Frequently Asked Questions About 2dF
Questions
What is the reconfiguration time?
Which way does the slit-numbering go?
How do I convert to/from FITS files?
How do I work out which spectrum is from which object?
What is the minimum spacing (in arcseconds on the sky) of 2dF fibres?
Is it possible to look at the data after each reduction step in 2dFDR?
How does the emission-line throughput calibration work in 2dFDR?
What does flat-fielding mean in 2dFDR?
How can one omit 'sky' fibres which contain objects in 2dFDR?
How does 2dFDR handle flexure?
Answers
What is the reconfiguration time?
For a full field reconfiguration (i.e. to remove the old configuration and replace it with a new one) for relatively uniformly sampled fields (i.e. not to strongly clustered) the reconfiguration time is of the order 60-70mins.
Which way does the slit-numbering go?Both ways! Seriously - engineers and programmers decided on
different
conventions.
In drcontrol S=1
refers to the bottom fibre, as
displayed. Ditto for the headers where the 1..400
arrays go from
bottom to top. In general low pivot numbers (i.e. those near
1) will be
displayed near
the bottom of the screen by drcontrol
while high
pivot numbers (i.e.
those near 400) will be displayed near the top, though there is not a
one
to one mapping. The actual mapping is stored in the 2dF system file spec_fibres.txt.
How do I convert to/from
FITS files?
Use the programs ndf2fits and fits2ndf - these programs correctly propagate the FIBRES info (unlike FIGARO rdfits/wdfits or IRAF rfits/wfits) AND the VARIANCE array (which is present, and correct, in reduced data). An example usage might be:
ndf2fits proexts my_sdf_file my_new_fits_file.fits
Note that the .sdf file does not want the .sdf extension but that the .fits file does need the .fits extension. The proexts keyword propagates a set of fits extensions (images arrays and binary tables) for fibre information and sky spectrum etc.How do I work out which spectrum is from which object?
There are a number of ways to do this, the best method depends on what you are trying to do. The information is all stored in a binary table extension of the reduced (and raw) fits or sdf files, and also in the .sds file from which the field is configured. The methods most often used to retrieve the data are:
- In configure, uses the File->List menu option to save an ASCII text file of which fibre is attached to which object
- Use the 2dfinfo task packaged with drcontrol and pipe the table output to a text file. e.g. 2dfinfo data.fits fibres > fibres.txt
- Use a package that can read binary tables to access the information, e.g.. IRAF or IDL. Examples of how to do this can be found in the AAOmega cookbook.
What is the minimum
spacing (in arcseconds
on the sky) of 2dF fibres?
i.e. how close can observational targets be?
Actually it depends on the geometry at which the fibres come in due to the funny shape of the magnetic buttons. The absolute minimum is 30 arcsec (2mm), but typically it's 30-40 arcsec depending on location in the field and target distribution.
Is it possible to look at the data after each reduction step in drcontrol?
One can turn on various automatic plot options - the fitted
tramlines,
the fit to the scattered light background and the
profile fits during extraction (with the FIT method), the
throughput
map and the subtracted sky. If you want to look at the raw
extracted
spectra (i.e. before calibration/sky-subtraction) use the "Command->Plot"
menu option to look at the *ex.sdf files after
reduction.
How does the emission-line throughput calibration work in drcontrol?
All sky lines are used. Sky lines pixels are identified by plotting the wavelength derivative of the flux - those with large derivatives are identified as sky emission line pixels. Obviously if your wavelength range contains no sky lines (e.g. at high-dispersion in the blue) this option should not be used! In that case twilight flats and offset sky frames will be needed. Note, it is only possible to do twilight flats for a maximum of 4 fields a night, two at the start and two at the end, since the fields must be pre-configured in order to take twilight flats, and the flat is not relevant once fibres have been moved, even if the field is reconfigured at a later date.
What does flat-fielding mean in drcontrol?
There are three-meanings. The first is the dispersed white-light fibre spectra used to fit the tramlines. This is what is usually referred to as the `FLAT'. The second meaning is `pixel-to-pixel CCD flat field' otherwise known as a 'longslit flat'. This was not available in 2dF, but is to be used for AAOmega. The third meaning is `fibre flat-field' by which we mean dividing the extracted object spectra by extracted, normalized, white light spectra (this is usually the same as that used for the tramlines). This will remove any funny wavelength variations in individual fibres, such as was found in May 1998 due to some of the prisms being cracked giving rise to fringing. At some level, the work of the 'longslit flat field' can be done with the 'fibre flat', especially given the uniformity of modern CCDs. The best results will require the 'longslit flat' though.
How can one omit 'sky' fibres which contain objects in drcontrol?
Normally this should not be necessary as drcontrol takes a median sky, and clips outliers. If you really must, create a file called `skyfibres.dat' in the working directory, listing the numbers (one per line) of the fibres you wish to use for sky. drcontrol will then use this file, in preference to the headers, when you reduce (or re-reduce) the data.
How does 2dFDR handle flexure?
AAOmega is a bench mounted spectrograph in a stable thermal environment. In principle there will be no flexure, certainly not on the same scale as with the 2dF system. During commissioning, when the spectrograph had not yet reached thermal equilibrium, a small shift of the spectra on the CCD was seen at the 0.5 pixel level. To retain the possibility of correcting for this the "shift and rotate" option, which allows the tramline map to be tweaked to the data, has been retained form 2dfdr. If shifts are seen to be a problem for a data set, additional arc frames would be needed during an observation, rather than the single arc taken at the start of each new field. We are watching this carefully as AAOmega matures.
Can one save drcontrol parameter settings?
No, but you can create .idx file with these setting preset.
Can one combine frames BEFORE sky subtraction in 2dFDR?
No. If you want to experiment with this I suggest you turn sky-subtraction off completely and do your own processing on the final individual spectra. If you get better results that drcontrol, let us know. Most previous efforts at this have failed, and AAOmega has been seen to give 1% sky subtraction, if you care at this level then ask your support astronomer about "Nod and Shuffle" observations.
Rob Sharp (rgs@aao.gov.au)