Trends in Dome Seeing During the Night

a summer 2020 research project for James Findikyan
working remotely with John Hill at LBT Observatory, University of Arizona.

James' science teacher in New York is Joseph DiBello.

The Big Question

Does the average image quality of the Large Binocular Telescope vary systematically during the night?
(a real question that an astronomer asked recently)

The Narrower Question (to give a more practical size to the project)

Does the "dome seeing" at the Large Binocular Telescope vary systematically during the night, or during the year?
("Dome seeing" is the term used by astronomers to describe image blurring caused by turbulent mixing of warm/cold blobs of air inside the telescope enclosure.)

The Methodology to answer the question

Use archived telemetry measurements from the DIMM (Differential Image Motion Monitor) mounted on the LBT telescope to look for systematic trends during the night and/or during the year. The DIMM samples the turbulence from near the top end of the telescope on up into the atmosphere.

Section I - Introduction to Python (python3), the Linux environment (Fedora) and Wiki (Foswiki).

https://wiki.python.org/moin/BeginnersGuide (yes, "moin" is not a typo in this URL)

https://docs.python.org/3/tutorial/

https://github.com/LBTO/web-dms/blob/master/src/python/examples/demo_telemetry_api (account needed; this reference is for John)

https://dms.tucson.lbto.org/view-telemetry (needs VPN connected) "Make LBTO Telemetry Plots interactively"

https://maker.pro/linux/tutorial/ "Basic Linux Commands for Beginners"

https://wiki.lbto.org/System/TwentyMinuteTutorial (wiki account needed to edit)
Hint: It is better to use the "Edit Wiki Text" at the bottom rather than the plain WYSIWYG "Edit" button at the top.

https://wiki.lbto.org/Commissioning/WhatIsTheSeeing (some precursor work on seeing plots)

Section II - How to connect to LBTO computers (AnyConnect VPN, X2Go client, Zoom or Skype)

Connect to LBTO VPN using CISCO AnyConnect (see emailed instructions from Stephen and John).

SSH to specified linux computer shell.tucson.lbto.org (running Fedora with a Bash shell) using a secure SSH program with X-forwarding. From a Win10 PC, John uses Putty for the SSH connection, and X-Ming for the X display. John uses WinSCP for secure file transfer.

Talk to John via Zoom or email to report your progress and ask questions.

Take notes on the wiki, wiki.lbto.org, (using a sub-page of this page) as you go along. This is to document your work - you will often want to repeat a task as you learn more or improve plotting scripts. Create a new wiki topic (like JamesFexercise1), and then edit that wikiname into this page to have a link to it.

In this era of COVID-19 this is a completely remote research project, but to get a feel of additional things you might encounter if you were on-site review: "20200101 LBTO Site Safety Orientation - Industrial.pdf" found at https://wiki.lbto.org/Safety/SiteSafetyOrientation

JamesFnotes

Section III - Data Analysis Steps

Exercise 1 - Make a few nightly plots of the DIMM seeing measurements

Make time series plots of DIMM seeing measurements (corrected to zenith) for 10 different nights spread over the last year. What variations do you see from night to night?

This would use John's existing python scripts which invoke IDL code to make plots. John's scripts mostly live in /lbt/lbto/supportscripts/TelemetrySupport/

  • change to that directory by saying: cd /lbt/lbto/supportscripts/TelemetrySupport

  • Example for how to get help about the arguments (common for many linux commands):
    ./dimmplot_hdfextract.py -h
[jhill@shell TelemetrySupport]$ ./dimmplot_hdfextract.py -h
Example:  ./dimmplot_hdfextract.py  -b -12.0   (makes a plot of the last 12 hours)
general command line arguments for plotting
 -d YYYMMDD or --day YYYYMMDD date to plot (default today)
 -e TIME or --endhour TIME    end time in UT hours (default now)
 -b TIME or --backhour TIME   hours before end time (default 24)
 -x '[XMIN,XMAX]' or --xlimits '[XMIN,XMAX]' plot range (IDL style)
 -y '[YMIN,YMAX]' or --ylimits '[YMIN,YMAX]' plot range (IDL style)
 -s XX or --side XX           which mirror side to plot SX/DX
detailed arguments to control script behaviour
 -p or --plotonly             make plot but don't extract new telemetry
 -n or --nodisplay            extract new telemetry but don't plot
 -m or --mute                 don't display the plot with eog
 -r or --remove               remove temporary data files
 -o DIR or --outpath DIR      put plot files in this directory path
 -h or --help                 print this help message

  • Example for how to make a plot:
    ./dimmplot_hdfextract.py -d 20200202 -e 14 -b 13
    Note that "-d YYYMMDD" specifies the UT date, "-e HH.H" specifies the ending UT hour, "-b BB.B" specifies how many hours to look back from the end to make the plot. The sample output may include error messages if something went wrong - it may have be cloudy on the night you asked for, etc.
[jhill@shell TelemetrySupport]$ ./dimmplot_hdfextract.py -d 20200202 -e 14 -b 13
20200202  day
14.0  endhour
13.0  backhour
./dimmplot_hdfextract.py script version of 08-Jan-2016
time.struct_time(tm_year=2020, tm_mon=2, tm_mday=2, tm_hour=23, tm_min=59, tm_sec=59, tm_wday=6, tm_yday=33, tm_isdst=-1)  specified time as tuple
5087404799.0  end MJD sec for the end of this UT day
58881  end MJD day
5087368800.0  end MJD sec corrected for endhour  14.0
5087322000.0  start MJD sec using backhour  13.0
20200202  is the last day to be plotted. Covering MJD  58881  to  58881
20200202  is the day to be plotted for  58881
h5csv -d seeing_01 /lbt/data/telemetry/tcs/dimm/2020/02/02/20200202????.dimm.seeing.h5 | awk '{time = $1 / 1000000.0 ; if ( NR <= 1 || (time >= 5087322000.0 && time <= 5087368800.0) ) {print $0} }'  > /tmp/dimm.20200202jhill.tmp
Sending command to retrieve telemetry - no output until it is completed.

dimmplot_hdf, "/tmp/dimm.20200202jhill.tmp"

Sending commands to IDL to make the plot - no output until they are completed.
X-range:    5.0873244e+09   5.0873679e+09 MJD seconds
X-range:        58881.070       58881.573 MJD days
today:        58881
X-range:      0.069976852      0.57305556 Days in UT
X-range:        1.6794444       13.753333 Hours in UT
       3.1872300      0.70072700 raw seeing measurement (telem)
     0.500000          DIMM floor value (JMH)
       3.1477667      0.49093618 raw seeing less floor (JMH)
       61.087742       18.961342 zenith distance
       2.0683820       1.0573753 airmass (JMH)
       2.0352796      0.47477467 zenith seeing less floor (JMH)
       2.0607960      0.67765900 zenith_seeing (telem)
       2.0400000    0.0010000000 catalog magnitude (telem)
       13.164271       116.37470 measured flux (telem)
       2.4515080     0.085353545 apparent magnitude
Y-range:        0      3.00000 arcsec

Plot file is located at  /tmp/dimm.20200202jhill.jpg
Sending plotfile to eog for display - exit eog to finish.

Issue: The eog (eye of gnome) image display program used by John's Python scripts isn't found on the machine shell.tucson.lbto.org. That's OK, because eog is too slow over remote connections.

A work-around is to invoke John's home-grown display tool using the following alias command at the beginning of your session. To make it happen every time you login, include the specified alias command at the end of your .bash_profile file.
alias eog="/home/jhill/bin/LBTdisp.py" Unfortunately, the alias doesn't seem to function from within John's telemetry scripts. That's because the python scripts call "/usr/bin/eog -n" rather than simply "eog -n". The replacement for eog (eye of gnome) on shell is eom (eye of mate) since shell uses the MATE window manager rather than the GNOME window manager.

To launch an LBTdisp window: /home/jhill/bin/LBTdisp.py &

Exercise 2 - Make a time series plot of the DIMM Seeing measurements for an entire month

Make a time series plot of DIMM seeing measurements (corrected to zenith) for an entire month. Do you see any night-to-night trends beyond what was seen in a single night? (e.g. May 2019, November 2019)

Exercise 3 - Make histogram plots of the DIMM seeing measurements for each month

Make histogram plots of the DIMM seeing measurements month by month during the year? (Jan 2019 - Feb 2020) Is there any trend during the different seasons?

Plot the results of the monthly histograms vs time (mean, mode, median) [This is the first item that requires a code change.] That is plot the mode, mean, and median of each month's histogram vs month. (To be more technically correct: mode of the histogram, mean of the data, median of the data - since floating point data doesn't have a mode until binned into a histogram) Do you get a different shape of histogram for different months/years?

Make yourself a local copy of the programs to edit:
cp /lbt/lbto/supportscripts/TelemetrySupport/dimmhist_hdfextract.py /home/jfindikyan/
cp /lbt/lbto/supportscripts/TelemetrySupport/dimmhist_hdf.pro /home/jfindikyan/

Common linux text editors: vi, emacs, gedit (only on Gnome), pluma (not common, but on MATE)

Emacs Reference Card:
https://www.gnu.org/software/emacs/refcards/pdf/refcard.pdf

Modify the IDL program to print mean median and mode:

To edit a file:
emacs -nw dimmhist_hdf.pro

Exercise 4 - Make histogram plots hour by hour during the night

Make histograms of the DIMM seeing measurements hour by hour during the night (e.g. a histogram of the DIMM data between 2 and 3 UT summed for a month).

James Findikyan ended his summer 2020 research here

With only so many hours in the summer, James Findikyan ended his research project here (Exercise 4). His final research report is attached below.

See CausesofDomeSeeing for a more extensive paper which includes James' summer 2021 research.

John's conclusions from James' work are as follows:
  • There is no discernable trend in the monthly median seeing values as evaluated by histograms of zenith-corrected DIMM measurements. The month-to-month variations swamp any seasonal trend (based on 2018 and 2019 data).
  • There is a weak trend of seeing improving during the night based on histograms of zenith-corrected DIMM measurements. The median seeing measured by the DIMM improves about 15% between the first hour of the night and the last hour of the night. These data were not corrected for the time at which the dome actually opened, and were binned by UT hour (not hour since sunset).


Exercise 5 - Make time series plots during the night averaged over a month

Make time series averaged plots of the DIMM seeing measurements plotted by UT hour during the night (averaging a month of nights together). Is there any obvious trend during the night? Does the trend depend on particular nights being included in the average? [Probably needs some boxcar smoothing of the data.]

Make time series averaged plots of the DIMM seeing measurements plotted by hour since sunset during the night (averaging a month of nights together). Is there any obvious trend during the night?

Make time series averaged plots of the DIMM seeing measurements plotted by hour since the enclosure was opened during the night (averaging a month of nights together). Is there any obvious trend during the night?

Exercise 6 - Plot DIMM seeing measurments versus temperature differences

Make a plot of DIMM seeing measurements plotted against the difference between telescope steel temperature and ambient air temperature.

Make a plot of DIMM seeing measurements plotted against the wind velocity.

Make a plot of DIMM seeing measurements plotted against the wind direction (telescope azimuth vs wind azimuth).

Section IV - Summary and Final Report

Write up a final report about what was learned.

Prepare a Powerpoint presentation to present to your class and to astronomers.

Follow-up

At this point, I expect that we have already far exceeded 100 hours of work, but there could be other things to look at to confirm or deny what we learned from the DIMM seeing measurements.....

Other Sources of Comparison Seeing Data Off-axis guide star FWHM Off-axis Shack-Hartmann wavefront sensor FWHM

Causes and Explanation of Dome Seeing

Dome Seeing Mitigation Strategies

References about Seeing and Telescope Image Quality

- N. J. Woolf 1982 "High Resolution Imaging from the Ground" in Annual Reviews of Astronomy & Astrophysics

- J. M. Hill et al 2016 "Telescope Performance at the LBT" SPIE 9906

- P. Bely and G. Lelievre 1987 "Seeing control in domes and telescopes" conference proceedings "Identification, optimization and protection of optical telescope sites"

- P. Bely 1987 "Weather and seeing on Mauna Kea" PASP 99, 560

- Pedersen, Rigaut, Sarazin 1988 "Seeing Measurements with a Differential Image Motion Monitor" (ESO Messenger #53)

- Vernin and Munoz-Tunon, 1995 "Measuring Astronomical Seeing: The DA/IAC DIMM" PASP 107, 265.

- R. Racine, D. Salmon, D. Cowley, J. Sovka 1991, "Mirror, Dome and Natural Seeing at CFHT". PASP

-- JohnHill - 04 May 2020

I Attachment Action Size Date Who Comment
Analysis of Trends in Large Binocular Telescope Dome Seeing.pdfpdf Analysis of Trends in Large Binocular Telescope Dome Seeing.pdf manage 1 MB 01 Mar 2021 - 19:14 JohnHill A more polished version of James Findikyan report which was submitted to WESEF in Fall 2020
JamesF Histograms.xlsxxlsx JamesF Histograms.xlsx manage 32 K 15 Sep 2020 - 21:03 JohnHill Spreadsheet of DIMM histogram results by James Findikyan
Research Paper p2 - Google Docs.pdfpdf Research Paper p2 - Google Docs.pdf manage 760 K 15 Sep 2020 - 20:15 JohnHill Final Research Report by James Findikyan
make-september10-11.cshcsh make-september10-11.csh manage 6 K 18 Sep 2020 - 20:21 JohnHill sample script from James Findikyan project
Topic revision: r24 - 17 Nov 2021, JohnHill
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback