Variability and uncertainty in reach bankfull hydraulic geometry

Abstract

Observations of the hydraulic geometry of river channels are widely used in fluvial geomorphological research and river management. Variations in hydraulic geometry observed at sites through a region or through time are often assumed to be true variation, but measurement uncertainty will also contribute to these variations. Few studies have quantified the effect of error. This paper evaluates uncertainty in bankfull hydraulic geometry for 114 river sites in southeast Australia. These sites represent a broad range of river types, including lowland and upland rivers in temperate and semi-arid climates. Results will be applicable to hydraulic geometry investigations worldwide when consistent channel survey techniques are used. Uncertainty is quantified relative to observed variation in hydraulic geometry between sites. For bankfull discharge, width, depth, velocity and slope, respectively 4%, 3%, 7%, 23% and 42% of the site-to-site variation can be attributed to uncertainty. Uncertainty in the variability of bankfull metrics for each of the 114 sites (e.g. standard deviation in bankfull width) is higher, contributing between 40% and 50% of the observed between site variation. Mean width and depth are more sensitive to sample errors associated with cross-section surveys. Bankfull discharge, velocity and slope are more sensitive to errors in parameters for the gradually-varied flow model required to model flow conditions at bankfull discharge. We develop models to predict the error resulting from using hydraulic models to predict bankfull flow, which may be used to estimate the appropriate number and spacing of cross-section surveys required for accurate modelling. These models relate error in bankfull metrics to channel attributes and the number and spacing of cross-sections. To minimise errors, sites must be longer in low gradient channels and have more cross-sections in more variable channels.