Correlation parameter

A customer asked:

I am checking the ASCII files and according to the headers the .c* contains correlation.

C:\data\smy00701-2.c1]
1 Burst counter
2 Ensemble counter
3 Correlation Cell 1 (Beam1) (%)
4 Correlation Cell 2 (Beam1) (%)
.
.
n Correlation Cell n (Beam1) (%)

Correlation of what?
Does this value show the quality of the data?

Which should be the order of the value of correlation for reliable data?

The parameter called "correlation" is an excellent indicator of data quality. The AquaPro High Resolution Profiler operates by finding the phase shift from two pulses that are transmitted. This correlation is a measure of the coherence or "similarity" of the echoes from the two pulses. If the signals do not emanate from the same scattering particle, the correlation will be low.

To edit out spikes in the velocity data, simply set a threshold on the value of the correlation parameter when post processing the data. The appropriate threshold value depends on the circumstances but a value of correlation > 50% can be a good starting point.

If you want more information on the AquaPro HR I recommend to read the Principles of Operation, Planning Guidelines and Velocity range sections in the Help file in the AquaPro HR software. There you will find information that will help you to configure the instrument properly in order to get good data quality from your deployments.

Best regards,
Sven Nylund

More information on correlation (from Atle):

The correlation in the HR means exactly the same thing as it does in the Velocimeters/ADVs. The sources of decorrelation that have been mentioned in the literature are:

• Residence time (do we sample the same particles)
• Reorientation of the particles (turbulence)
  

The mechanisms that can lead to spectral broadening, but not necessarily show up in the form of reduced correlation, are:

• Beam divergence (changes in the angle of the projected velocity)
• Shear (velocity variation inside the sampling volume)

In addition, there is interference between the two acoustic pulses, which is always a problem - but in most cases limited to bands in the velocity profile.

Best regards,

Sven Nylund