Jeff and I looked at thermal time constant of the calibration time dependent factors.
Soon after each lock, the GDS pipeline starts calculating some estimated cal params, including the spring frequency and DARM pole.

I looked at two-hour trends for the last six locks, and found the spring frequency settles in about 45 minutes, going from ~7 Hz to ~0 Hz.
The DARM pole was more noisy overall, going from 420 Hz to 412 Hz in about 40 minutes.

Code is in /ligo/home/craig.cahillane/Git/IFO/DARM/scripts/grab_time_dependent_factors.py and grabs the data using nds2, make sure to kinit.