FLOOD10102 - Laserfiche WebLink

Home Browse Search

bj b2 b. Y=aX X . . .X. I j 2 I (7) where: Y=flood characteristic (dependent variable), Xj, a=regression constant, X2.. . X.=basin and climatic characteristics (independent variables), I and bj, b2.. .b,=regression coefficients. I The application of this method to the 15 stations for which synthetic flood- frequency relations were developed is discussed in the following sections. Dependent Variables The dependent variables selected for regression analysis were a weighted 10-year (QjO l. 25-year (Q25l. 50-year (Q50), and 100-year (Qjoo) peak discharge for each of 15 study basins (table 2). Results of two different flood-frequency analyses have been discussed thus far: analysis of recorded annual flood series and analysis of synthetic annual flood series. Lichty and Liscum (1978) develop- ed a method of computing a weighted average of the observed and synthetic flood-frequency relations based on an analysis of variance; the weighted result by this method was determined to be an improved estimate of the flood characteristic. Because of the very large time-sampling error (short period-of-record) experienc- ed in this study, application of this variance-weighting method only was success ful for the 10-year and 25-year frequencies. The weighting factors used for the 50-year and 100-year frequencies were determined by judgment based on the 10- and 25-year factors and the results of previous studies (Thomas and Corley, 1977). The following are the weighting factors used: FZood aharaateI'istia Observed Synthetia QjO 0.60 0.40 Q25 .40 .60 Q50 .25 .75 Qj 00 .10 ,90 , The weighted flood-frequency relation was determined by this method for 12 basins as shown in table 2. For example, the weighted Q50 for station 07120600 is 0.25 times 2,040, plus 0.75 times 3,080, or 2,820 ft3/S. For station 07124700 ., 18