for my own edification, I took a look at End X during a wind time on Oct 17. 
wind storm started at about 1 am Pacific time and went for about 8 hours. 

(note to self - repeat this for Oct 14 - see Jim Warner alog
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=30547)

I can see several things:
1) The wind direction indicator seems to work as long as the wind is blowing hard enough. This is great. 
Conversion is : add (180-37) = 143 degs. 
This makes 90 (towards -Y) into 233 (from the SW, up along the Y arm)
see plots 1&2 for wind history during monitor period.


2) building tilt is still related to wind. From 1 mHz to 20-30mHz, the relation between floor tilt and wind-speed^2 is quite clear - good coherence, time series look similar. 
(plots 3&4)
TF is odd though,(plot 5) it doesn't appear to follow Kramers-Kronig. 


3) The wind speed indicator gives decent data at frequencies below 20-30 mHz. This limit is set by the discrete reporting levels in some ADC.
(see plot 6) There may also be some low pass filters, but those are the biggest limit on data quality.