Laserfiche WebLink
<br />(1972) suggested that the positions of the bursts <br />could be reversed if that is determined to be <br />hydrologically the more critical situation. However, <br />their view on the arrangement of bursts was not <br />shared by Pilgrim and Cordery (1975) who contended <br />that the sequence of intensity blocks arranged <br />arbitrarily to give a maximum value of peak runoff in <br />several design patterns would give the joint <br />occurrence of a rainfall intensity of low probability <br />and a pattern of low probability. Consequently, the <br />frequency of exceedence of the resulting flood <br />estimate would then be lower than that of the storm <br />causing it. <br /> <br />A method proposed by Kiefer and Chu (1957), <br />and later adopted by Bandyopadhyay (1972) andPreul <br />and Papadakis (1973), to determine the skewness ( r <br />value) of the temporal pattern was entirely based on <br />antecedent rainfall records of arbitrarily specified <br />durations of 15, 30, 60 minutes, etc., up to tc' the <br />time of concentration. The r value obtained fot each <br />specified duration was weighted in proportion to the <br />amount of antecedent rainfall preceding that duration <br /> <br />so that weighted average value of r was computed. <br />The r value so obtained should vary with the a, b, <br />and c values used in the rainfall <br />intensity-duration-frequency formula, Eq. 6 or 7, as <br />well as the te value found in the drainage area under <br />study. Therefore, by using this method, the r value <br />so obtained cannot be made independently of <br />frequency and tc' This result in a sense is <br />contradictory to their original assumption that the r <br />value is independent of frequency. <br /> <br />Another method of evaluating the r value <br />developed by Kiefer and Chu (1957) is to calculate <br />the mean locations of the peak within all specified <br />durations such as IS, 30, 60, and 120 minutes and <br />then to have the results weighted proportionally to <br />the time of duration. In this method, the average <br />value of r for all durations considered, despite being <br />adjusted through the weighting process, still could be <br />biased by inaccuracy in the computed mean locations <br />of the peak for such durations as small as IS minutes <br />which have only three 5.minute periods. <br /> <br />6 <br />