image quality document:

SITCOMTN-017 (3).pdf

Brightest stars near S celestial pole:

At S pole:

Bright stars dec < -20 

C.D.S.  -  SIMBAD4 rel 1.8  -  2022.07.07CEST05:39:32

vmag<2 & dec < -20
------------------

Number of objects : 26

# |            identifier             |typ|    coord1 (ICRS,J2000/2000)     |  Mag U  |  Mag B  |  Mag V  |  Mag R  |  Mag I  |  spec. type   |#bib |#not
--|-----------------------------------|---|---------------------------------|---------|---------|---------|---------|---------|---------------|-----|----
1 |* eps CMa                          |*  |06 58 37.54876 -28 58 19.5102    |0.36     |1.29     |1.50     |1.59     |1.80     |B1.5II         | 473 |   0
2 |* del CMa                          |s*y|07 08 23.4840514 -26 23 35.518484|3.06     |2.52     |1.84     |1.33     |1.00     |F8Ia           | 331 |   0
3 |NAME CMa Dwarf Galaxy              |G  |07 12 36.0 -27 40 00             |     ~   |     ~   |-0.1     |     ~   |     ~   |~              | 197 |   1
4 |* alf PsA                          |** |22 57 39.04625 -29 37 20.0533    |1.31     |1.25     |1.16     |1.11     |1.09     |A4V            |1182 |   3
5 |* tet Sco                          |*  |17 37 19.12985 -42 59 52.1808    |     ~   |2.29     |1.85     |     ~   |     ~   |F1III          | 120 |   0
6 |* lam Sco                          |bC*|17 33 36.52012 -37 06 13.7648    |0.52     |1.49     |1.63     |     ~   |     ~   |B2IV+DA7.9     | 380 |   0
7 |* eps Sgr                          |PM*|18 24 10.31840 -34 23 04.6193    |     ~   |1.82     |1.81     |     ~   |     ~   |B9IVp_lB?      | 204 |   0
8 |* alf Sco                          |s*r|16 29 24.45970 -26 25 55.2094    |4.08     |2.75     |0.91     |-0.64    |-1.87    |M1.5Iab+B2Vn   | 726 |   0
9 |* gam02 Vel                        |WR*|08 09 31.95013 -47 20 11.7108    |0.64     |1.58     |1.83     |1.85     |2.00     |WC8+O7.5III-V  | 869 |   0
10|* alf Car                          |*  |06 23 57.10988 -52 41 44.3810    |-0.49    |-0.59    |-0.74    |-0.96    |-1.13    |A9II           | 436 |   0
11|* del Vel                          |EB*|08 44 42.22658 -54 42 31.7493    |     ~   |2.00     |1.95     |     ~   |     ~   |A2IV+A4V+F8V   | 191 |   0
12|* eps Car                          |** |08 22 30.83526 -59 30 34.1431    |3.32     |3.13     |1.86     |     ~   |     ~   |K3:III+B2:V    |  82 |   0
13|NAME LMC                           |G  |05 23 34.6 -69 45 22             |     ~   |     ~   |0.4      |     ~   |     ~   |~              |16075|   1
14|* bet Car                          |PM*|09 13 11.97746 -69 43 01.9473    |1.72     |1.69     |1.69     |1.63     |1.61     |A1III-         | 218 |   0
15|* gam Cru                          |PM*|12 31 09.95961 -57 06 47.5684    |5.01     |3.23     |1.64     |-0.02    |-1.44    |M3.5III        | 267 |   0
16|* bet Cru                          |bC*|12 47 43.26877 -59 41 19.5792    |0.03     |1.02     |1.25     |1.38     |1.64     |B1IV           | 333 |   0
17|* bet Cen                          |bC*|14 03 49.40535 -60 22 22.9266    |     ~   |     ~   |0.58     |     ~   |     ~   |B1III          | 351 |   0
18|* alf Cen A                        |SB*|14 39 36.49400 -60 50 02.3737    |0.96     |0.72     |0.01     |     ~   |     ~   |G2V            |1206 |   0
19|* alf Cen                          |** |14 39 29.71993 -60 49 55.9990    |     ~   |0.4      |-0.1     |     ~   |     ~   |G2V+K1V        | 890 |   0
20|* alf Cen B                        |PM*|14 39 35.06311 -60 50 15.0992    |2.89     |2.21     |1.33     |     ~   |     ~   |K1V            | 963 |   1
21|* alf TrA                          |*  |16 48 39.89508 -69 01 39.7626    |     ~   |3.33     |1.88     |     ~   |     ~   |K2III_Ba1      | 234 |   0
22|* alf Gru                          |PM*|22 08 13.98473 -46 57 39.5078    |1.11     |1.58     |1.71     |1.75     |1.83     |B6V            | 295 |   0
23|* alf Pav                          |SB*|20 25 38.85705 -56 44 06.3230    |1.03     |1.791    |1.918    |     ~   |     ~   |B2IV           | 282 |   0
24|* alf01 Cru                        |*  |12 26 35.896 -63 05 56.73        |     ~   |1.10     |1.28     |     ~   |     ~   |B0.5IV         | 141 |   0
25|* alf02 Cru                        |*  |12 26 36.442 -63 05 58.28        |     ~   |1.41     |1.58     |     ~   |     ~   |B1V            |  67 |   0
26|* alf Eri                          |Be*|01 37 42.84548 -57 14 12.3101    |-0.36    |0.30     |0.46     |0.49     |0.60     |B6Vpe          | 457 |   0
================================================================================

Aux tel image quality budget elements, instrumentation, and methods

Feb 17 2022

Thinking about temperature gradients within the dome. Imagine a temperature discontinuity of dT, at an angle theta to beam propagation. What combination of dT and theta produces a ray deflection of 0.1 arcsec? 

n1sin(theta1) = n2 sin(theta2). We want to solve for case where theta1-theta2=0.1 arcsec = 5E-7 rad. So theta2=theta1+5e-7 and so 

sin(theta1)/sin(theta1+5e-7) = n2/n1. To get an initial sense, take 45 deg incidence where theta1=pi/4. that means n1/n2 = 5e-7. What dT is that, around zero C? That implies that n1/n2=0.999999500000375

https://emtoolbox.nist.gov/wavelength/ciddor.asp

dn/dT is 1.08E-6 per degree C. (update Feb 22 2022- Elana says at Rubin elevation dn/dT is more like 7e-7 so plots are updated to this value)

This all means that at 45 degree incidence, 0.1 arcsec of ray deflection happens for deltaT of 0.5 degrees C. Same thing happens even if it's not a step change in temperature. 

Angle-of-arrival dependence for a 1 degree step change in temperature: 

Cut at dome seeing limit of 0.08 arcsec:

Purple area is OK, green area is not. 

MATLAB code for making these plots: airwedge.m

What about acoustic determination of path integral of temperature via time delays? Using Robert Lupton's idea of single source. How does sound speed depend on temperature? 

Also: 

Slope of c(T) is 0.6 m/s per degree and intercept at T=0C is 331 m/s. So fractional change in travel time per degree is 1.8E-3 per degree. 
Assume 10m path length, travel time is 10/330 = 30 msec. To detect a 0.1C change we need go from 330 m/s to 330.06 m/s so change in travel time is 5 microseconds. That corresponds to an acoustic bandwidth of 200 kHz and a corresponding digitization rate. 

Attenuation at 200 kHz is 1000 dB for 100m path length. Intensity (energy) changes by 6dB for each factor of two in distance. If we have 10m vs 100m path length that is a factor of 10 in distance which is 3.3 factors of two or 3.3*6=20dB less so still 1000-6 = 994 dB. 

If we drop down to 40 kHz then per-pulse resolution is 5x lower but attenuation is far far less, more like 94 dB for 10m. So it's a trade between temporal bandwidth and SNR. At 10 kHz it's audible, can't go there. 

AIRFLOW

There is a tension between large airflow rates in the dome to reduce thermal seeing but that increases image motion due to wind loading. Here is a table from mechanical estimates of wind induced image motion

Neill et al (SPIE Wind Induced Image Degradation (Jitter) of the LSST Telescope, see IQ team references page) did FEA analysis of wind shake. A weird phase-cancellation of image motion happens in their FEA model, so they did the analysis two ways, one of which omits this cancellation. Results: 

So depending on whether one allows for magic cancellation or not, wind speeds of up to 2-3 m/s inside the dome are tolerable. Note quadratic dependence on wind speed. 

Crude fits: best-case FWHM= (0.0035*(m/s)^2 + 0.0021*(m/s)) arcsec

                 worst-case FWHM = (0.0131(m/s)^2 - 0.0026 (m/s)) arcsec

How does this compare to CFD estimates of in-dome wind speeds? See Document-28556 to see that expected external median wind speed is 5 m/s and enclosure suppresses this by about a factor of two. So median interior wind speed of 2.5m/s breaks wind shake allocation if no structural cancellations occur. from Doc-28556: (definition of wind speed here is external to dome, differs from above plot for wind shake which is internal wind speed. )

Action- what is histogram of wind velocity at the site? Extracted one year of data from Rubin EFD Coronagraf interface, Feb 19 2021 to Feb 19, 2022. 

Here is a night-only data set over a longer time period, taken from LTS-54 document: 

So, not that different. 

LTS-53 implies that the wind limit is at least 20 m/s or 44 mph. Not sure that was really what happened though. 

CTIO, SOAR, Gemini all have wind limit of 20 m/s. (https://noirlab.edu/science/observing-noirlab/observing-ctio/cerro-tololo/bad-weather-protocol-at-ctio). 

We expect external winds of above 7 m/s about 40% of the time. With factor-of-two nominal reduction of interior vs exterior wind speed, we therefore expect internal speeds above 3.5m/s about 40% of the time. 

Take external wind speed, divide by two to get internal winds, then compute best and worst case wind shake FWHM cumulative PDF:

Conclusion- Best case is that wind buffeting of telescope and optics will meet its requirement 50% of the time. Thermally induced seeing also has wind speed dependence, that is complicated. 

How well would adaptive scheduler work? If wind speed and direction are random, then it won't work at all. Did cross correlation of wind speed, after subtraction of the mean. Data points are 1 min apart but sliding averages over 2 min. 
1440 minutes is one day, so the oscillations in this autocorrelation plot correspond to diurnal cycle (24 hours) of wind speed

x-axis is lag in minutes, so wind speed is highly correlated on hours-long time scales. That's good!

Downloaded a year's worth of wind speed and direction data from Chronagraf at summit. For some reason, wind direction seems to have an interval of 1 hour and 12 minutes, while wind speed is measured each minute. Also time stamp convention is unclear. It would be nice to conform to ISO standards! Finally, picking a large date range gives daily spacing for wind direction. Need help with that! So there is work to do here. 

Also there are missing measurements in the data structure, for example, for wind direction, third row here is missing a measurement. 

2/2/2022 1:00:00.000000000 PM,304.30555555555554

2/2/2022 2:12:00.000000000 PM,224.0625

2/2/2022 3:24:00.000000000 PM,

2/2/2022 4:36:00.000000000 PM,112.29032258064517

2/2/2022 5:48:00.000000000 PM,96.43055555555556

Issues: 

• timestamps don't conform to ISO standards, they have AM and PM suffixes, and no explicit time zone. What is EFD convention? 
• What is the definition of "Local" on the Environmental dashboard?? Seems to be UT-5 for me, and I certainly assumed it meant Chilean time. If it converts to local for the user, we should totally eliminate that option and make all timestamps UT, period.
• Another issue with time stamps: With UTC selected on Chronograf dashboard, downloaded file for Wind Direction. Got this: 

  1/30/2022 5:00:00.000000000 AM,312.7416666666667

  1/30/2022 7:00:00.000000000 AM,313.8333333333333

  1/30/2022 9:00:00.000000000 AM,287.35

  But using the cursor on the display, the data point of 313.833333 occurs at UT 1/30/2022 12:00. So there is a 7 hour discrepancy. 
• there are missing Wind Direction data entries, that are simply blank. See for example wind direction entry for 2/2/2022 3:24:00.000000000 PM. 
• Dashboard window panel title is not correct if a different averaging time than 10 min is chosen in query, at least for Wind Direction and perhaps others.
• Download file name is not correct if a different averaging time is chosen in query, at least for Wind Direction and perhaps others.
• Attempting to download long time series for Wind Direction gives only a decimated version of the data file, with coarser sampling as longer date span is selected. (this could be user error) 
• We need faster sampling of wind direction, each minute would be good. 

Nevertheless, for Feb 2 2022 to Feb 19 2022 here is what we have:

Polar histogram of wind direction. ) is N, azimuth increases to E so this is rotated and flipped. 330 is NNW, 0=360 is N, 90 is E:

What azimuths do we expect to point towards? 

Is wind speed correlated with direction? Not really: 

Computed North and East components of Wind vector, and looked at differences dN and dE on one minute timescales. 

standard deviation in dN is 0.3 m/s, and std deviation in dE is 0.4 m/s. Mean of windspeed in this interval is 4.3 m/s and standard deviation is 2.5 m/s. So windspeed scatter on 1 minute timescale is 0.1 times as much as overall variation, so highly correlated. 
But note the data are 2 minute sliding boxcar averages. What does plot of windspeed look like? Data points are 1 min intervals:

Direction plot: 

Plot wind direction vs. folded MJD, which is 0 at midnight UT, 0.5 at noon UT. Chile local time is either 3 or 4 hours earlier than UT, depending on the season 

Looks like there is a thermal transient that starts mid-day? Bump happens at 0.7 UT which is around 1 pm local time in Chile.  

Diurnal cycle in wind speed: 6 consecutive days of wind speed vs. day fraction. Can we use daytime wind to predict nighttime wind speed profile? 

What would it take to cool Aux Tel interior? 

Gemini has 126kW cooling power with dome area of 2850 m^2. Rubin is building about 240 kW with enclosure area of 2494 m^2. Aux tel dome is 9.3m diameter. 

total spherical shell surface area is 4piR^2 so half-dome is 2piR^2 = 2*pi*(9.3/2)^2 = 135 m^2. Scaling to Gemini and Rubin implies a daytime cooling capacity of 240*(135/2500) = 13 kW = 44,000 BTU/hr

single high-capacity McMaster portable AC unit has about half this capacity, at cost of $4000. https://www.mcmaster.com/7669K7/ 

This honker would do it, for $10K, with capacity to spare: https://www.sylvane.com/kwikool-60000-btu-5-ton-portable-air-conditioners.html 

Follow-on tasks

Nov 25, 2022

Information about Aux tel temperature evolution, now augmented with acoustic temperature near top end, is at
https://summit-lsp.lsst.codes/chronograf/sources/1/dashboards/91?refresh=10s&lower=now%28%29%20-%207d 

EFD at USDF is way behind.