Estimation of Channel Depth during Floods by Canonical Correlation Analysis

Abstract

Estimation of the scour depth due to design flood in an alluvial river is often done with a regime relationship that gives the equilibrium flow depth for peak discharge. For short‐duration peaks, this depth may never occur; and since riverine structures are usually very costly, the difference may be very significant in monetary terms. Thus the need to incorporate the variation of predictor variables with time in the analysis of scouring process exists; and since no satisfactory method is available to take this into account, a statistical analysis technique called canonical correlation model is introduced to solve the problem. Canonical correlation analysis is functionally similar to regression analysis with additional procedures to handle multicollinearity and serial and autocorrelations. The technique has been used in other areas of science with proven usefulness. The model application is illustrated with an example.