Multiple Peak Count Analysis for Increased Spectral Resolution, A Non-linear Filtering Method

In applications for acoustic profiling, an analysis of current versus depth is key. By analyzing the Doppler frequency variation in the backscattering signal, it is possible to calculate the mean current versus depth. However, due to turbulence in the water creating simultaneous currents and layers, the current direction could change rapidly, making the data non-stationary and difficult to analyze. The transmitted pulse is very short, and the volume where the current should be estimated is typically small. Also, the water volume can contain multiple Doppler responses due to turbulence and other phenomenon. W hen using this innovative non-linear filtering method, Multiple Peak Count Analysis, these effects can become visible, and the method shows superior performance as compared to classical methods. Classical 3-dimensional spectral plots of the data, does not use a-priori information. In this case, there is a particular interested in finding multiple peaks in the data, a-priori information. A non-linear pre-filtering method has thus been developed where the peaks are extracted in the spectra. A short-time FFT has been used to find the spectral content, where a small time data block is processed using a much larger time analysis block size. The paper describes this new method and compares the result with classical analysis. Advantages of this new method, and suggestions where the method can perform better are also discussed. This approach can also be used in other applications for sound and vibration, despite this method’s main current use being under water.