PDA

View Full Version : Correlation coefficient for vectorial timeseries



sargenta
15th February 2013, 07:04 AM
Dear all,

I would like to find a correlation coefficient for two timeseries of vector variables (currents). I know I have the option of correlating the modulus of the currents (not the direction, since it's a circular variable), or correlating each vector component separately, but I think in both cases there is loss of information. Therefore, I want to correlate the vectors as a whole. I have tried Hanson's (1992) correlation algorithm, but its implementation gives me a 0 correlation for two identical timeseries (strictly northward stationary currents). Are there any other options? I would really appreciate your ideas on this issue.

Thanks

42d3e78f26a4b20d412==