ON THE NECESSITY OF APPLYING A ROTATION TO INSTANTANEOUS VELOCITY-MEASUREMENTS IN RIVER FLOWS

In studies on river channel flow turbulence, it is often the case that the measured mean vertical velocity is different from zero, indicatin g that the frame of reference of the current meter is not parallel to the flow streamline. This situation affects the estimate of Reynolds s hear stress in the streamwise and vertical planes and consequently the analysis of the flow turbulent structure. One way to solve this probl em is to correct data by applying a rotation and this is reviewed in t he first part of the paper. However, in fluvial geomorphology, the stu died flow is often complex and streamlines may exhibit significant cha nges from one point of measurement to the other. In this context, appl ying a rotation complicates the situation more than it simplifies it. The second part of this paper examines the question of velocity data c orrection in complex flows using a held example of the turbulent bound ary layer over a very rough gravel bed and a laboratory example taken from flow at a river channel confluence. In both cases, velocity vecto rs are spatially variable. In the first case, errors in the Reynolds s hear stress estimates are relatively low (ranging from -13 to 7 per ce nt/deg) while in the second case, they are much larger (-200 to 164 pe r cent/deg). The significance of these errors on the interpretation of turbulence statistics in river channel flows is discussed. We propose that corrections should be applied in all clear cases of sensor misal ignment and when the frame of reference changes spatially and temporal ly. However, no corrections should be used where different how velocit y vector orientations, not sensor misalignment, are responsible for th e mean vertical velocity differing from zero.