Identification of a multiple input, single output, linear, time invariant model for hydrological forecasting

Abstract

The technique of ridge regression 2 2 Hoerl, 1964. is reviewed and applied in the identification of the transformations in a multiple input, single output, linear flow routing model assumed to relate four inflow hydrographs to a discharge hydrograph at Hankou on the Changjiang (Yangtze River) in China. It is shown that this method of identification can be interpreted as an ordinary least squares estimation where an uncorrelated white noise component has been added to the input series.