Secondary circulation cells in river channel confluences: measurement artefacts or coherent flow structures?

This paper is concerned with the representation of secondary circulation in river channel confluences. Recent research has emphasized the complex three-dimensional flow fields that exist where two river channels join. Field and laboratory measurements have been developed to describe time-averaged flow fields in terms of primary and secondary circulation, and to interpret these in terms of key generating processes. Central to this research is the need to understand the effect that flow structures have upon both mixing processes and confluence geomorphology, notably the development of scour-holes within the junction zone. One of the common problems faced by this research is the dependence of observed secondary flow structures upon the rotation plane for which they are determined. Different researchers have used different rotation planes, such that intercomparison of results from different field sites is difficult. Problems also arise when only two-dimensional measurements (e.g. downstream and cross-stream) are available, and vertical velocities need to be inferred from analysis of secondary circulation patterns. If different analytical methods produce different patterns, so different inferences could be reached. This paper uses a numerical model to show: (i) that different analytical methods do result in very different estimates of the strength of secondary circulation; (ii) that there are problems in inferring vertical velocities from secondary circulation cells identified using these methods in confluences, most notably as a result of the effects of planform acceleration and deceleration; and (iii) that field and laboratory measurements suffer from being unable to measure the three-dimensional flow field instantaneously, and hence allow understanding of the evolution of flow structures through time. A three-dimensional solution of the Navier–Stokes equations for open channel flow, combined with a free surface approximation and an unsteady turbulence model, allows representation of the three-dimensional time-averaged flow field, and some aspects of the unsteady evolution of these flow structures. Hence, the researcher can be freed from the dependence of results obtained upon the analytical method chosen. This emphasizes the downstream transport of mass in the form of a helix, which will be central in zones of flow convergence or divergence, rather than the more traditional recognition of closed helical circulation cells. Copyright © 2000 John Wiley & Sons, Ltd.