Secondary circulation in river confluences results in a spatial and temporal variation of fluid motion and a relatively high level of morphodynamic change. Acoustic Doppler current profiler (aDcp) vessel-mounted flow measurements are now commonly used to quantify such circulation in shallow water fluvial environments. It is well established that such quantification using vessel-mounted aDcps requires repeated survey of the same cross-section. However, less attention has been given to how to process these data. Most aDcp data processing techniques make the assumption of homogeneity between the measured radial components of velocity. As acoustic beams diverge with distance from the aDcp probe, the volume of the flow that must be assumed to be homogeneous between the beams increases. In the presence of secondary circulation cells, and where there are strong rates of shear in the flow, the homogeneity assumption may not apply, especially deeper in the water column and close to the bed. To reduce dependence on this assumption, we apply a newly-established method to aDcp data obtained for two medium-sized (~60-80 m wide) gravel-bed river confluences and compare the results with those from more conventional data processing approaches. The comparsion confirms that in the presence of strong shear our method produces different results to more conventional approaches. In the absence of a third set of fully independent data, we cannot demonstrate conclusively which method is best, but our method involves less averaging and so in the presence of strong shear is likely to be more reliable. We conclude that it is wise to apply both our method and more conventional methods to identify where data analysis might be impacted upon by strong shear and where inferences of secondary circulation may need to be made more cautiously.