Quicklook Data Reduction
For observers to assess the quality of their data in (near-)real-time
and make better informed decisions, the IRTF implemented
"quicklook" data reduction software. During the observations, the
software reads the FITS file headers and determines if sufficient data
are available to run a scripted version of Spextool. The extracted
spectra are then sent to DV so the observer can inspect them. All this
Irtfquicklook runs 24/7, without observer interaction. The program
will recognize when an observer logs into the Bigdog user interface,
and when an observer logs out. It will automatically change the raw
data directory when needed. If Irtfquicklook is not running for some
reason (after a computer reboot), the operator will need to
On the Bigdog VNC, open an xterm.
On the xterm, type ssh quicklook@cybele
Enter the project password
On the Cybele prompt, type irtfquicklook -m live
Do not kill the xterm. It will display messages with the data
reduction progress. See screenshot:
When Irtfquicklook is Running
When the Irtfquicklook software is running, its progress can be
monitored as follows:
Messages on the xterm will show which FITS files are found, and which
are being reduced.
Reduced spectra are automatically loaded into buffers F
and G of DV. To see them, make sure the DV display is in
"quicklook" mode. See screenshot.
One can load and plot any reduced spectrum using the "Open" button in
DV. In the file selection window, enter the path of the reduced data
(/home/quicklook/Data/2019AXXX/YYMMDD_reduced/), select the
appropriate buffer, and click on the "ReadQuickLook" button on the
FITS files of the reduced spectra will be stored in
/home/quicklook/Data/2019AXXX/YYMMDD_reduced/. The observer may copy
them to their home computer (scp) within 2 weeks of the observations,
after which the files will be deleted..
A copy of the xterm messages for this observing session is saved in
Buffer F will contain the most recently reduced A-B nodding
Buffer G will contain the latest average over a contiguous
collection of nodding pairs, sharing the same position on the sky,
instrument settings, itime, and coadds. It will be updated when a new
nod pair becomes available.
If the observations are ongoing, but no new spectra appear in DV
buffers F and G, check the messages in the xterm in which
Irtfquicklook was started:
There could a backlog. Each reduction takes about 30 seconds.
There may be insufficient calibration files (flats and arcs) for the
reduction to proceed. The software will select flats and arcs taken in
the same night. When those are not taken yet, the most recently
reduced calibration files will be selected from the library at
The automated reduction does not work for every observing mode
yet. See the section on "Known Issues and Limitations" below.
There may be observing situations that are not yet covered by
Irtfquicklook. For example, as a result of an "abort" after clicking
"Go" in the Bigdog user interface, after an observatory systems crash,
or after network interupts. We are checking the reduction logs every
morning and will address issues accordingly.
Known Issues and Limitations
The development of Irtfquicklook is ongoing. We are aware of the
following problems and limitations:
The software works only for SpeX (Bigdog), not iSHELL, and only for
spectroscopy, not imaging.
Only observations taken with Beam Pattern AB will be reduced. This
excludes slit scans.
The extracted spectra are not divided over a standard star
spectrum. The primary goal is to determine the S/N of the
observations. Standard star division will be implemented later.
If one quits Irtfquicklook (by Control-C, for example), and then
restarts it with the same raw data input directory, the reduction will
start from scratch, ignoring previous reductions.
If the user logs out of the observing session on Bigdog, and then logs
back in on the same date with the same account, the permissions have
changed for the previously taken raw FITS files and they are no longer
readable by the quicklook account. In this case the observer could
start irtfquicklook using their user account after logging in to
cybele (see above). The reduced data will then be saved in their home
Occasionally, the wavelength calibration for the LXD mode is very
wrong (by a few tenths to several micron).
The reduction of spectra obtained in off-slit nodding mode currently
gives incorrect results.
For non-sidereal rates faster than ~0.10 arcsec/sec the software
currently has problems recognizing if nodding is done on- or off-slit.
Until a better algorithm is implemented, on-slit nodding is typically
assumed. If not, a warning is given in the xterm that the extraction
apertures may be incorrect.
04/26/19: first night-time use with SpeX
04/27-05/30/19: multiple upgrades and bug-fixes
Contact Adwin Boogert for further information and assistance.