation resulling from mesh filtering and
<br />ma)' not be a direct effect of the effect of resolution on
<br />the solution of thee qual ion but is caused by the filter-
<br />ing of topography by the mesh.
<br />
<br />oJ.3. Rela/il'e sl!llsitiviry of spatial resolmion and
<br />calibration
<br />
<br />This section is diucted towards assessing the lnuIS-
<br />ferability of parameter values betwccn meshes repre-
<br />senlin! Ihe same reach, but of different resolutions.
<br />Calibration of models can often be a computalionally
<br />demanding procedure, and this openltion may be
<br />
<br />lS.S&
<br />73.40
<br />97.03
<br />
<br />1~,S9
<br />7S.61
<br />91.&4
<br />
<br />12.42
<br />74.01
<br />96.&2
<br />
<br />11.44
<br />76.07
<br />%.&2
<br />
<br />11.0&
<br />74.71
<br />96.10
<br />
<br />RJ H",d>' e' al. I Jo",",,1 of HI'Jro!og.. 216 (/999) 1!~-/J6
<br />
<br />'"
<br />
<br />25{
<br />~ 20 /..:.--.~.~,.:.:--.'::"'"
<br />I" --- .' .- -- -- -- -- -- .- '_ . -,_
<br />j lO l',~:,.:.::::~~_,~;~~
<br />j {~-,.
<br />
<br />o
<br />
<br />~,
<br />.....-1'
<br />-.-......,
<br />-......1
<br />:::.:=~:
<br />-.....,.
<br />
<br />o 10000 20000 30000 40000 50000 60000 70000
<br />
<br />T.....C_)
<br />
<br />fi..4. TbceIfCl:I"frcsolul;oaoni"un<l;uionulCnfID2S~mforcvcIll2.
<br />
<br />made more efficient if models can be calibraled on a
<br />low resolution. compulationally efficient model, and
<br />then parameter values transferred to a higher resolu-
<br />tion model. For Ibis invesligation, evcnI 2, the inter-
<br />mediate event where up to SQ<l, of the domain was
<br />flooded, was used as the input hydrograph. The simu-
<br />II1Iion of this event using mesh 4 from the afore-
<br />mentioned analysis was designated as the conlrol
<br />and further simulations conducted using a range of
<br />
<br />floodplain friction values (11 = 0.111, 0.083 and
<br />0.056, uspectively). The ehll.T\nel friction coefficient
<br />(/I :z 0.02S) was utained for each simulation as
<br />puvious work (Hardy, 1997) has demonstrated that
<br />floodplain friction has a far greater effect on model
<br />results compared with channel friction. If the relation-
<br />ship between me spatial resolution, Ihe friction cali-
<br />brationcoefficienl and the model efficiency is studied,
<br />Fig. S, a definite pauem emerges. The graph may be
<br />
<br />D.I.BO'IE1.20
<br />.. o.eo.1.20
<br />_ 0.040.0.ao
<br />_ 0.00-0.00
<br />_ -o,4(I.o.QO
<br />_ -0.80--0.40
<br />_8el.OW-ll.lll
<br />
<br />
<br />~
<br />.-
<br />So
<br />l
<br />-.Q, ~
<br />""..
<br />'1.", "'
<br />1'~....
<br />
<br />M',~, "The rclationibip be<ween friction. .pali;a.l resolulion and model efficiency.
<br />
|