Using velocimeter signal to noise ratio as a surrogate measure of suspended mud concentration

摘要

This study examines the relationship between suspended sediment concentration (SSC) and the signal to noise ratio (SNR) recorded by a 6MHz Nortek Vector velocimeter in a laboratory water tank using four different synthetic and natural mud mixtures and different combinations of the user-set Vector parameters transmit power level and velocity range. For concentrations less than 1500mg/l (1.5g/l), a region of linearity between the logarithm of concentration and time-average SNR was found for all sediment types and transmitter power level settings. Within this concentration range, the experimental data was used to develop calibrated equations of the form, log(SSC)=c1SNR+c2; R~2 values for all calibrated equations were greater than 0.98, suggesting that properly calibrated relations can yield accurate time-averaged SSC measurements using Vector measured SNR. An analysis of the general calibration equation indicated that the predicted SSC values are sensitive to changes in the coefficient values for c1 and c2. Even small (10%) deviations in coefficient values resulted in 20%-65% changes in the predicted SSC. Variation in c1 and c2 values among all four mud mixtures were significant enough that the calibrated equations could not be used interchangeably. This was true even among three samples that had similar particle-size distributions. Translation of raw 32Hz SNR data to 32Hz SSC time series produced excessively large variation in the SSC time series. Several smoothing and filtering schemes were examined to reduce the magnitude of these fluctuations to more reasonable levels. Of the methods tested, a two-sided moving average functioned best at removing fine-scale variation while retaining larger-scale trends. A 96-point (3Hz) averaging window brought 98.6% of the Vector estimated SSC time series values to within ±10% of the time-average physical samples. Impacts of turbulent kinetic energy and sampling volume size on instrument recorded SNR were also empirically examined.