Laserfiche WebLink
<br />The flood-frequency curves generally fit most of the observed data; <br />however, because of the unusually high floods that occurred during 1962-70, <br />the observed data plotted erratically in the upper parts of the curves. <br />The probability of exceedance of these higher flows is unknown even when <br />adjusted historically (Dalrymple, 1960). <br /> <br />REGIONALIZATION OF FLOOD-PEAK CHARACTERISTICS <br /> <br />Multiple-regression techniques are useful in regionalizing stream- <br />flow characteristics because discharge for a given recurrence interval <br />can be related to the physical and (or) climatic characteristics of the <br />basins. The regionalization procedure averages the chance variations <br />from sampling while maintaining the variation due to the basin character- <br />istics. Multiple-regression analysis allows definition of predictive <br />equations in the form <br /> <br />log Q = log C + a log X, + b log X2 + clog X3; <br /> <br />or in the alternate form <br /> <br />Q C X,aX2bX3c <br /> <br />where <br /> <br />Q = the discharge of selected frequency (dependent variable); <br /> <br />X" X2, X3 = the basin characteristics (independent variable); <br /> <br />C the regression constant; ~nd <br /> <br />a, b, c = regression coefficients. <br /> <br />In the regression process, variables that are the least significant <br />are automatically deleted from the equation. The accuracy of the estab- <br />lished equation is measured by the standard error of estimate, which rep- <br />resents the degree to which flood-peak variation is explained. <br /> <br />-41- <br />