| Abstract: | This paper constructs a model of channel geometry composed of three subsections: a steady-state submodel, a gradedstate submodel, and a stochastic error submodel. With the aid of this representation of the morphology of channels, the at-a-point changes in geometry can be reproduced by a simple recursive equation of autoregressive, moving-average form which is derived from methods used in the statistical analysis of time series. A set of height loss data for three Japanese rivers derived from Yatsu's (1955) paper is used to determine the effect of bed material changes on adaptations within the graded-state submodel of the channel. The results of analyzing the autocorrelation function, spectrum, and adaptive parameter shifts within this model can be used to infer that significant changes in the amplitude of height change variation, and a shift to higher frequency oscillations of bed forms are associated with the shift in bed material conditions. |
