Nick, 

I am finally getting to use Exp on the water, and was just looking for something like this. In the Oakland Estuary, a windplot shows a major shift cycle of +.- 20-30 degrees about every 5 minutes, and on top of that a minor shift that could be +/- 10 degree ever 30 seconds. I was planning to log and analyze off line as a time series. 

If you return to the code, to manage 0/360 transition, May I request the ability to solve for :
baseline shift,  the first two dominant components, and to overlay that waveform on the actual measured TWD/TWS.

It would be perhaps very useful. That would of course then allow the mathematical projection of the next expected shift from the observed pattern...  and scalars of "Time to next shift" & "Next shift TWD/TWA/TWS/AWA/AWS"  and confidence/fit of pattern.