Laurence Datrier, Sheila Dwyer
We looked at O2 data for H1 to get an estimate of down time due to wind. We used the maximum wind speed of the three max. 1min and 30min trends for the H1 EX, EY, CS wind channels, for all of O2.
The first histograms and trend line show duty cycle vs wind speed percentiles, for wind speed vs lock status and previous wind speed (max. of previous 1 or 30 min) vs lock status. Looking at the previous wind speed shows more down time due to high winds. Percentiles and wind speed values are (in mph):
percentile | 5 | 10 | 20 | 30 | 40 | 50 | 60 | 70 | 80 | 90 | 95 | 99 |
1min trend wind speed [mph] | 3 | 4 | 5 | 6 | 7 | 9 | 11 | 13 | 16 | 21 | 25 | 33 |
30min trend wind speed [mph] | 5 | 6 | 7 | 8 | 10 | 12 | 14 | 17 | 21 | 26 | 31 | 41 |
The following histograms are the normalised distributions for wind speed at:
Compared to the normalised distribution at all times (blue). A significant tail appears in the distribution for wind speed at time of lock loss for winds >~35mph.
Interesting. Is this consistent with the histogram of 8 years of wind speed data in this log:
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=12996
and the O2 locking vs wind speed in Figure 16 of this paper?
https://dcc.ligo.org/LIGO-P1800038
It would also be useful to calculate the correlation of BNS range with wind speed. We may be able to maintain lock at higher wind speeds than in the past, but does the range suffer?
I used the wind wind speed histogram from 8 years of data at LHO:
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=12996
to calculate a wind probability density function. I then convolved it with a fit (approximation) of the O2 duty cycle vs wind speed from figure 16 in:
https://dcc.ligo.org/LIGO-P1800038
If interferometer locking was completely independent of wind speed, then only a 1.6% improvement in duty cycle would result.
However the BNS & BBH range *may* decrease at higher wind speeds.
Based on the table that Laurence posted, these wind speed percentiles are fairly consistent with Margarita's 8 year histograms in 12996. Margarita finds that the hourly maximum of the wind speed is about 31.3 mph 4.8% of the time, Laurence finds that the 30 minute maximum exceeds 31 mph 5% of the time.
Note that Laurence is plotting duty cycle for all the times with wind speeds above a certain quantile, while Krishna plots duty cycle vs wind speed in P1800038 , so they aren't directly comparable. The best comparison to Figure 16 would be to Laurence's 5th attachment, which is quite similar, the duty cycle is between 60-70% for wind speeds less than ~32 mph in P1800038, and less than the 99th percentile for the minute trend in Laurence's plot, which is 33mph.
I believe that the duty cycle vs wind speed plot in Figure 16 of P1800038 is based on minute trends of the wind speed, although I'm not completely certain of that. If it is based on minute trends, it makes most sense to convolve the wind speed vs duty cycle plot from P1800038 with a wind speed distribution based on the first line in Laurence's table, based on minute trends. (Or, to use wind speed vs duty cycle based on hour maximum instead).
The detailed O2 time accounting for H1 states that the interferometer was only down for wind 0.3% of the time:
https://ldas-jobs.ligo.caltech.edu/~detchar/summary/1164556817-1187733618/time_accounting/lho/
As Daniel point outs, wind might be responsible for some fraction of the time spent trying to re-lock.
Curiously the O2 accounting for L1 states that wind caused 2.7% downtime:
https://ldas-jobs.ligo.caltech.edu/~detchar/summary/1164556817-1187733618/time_accounting/llo/
I think it would be useful to run this again for O3 so far. A quick look indicates that the detector is rather more sensitive to wind now than at the end of O2. I went poking though the DetChar pages looking for impact of high wind - and one can see that for speeds of ~10 meters/sec one can often see dips in the range, and when the speed passes that LHO often looses lock. I counted up 15 days in the first 2 months where there seemed to be a clear correlation-by-eye between (wind vs. range) or (wind vs. operation). I expect there are many reasons why the detector is more sensitive, and hopefully we can get back to the sort of robustness that was achieved at the end of O2, but I think that, were there a wind break in place now, things would be easier to run.
I've posted an alog for the same analysis for O3, along with a look at the BNS range as a function of wind speed.