Acoustic Doppler sensors used for flow measurements at energetic tidal sites present an inherent “Doppler noise” in the measured signal, varying with hardware configuration and flow conditions. At scales comparable to the sensors’ sampling frequencies, the corresponding perturbations notably contaminate the signal, and cannot be corrected in the time series.
At such scales, dynamic phenomena are of particular interest in the process of increasing reliability and effectiveness of tidal turbines, and are mostly addressed in terms of statistics. In the case of inflow speed variations, the bias due to Doppler noise should be taken into account, and can be assessed via manufacturer specifications.
Here, a method is presented that enables a direct estimation of the Doppler noise strength from the measured signal itself. Inspired from polynomial least square regression, it is based on a spectral analysis of the measured signal respect to turbulence theory, under the hypothesis of a white Doppler noise contamination. The subsequent limitations are discussed and illustrated by practical cases.
The values found are generally higher than suggested by manufacturers, but still in the same order of magnitude. The use of the highest sampling frequency available is recommended.