A note on changes in channel geometry at tributary junctions

Evidence is presented to justify the assumption that channel width varies stochastically about an approximately constant mean width within a stream link of an alluvial channel. Adjustments of channel width occur predominantly at junctions, but Miller's (1958) simple model describing such changes is misleading, and an alternative is proposed whose implication is that channel width (and probably other channel dimensions) is a power function of link magnitude. It is suggested that an alternative to conventional hydraulic geometry is provided by the combination of a step function describing changes at junctions and a stochastic model of variation within individual links.