De-spiking parameters

From Atomix
Revision as of 20:09, 6 June 2024 by Marcus (talk | contribs)

The shear probe will collide with plankton and other particulates in the water, and it will occasionally hit jellyfish and even some fish that do not get out of its path. These collisions cause a very large and transient signal from the shear probe. These anomalies typically last for ~50ms, but can be longer in the case of jellyfish. They do not represent shear and bias high the variance of shear and the rate of dissipation. These anomalies should be replaced with data of constant value and the data so modified must be tracked because it is a quality control metric. If more than a few percent of the data used for a particular estimate of [math]\displaystyle{ \varepsilon }[/math] have been modified by the de-spiking algorithm than such estimates are suspect.

A widely used algorithm, De-spike the shear-probe data, identifies shear-probe signal anomalies by comparing the absolute shear against a smoothed version of the absolute shear. It requires a threshold and smoothing parameter and the number of points to be replaced around a spike. You may use other de-spiking algorithms but it important to keep track of the fraction of the data that has been modified.

<! –– * [thresh] Threshold value for the ratio of the instantaneous rectified signal to its smoothed version. A value of 8 is a good starting point for a VMP. ––> <! –– * [smooth] The cut-off frequency of the first-order Butterworth filter that is used to smooth the rectified input signal. The time scale of smoothing is approximately 1/(2*smooth). A value of 0.5 is a good starting point for a VMP. ––> <! –– * [Fs] Sampling rate (Hz). ––> <! –– * [N] Spike removal scale. A total of 1.5*N data points are removed. N/2 points are removed before a spike, and N points are removed after a spike. The replaced data is an average of the adjacent neighbouthood. Averaging uses ~Fs/(4*smooth) points from each side of a spike. ––>


return to Flow chart for shear probes