MONITORING ALLUVIAL SAND SUSPENSION BY EDDY-CORRELATION

Eddy correlation techniques are standard tools in micrometeorology and oceanography to measure momentum and contaminant transport across turbulent boundary layers. They can, in theory, be used to estimate the net vertical suspended sediment flux directly over different areas of an alluvial channel boundary, and thus disclose ongoing erosion/deposition patterns. The basic principles and main problems in applying the technique to alluvial suspension are first introduced. Results from a trial application of the method in a large sand bed river are then presented; the focus of the analysis is on the substantial (and surprising) contributions of multi-minute flow fluctuations to suspension work in the study environment. The data were collected in a 10 m deep channel of the Fraser River near Mission, British Columbia, Canada. Turbulent fluctuations of flow components streamwise and normal to the bed, along with the output of an optical suspended sediment sensor, were monitored over 7 h, 1 m above the bed. Flow velocities averaged 0-9 m s-1 and mean suspended sediment concentrations 500 mg l-1, at sensor level above 1-5 dm high dunes. Spectral analysis of the records reveals that approximately 30 per cent of the vertical suspended sand mixing across the sensor level (and roughly as much of the momentum exchange) was linked to gradual flow oscillations with periods between 1 and 13.6 min (underlying briefer, turbulent fluctuations). Extended periods of sediment-rich, slightly upward directed but slower mean flow alternated with periods of sediment-poor, slightly downward and faster mean flow; these slow fluctuations involved 10-20 cm s-1 changes in 5 min average flow speed, 2-4-degrees changes in vertical flow angle and 100 mg l-1 changes in mean sand concentration. To obtain accurate eddy-correlation estimates of the vertical suspension flux in the study conditions, hour-scale flow and turbidity records that include many of these multi-minute cycles appear to be necessary. The spectra of the Fraser River near-bed signals do not conspicuously differ in overall shape (in terms of low-frequency content and location of peak) from turbulent spectra encountered in some atmospheric boundary layers. Nonetheless, the long period fluctuations observed on the Fraser River may not be turbulent; rather they may reflect slowly evolving perturbations in the near-bed streamlines, caused by bedform translation or gradual fluctuation within the large-scale streamwise cells of the secondary flow.