Window length for calculating Coherence

Hello,
I want to calculate coherence for each time window of my recordings. the time windows could be between 1 to 5 seconds (sampling rate 200 Hz).
Would you please let me know how I should find the optimal window length for calculating coherence? and generally how the window length could affect the calculated coherence value?

Best,
L

Just like for the evaluation of the PSD, if you increase the length of your estimator window length:

  • you increase the frequency resolution of your output (= more frequency bins, which can be good or bad depending on what you expect),
  • you decrease the number of averaged windows, and therefore obtain noisier results.

https://neuroimage.usc.edu/brainstorm/Tutorials/ArtifactsFilter#Evaluation_of_the_noise_level

I'm not aware of any solution for defining an "optimal length", it all depends of your data and your expectations.

@hossein27en @Raymundo.Cassani Any other suggestion?

thank you very much