Here is my contribution to the range worsening during the last long lock, following up on Lisa's and Sheila's entries.

Basically, my conclusion is that high wind affects the angular stability of the interferometer, the noise gets very non stationary (as visible from the summary page spectrogram and as it was observed in the past). 

I loaded all 4 hours of lock data, computed the band-limited RMS of DARM in the 100-200 hz region and correlate it with angular signals (both local sensors and WFS/QPD). For the first couples of hours the noise is lower and more stationary (still, there is some fluctuation that is correlated with angular motion). The first attached plot shows the trend of the BLRMS during the whole lock, together with the reconstruction using the WFS and QPD signals. The second and third plots are zooms into a quiet and a not so quiet time periods. 

The final attachment is the usual ranking of the signals that contribute most to the fit. AS_B_45_YAW is the clear winner, both I and Q.

P.S. The second attachment is broken and I can't remove it, ignore it.