I have two time series of wind data from nearby stations, and I want to find out how well the records correlate, as I'm interested in the confidence with which I can make inferences about the wind conditions at sites between these stations based on the geographically sparse data available. Wind speed is easy enough to establish correlation for but the time series of direction data is proving challenging. Obviously I can't use bearings in degrees so I really have two questions here:
1/ How would I go about converting the wind direction into a value on a non-repeating scale? I'm reluctant to convert speed and direction into vectors and then correlating those, as that gives me very limited information about the correlation in direction which, for my application, is more important than speed.
2/ Having got the wind direction into a suitable format, what would the most appropriate function be for calculating correlation coefficients? (xcorr?)
I realise this is not strictly a 'Matlab' query, but I thought I'd ask here, as I'm going to have to implement this in Matlab and I'm sure somebody out there must have had to deal with wind data in Matlab before.