Laserfiche WebLink
<br />3-04 <br /> <br />an a:wroximate adjustment is satisfactory. Where the regu1s.tic_ <br />is comp1.ex, as in the case .of a large n~erof upstreamreser,.. ' <br />voirs, it may be advisable to adjust the da.ta to a uniform con- <br />dition with specific reservoirs and diversion facilitiesopers.ting. <br />For design. purposes, s. frequency ,curve of runoff . under "non- <br />project" conditionsths.t ,is expected to preTa11 during the lii'e-. <br />time of the proposed project, if the project is not constructed, <br />is required. A frequency curve bued on ~ specified uniform <br />condition can be converted to one for nonproject conditions usmg <br />relationships developed. by routing "balanced" floods of specified <br />frequency (i.e., floods hs.ving runo:t:'f forvanous durs.tions and in <br />various portions of the drs.ins.ge basin of equal exceedence fre,. <br />quency). Techniques for doing this are ()utside the scope of this. <br />pap er. <br /> <br />3-05. <br /> <br />PLOTTING FORMIJLl\ <br /> <br />!!. . The reasoning behind the selection of lll1 exact form~a <br />for plotting the frequency of observed flood. events is extrelliely .- <br />cOlllPlex. Approximate plotting positions can be. obtained by . <br />reasoning that each item in. a set, of, say 10, represents 10 :Per,. <br />cent. of the parent-popula.tion. events and should be plotted in the <br />middJ.e of its group,that is at 5, 15,25 percent, etc.,:tor suc'; <br />cessive eventsin.the order of magnitude.. 'rheformuladerived . <br />from this line of reasoning ~ been nseq. in the ,past, and is <br />generaJ.ly satisfactory,.. consi<iE!ring the overall reliablli ty of <br />the results. HOIrever, !I1Ore accurate p1.otting positioDs, haTe been <br />derived theoreticalJ.y and, since their usa is very silll;Ple, it is <br />considered ad'V8l:1tageous to use them, Plotting positions recOlll- ... <br />mended are shown on e:xhib~ t. 37, . and are based on the Pi'emise that <br />used re at ,.in a great numberOt randOlllssm;ple$; <br />they will prove to be too 10lr in hall' of the cues aJl. 0 <br />in the other half, collql8.red with the theoreticalJ.y true values <br />that ca.nnot be determined because of random -yariations in da.ta. <br />. -.' , , .- <br />They are, therefore, called median plotting pOsitiOns. Thereare <br />other systems of deriving plotting positions that yield good. <br />resuJ.ts, but of the fo1'll1Ul.as generally.used., use of the median <br />p1.otting position will most nearly duplicate results obt&ineci by <br />!ll1a1.ytical methods of frequenCY !ll1a1.ysis that do not require <br />p1.otting positions. <br /> <br /> <br />-\12.- <br />