Statistical Decision Theory Techniques for the Revision of Mean Flood Flow Regression Estimates

Two statistical estimating procedures are presented for using regression information along with direct observations to obtain estimates of the expected value of peak flood discharge rates exceeding a constant base. The first type of estimate represents the minimum mean squared error linear combination of regression and direct estimates, whereas the second is a Bayesian estimate based on an objective prior distribution associated with the regression model. A comparison of combined estimates to regression and direct estimates used alone indicates that a significant reduction in the mean squared error is obtained by using combined estimates. A comparison of the two methods for obtaining a combined estimate indicates that in many cases they both give essentially the same result.