2SLAQ Quick Look Observing Guide

Isaac Roseboom, February 2005

This quick look guide gives the very basic things you need to do when observing for the 2SLAQ survey. Links are provided to the more detailed notes below where relevant, it is highly recommended that all observers read or at least look at the entire observing guide.

1. Observing Plan

It will be the responsibility of one of the 2SLAQ observers to produce a observing plan for each night. A useful tool for doing so can be found at ~ observer/obsplan.tcl A list of fields to be observed during the run is usually produced beforehand. The location of this file will be made known to the observer at the start of the run. In producing an observing plan several things need to be taken into account.

Also take into account how many more nights you have in the run to complete the fields. There is no point in starting new fields on the last night of the run as uncompleted fields are unlikely to be reobserved in subsequent runs.

2. Configuring Fields

Fields should already be configured prior to the run, and hence are not usually the responsibility of the 2SLAQ observer. However if for some reason you do need to configure a field follow the instructions given in Section 2.1 of the observing guide.

3. Data Reduction

The primary role of the 2SLAQ observer is to reduce the data as it is acquired from the instrument. Before the data can be reduced you must copy it from the instrument into your visitor area. It is prefered that you use the disks on aat_lxa as this machine has plenty of diskspace, and is the fastest for data reduction. The following recipe should be used for acquiring and reducing the data. A more detailed description of the data reduciton process can be found in Section 6 of the observing guide.

A list of known data reduction problems is given in Section 6.3, please refer to these first if you have any problems.

4. Redshifting

Now that you have the combined frames for a given field you will need to perform the redshifting so as to gauge the completeness of the field. Seperate redshifting programs exist for the LRG and QSO data.

For the LRGs you will need to use Zcode which is usually located in /epping/rdc/2dflrg/Zcode/. This program will only run on the sun machines at present so you must ssh onto aatssx or aatssz to perform the redshifting. To start the Zcode type
/epping/rdc/2dflrg/Zcode/runz_apr04 filename
Where apr04 is replaced with the current version, and filename is the name of the combined reduced LRG field, minus the .sdf suffix. If you are unfamilar with the operation of the Zcode please see Section 8 of the observing guide.

For the QSOs you will need to use the AUTOZ and 2DFEMLINES code located in /home/aatssz/scroom/2df/code/. The code uses the starlink libraries so you will have to ssh onto aatssx or aatssz, and type starlink, and then type
setenv TEMPLATE_AREA /home/aatssz/scroom/2df/code/
The AUTOZ code must be run first via the command
/home/aatssz/scroom/2df/code/autoz filename
Where filename is the name of the combined reduced QSO field, minus the .sdf suffix. Once this is completed run the 2DFEMLINES code via
/home/aatssz/scroom/2df/code/2dfemlines filenameq Where filename is the name of the combined reduced QSO field, minus the .sdf suffix, with a q added on the end. The QSO redshifts produced by AUTOZ can now be checked and corrected where necessary. If you are unfamiliar with the QSO redshifting code read Section 9 of the observing guide.

5. Reconfiguring Fields

Once the redshifting of both LRGs and QSOs in a given field are completed the field can be reconfigured for subsequent observation. Before reconfiguring a field look at the S/N for the field. It is not worth reconfiguring fields which have exceptionally low S/N (<3) . Galaxies which have a quality flag of 4 or greater in the LRGs, and certain redshifts/types in the QSOs (no ? marks in 2dfemlines) are set to the lowest priority and hence replaced by new targets. To revise the configuration follow the instructions given in Section 2.2 of the observing guide. Make sure that fields are always configured on the same plate as they were previously observed on, as a configuration which works on one plate may not work on the other.

6. What to do at the end of a run

A few things tasks must be completed at the end of the observing run.
  • It would also be helpful (but not necessary) to write a list of amendments which need to be made to this guide to keep it up to date with changes at the AAT. This can be saved on the data DVDs.

    • 2SLAQ Observing Guide


    Terry Bridges, Scott Croom, Isaac Roseboom, Russell Cannon

    version 2, Feb 2005

    1.  Observational Setup

    2.  Configuring fields

    The configured fields set up prior to the run are usually found in:

    /home/aatssz/scroom/sdss2df/observing_0404/ (e.g. for April 2004).

    If fields are not already available (someone will usually have configured at least the first few nights worth of fields), then the following procedure should be used to obtain firstly a .fld file for input to the 2dF configure software, and then a .sds file that will be given to 2dF itself to position the fibres. The steps are outlined below. We also describe the process by which we tweak the configuration after the first or subsequent nights.

    2.1 Original configurations

    2.2 Revising configurations

    In order to increase the redshift/ID yield from each field we now examine the redshifts from the first night's data, and any objects with good quality redshifts can be de-allocated, so that their fibres can be used for other targets. The process by which this is done is outlined below:

    The process should be repeated for each night's observation, so for a field observed over 3 nights there may be a third version (v3) as well. If the data quality is not great, then the observers may choose not to remove any objects, however, there are usually at least a few targets (particularly NELGs in the QSO sample) which can be found even in rather poor data.

    3.  Observing Plan:

    The observing plan for the run will be in somewhere like:

    /home/aatssz/scroom/sdss2df/observing_0404/obsplan_0404

    The general aim is to get a total of 4 hours integration on each field (in good conditions), split over two nights (2 hours per night). After the first night a preliminary pass of the redshifting should be done and any objects that already have sufficient signal for a good ID can be de-allocated in the config for the next night (see above). If the conditions are worse than median seeing (which is ~1.5"), or cloudy, it may be necessary to extend the exposure times to get sufficient S/N to obtain good IDs, or observe on a 3rd or 4th night. Note that it is generally not a good idea to observe fields > 1 hour away from the tested HA.

    4.  Skyballing (obsolete as of March 2004)

    This section is now obsolete, as we have sky positions supplied from the SDSS which do not need to be checked (March 2004).

    `Skyballing' the field means visually examining each sky position on the Digital Sky Survey to check the fibre is really on sky and not an object. This uses the .dss file generated during the field configuration process, and which are kept along with the .fld and .sds files in the above location.

    You can use the Sparc in the control room (e.g. aatssf) to read the sky survey cdroms using

    5.  File Naming Conventions

    5.1  Reduced Data:

    5.1.1 Each Night

    (a-z)nn(g,q)_yymmdd_x.sdf, where:

    5.1.2 Final, Combined Data

    (a-z)nn(g,q)_yymmfi_x.sdf, where ``fi'' indicates final (note that care is needed if 3 or more nights are eventually combined, since intermediate combinations may have the same name: best rename any such so that only the real final set is called ``fi'', but please keep to six-character names to identify datasets for LRG Zcode compatability).

    5.2  Redshifted Data:

    (a-z)nn(g,q)_yymmdd_2z.rz for LRGs or ..._1.checkedresults for QSOs (automatic)

    6.  Data Reduction

    6.1  Setting up

    6.2  2dfdr Reduction

    See the 2dF User Manual for more details about 2dfdr: http://www.aao.gov.au/2df/manual.html

    Normally the data for one field will consist of 7 frames: one flat field, 2 arcs, and 4 target frames. A new 2dfdr session must be started for each field and CCD, in a separate sub-directory. Also 2dfdr must be run in the directory with the data.

    1. Create a subdirectory for the field and CCD about to be reduced. Use the standard nomenclature (a-z)nn(g,q)_yymmdd_x (see above):

    2. CD to this directory, then copy the raw data using the following script:

    3. Start 2dfdr by typing grsdrcontrol &; (note that this runs the same code as drcontrol but makes the standard 2dFGRS parameters and settings the defaults, thereby saving time and avoiding errors).
    4. Click on the Setup button in the Auto Reduction panel.
    5. In the pop-up display, check the path to the data and the filename prefix are correct and click OK.
    6. In the Commands menu select Find Fibres...
    7. In the pop-up display select the fibre flat field image (Class: MFFFF) and click OK.
    8. In the diagnostic plots window the fibre flat field image will be displayed with red tramlines overlaid and numbered. Check that the tramlines are aligned with the fibres in the central columns of the image. Dead fibres are marked with an asterix ( Note: dead fibres may still have light entering the fibre ). Zoom ('z' to zoom in, 'o' to zoom out, 'p' to pan) in and use the 'a' and 'd' keys to add and delete tramlines until exactly 200 are aligned and present. Press the Quit button in the plot window. If fibres fall off the chip and 200 tramlines cannot be specified, a manual fix is required (see trouble shooting).
    9. The following actions are only needed if you used drcontrol instead of grsdrcontrol:

    10. In the Data menu use the arrows to select the fibre flat field image (Class: MFFFF), and then click the Reduce button.
    11. The tramlines will be plotted again. Press the Quit button in the plot window (you have already checked this step using Find fibres). The tramlines will be plotted a second time. Zoom in and check that the tramlines have been rotated to fit the fibre spectra over the full spectral range, and that the tramlines do not cross to adjacent fibres. If the tramlines are mismatched, press the Quit button in the plot window, then the Abort button in the Auto Reduction panel and see the trouble shooting section ).
    12. Diagnostic plots for background (scattered light) subtraction and profile fitting will be displayed. Check that these look sensible. If the error "not enough dead fibres" is displayed, turn off Subtract Scattered Light in the Extract menu and restart the reduction of the flat field.
    13. Look at the reduced flat and check if it is reasonable (i.e looks grey, most values in range 0.95-1.05).
    14. Select the arc (Class: MFARC) in the Data menu and click the Reduce button.

    15. Plot the reduced arc image and check that the arclines are straight.

    16. Turn off Plot Tram Map in the Extract menu.
    17. Check the second reduced arc image
    18. It may also be worth reducing the first data frame manually, checking that the tramline map and reduced frame look sensible. It is also recommended that you restart 2dfdr at this point and pipe the console output into a file by simply adding > xnng_yymmfi_2.comb (where xnn is the field, yymm the date) to the end of the grsdrcontrol command.
    19. Click the Start button in the Auto Reduction panel to reduce the remaining object frames.
    20. Take note of the "Sky subtraction accuracy" displayed in the DREXEC? window. Ideally this should be less than 2%. Large values indicate serious problems in the extraction. When running in auto-mode, the sky subtraction accuracy will be missed on the display, but it can be recovered by looking in the ``history'' file for each frame.
    21. In the Commands menu select Combine Reduced Runs...
    22. In the pop-up window select the object frames in the sequence in which they were taken (double click or ADD button). It is important to combine the original frames from each night in the order they were taken to ensure that the flux weighting and cosmic ray rejection are done properly. Set the Combined File name to (e.g. a01g_yymmdd_1.sdf and click the OK button. (If you used drcontrol instead of grsdrcontrol, you will need to turn Flux Weighting for Frame ON.). Check thats the weights are sensible, also check that "Pixels rejected as CRs" is about 0.5% per frame as they scroll up on the screen.
    23. From the File menu select Open, and from the pop-up window select the combined reduced file. Alternatively, use the Plot from the Commands menu.
    24. Plot the combined reduced file. Use the 'x' key to plot the spectrum under the cursor. Use the Next and Prev buttons to plot the adjacent spectrum. If the spectra are noisy at the red and blue ends, the fibre flat field may be introducing noise (see trouble shooting). Check that broad spectral features have not been introduced to all spectra at the same wavelength by the background subtraction (see trouble shooting).

    25. Select Exit from the File menu. A new 2dfdr session must be started for each field and CCD.

    6.3. 2dfdr Troubleshooting

    7.  Files to Keep

    Here's the list of files that you should keep after 2dfdr reductions. The final combined frames are obvious, but other files are necessary in case re-reduction is required. Files to keep:

    8.  LRG Redshifting Code

    This uses a version of the Zcode developed by Will Sutherland and others for the 2dFGRS, as modified for the LRGs (mainly by RDC). See the "2dF Redshift Code - User Guide" (http://www.aao.gov.au/local/www/tjb/lrg/zguide.html) for a fuller description of how to run this.

    For users familiar with the GRS Zcode, the main differences are the virtual elimination of emission-line redshifts (only about 20% of the LRGs have emission lines, and often [O II] 3727A is the only line present: this is used as a check on some absn line zs), and a modified set of quality parameters in the range 1-4. These changes were a consequence of the different wavelength coverage of the LRG spectra (2200A centred at 6150A, rather than 4400A centred at 5800A). There have been numerous changes to the output data lists and formats, and a somewhat different set of template spectra is used for the LRGs. The program has been modified to give several user modes, the most useful being a `quick' semi-automatic mode where the observer only needs to check borderline redshifts. More details on some of these are given below.

    All spectra have to be assessed consistently and quickly by the end of each night, so that fibres can be re-assigned for the following night. Therefore instead of inspecting and assessing every spectrum, a `quick' mode of the Zcode is used which asks the observer to check only redshifts where the automatic z is poor or uncertain, typically 10-20%. The intention is that every spectrum will be checked after the survey is complete and all data for each target can be combined, and a cross-check between the `real time' quick redshifts and the final values will be used to pick up errors or ambiguous objects.

    9.  QSO Redshifting Code

    ###############################################################################
This Readme file aims to explain the procedure of identifying objects
from the QSO data in the SDSS-2dF LRG/QSO survey.  This makes use of
the AUTOZ and 2DFEMLINES code written by Lance Miller for the 2QZ with
substantial changes by Scott Croom.  I (SMC) have still some minor
revisions I would like to make to this code.

The basic process is as follows:

1) Reduce the raw data and combine to a final frame containing all the 
   Spectra from one field.

2) Run the AUTOZ code to get automated IDs for every object.

3) Run the 2DFEMLINES code to check each of the IDs by eye.

The eyeball checking can be time consuming, particular if the data
quality is not good.  It may take some practice before users are
efficient at this.

Below I will outline the process in detail, assuming that the these
programs are being run at the AAT:


1) Set up paths for starlink applications etc by typing:

starlink

2) Set the TEMPLATE_AREA environmental variable to point to the
   directory containing the templates:

setenv TEMPLATE_AREA /home/aatssz/scroom/2df/code/
 
3) Make sure your current working directory is the one containing the
   final reduced frames.

4) Run the AUTOZ code (this may take a few minutes), and the current
   version is 3.3 e.g.

/home/aatssz/scroom/2df/code/autoz d08q_030407 

(don't put the .sdf on the end of the file name).  This will result in
a number of files:

d08q_030407q.sdf          - spectra of just QSO candidates
d08q_030407.sortresults   - ID & redshift results
d08q_030407.ps            - plots

5) next run the 2DFEMLINES code (current version 2.4):

/home/aatssz/scroom/2df/code/2dfemlines d08q_030407q 

where  are your initials e.g. smc, in order that we can
identify who have done the eyeball checks.

This new version of 2dfemlines (v2.0+) will read two different types
of .sdf files, one with all the data (i.e. 200 spectra) and a second
with only the QSOs.  This second format also includes the noise data
as extensions in the .sdf and will use this data to process the
spectra.  If the noise data isn't in the .sdf file 2dfemlines will
look for it in the .noisedata file, if it isn't there it will go on to
calculate the noise data again (under the assumption that the .sdf
file must be full 2df data file with all 200 spectra).

The full .sdf files are called:
d08q_030407.sdf 
while the QSO only files are called:
d08q_030407q.sdf 

This is really a hang over from the 2QZ were it made sense to get rid
of all the extra 2dFGRS galaxies in each frame. The output of
2dfemlines (i.e. the .checkedresults file) does not contain the "q"
whichever type of data file is input.
 
6) After a while (a few things are being done, including calculating a
response fn. and removing some bad pixels/CRs etc.)  two windows will
pop up.  One contains a series of mouse-controlled buttons (see
below).  The other contains the first spectrum to be checked.  You
will probably need to shift/resize these.  

(NOTE: If the buttons window is persistently the wrong size on your
machine you can change the window size in the pgplot subroutine call
to PGPAP on line 291 if you have the source code.)

7) The program will step through each object in the .sdf file in turn
(starting with the best IDs through to the worst).  If the ID is okay
then you can simply click on okay to move to the next object.  If the
ID is not correct, then the other buttons can be used to edit the ID.

As well as the buttons there are a number of keys that perform
functions (adding any more buttons would have got very messy!!!).
These are:


p : dump the currently displayed spectrum to a .ps file (filename is
    the object name followed by a .ps e.g.: XQS468Z161.ps).
w : write a textfile containing the spectrum, to a .txt file (with
    file name e.g. XQS468Z161.txt)
o : plot O6 star template
b : plot B4 star template (unless type is WD, then plot DB template)
a : plot A5 star template (unless type is WD, then plot DA template)
f : plot F5 star template
g : plot G4 star template
k : plot K5 star template
m : plot M5 star template
c : clear the stellar template plot

The buttons displayed are as follows:


        z?      ?       ??      OK

        QSO     BAL     star    WD

        NELG    gal     autoz   notes

        orig    zoom    smth    back

        5007    Ha      caK     fit

        MgII    [OIII]  Hy      Hb

        Lya     SiIV    CIV     CIII]


explanations:

"z?"    Toggles on/off the addition of a "?" to the redshift quality.
        This should only be used when single certain line is found
        leading to uncertainty in the redshift.  Do NOT use this for
        all MgII lines, but only in situations where a second line
        could be hidden by bad sky etc.

"?"     Toggles on/off the addition of a "?" to the ID quality.  Add
        this if the ID is possible but not certain.  remember to
        toggle this off if AUTOZ has added a ? but the ID is certain.

"??"    Sets the ID quality to "??" and blanks the ID.

"okay"  Accepts the current ID and moves  

"QSO"   Sets the ID to "QSO"

"BAL"   Sets the ID to "Q BAL"

"star"  Sets the ID to "star" 

"WD"    Sets the ID to "WD". 

"NELG"  Sets the ID to "NELG".  Use this for all narrow line objects,
        use the comment section if you think it is a LINER/NLQSO.

"gal"   Sets the ID to "gal".  This should be used for the occasional
        absoption line galaxies that are found.

"autoz" Resets the ID to the autoz value.

"notes" Allows the user to type a short note (20 characters or less) 
        concerning this object.  These comments should be fairly
        limited in nature.  The following should be sufficient:
        cont  = good S/N continuum seen but no features visible (autoz2.0
                currently uses S/N>10.0, and labels is f/less -
                featureless).
        Qabs  = QSO with absorption line features (e.g. CIV or MgII)
        narrow= QSO with typical lines (e.g. CIV or MgII), but which
                are narrower than usual (a bit subjective, but useful
                to pull out any obviously odd objects).  
        badsky= bad sky subtraction
        fringe= sinusoidal variations in the spectrum due to a broken
                or cracked fibre.
        cline = recentered lines (sometimes autoz is not centered - 
                only to be used in v. EXTREME cases!).
        bad   = something obviously wrong with the spectrum which
                isn't due to the target. e.g.: the spectrum is just
                a negative sky spectrum -> a recent broken fibre which
                isn't labeled as parked in the .sdf header.
        odd   = something strange/unusual about the object - make a 
                further short note concerning what, e.g.:
                enter notes (<20 characters): odd no_CIII]
        DA/DB/DAB... =  if the object is a WD, but autoz hasn't
                        classified it as such, you can add a
                        classification in the notes section.
        O/A/F/G/.... =  if the object is a star but autoz hasn't
                        classified it as such, you can add an approx
                        spectral classification if it is obvious (not
                        required though).
        BAL?   = A possible BAL QSO (certain BALs should use the "BAL"
                 button).
        altz=  = if there are two possible redshifts add the other one
                 on the comments section e.g. altz=1.234
        Use the above comments, and leave a space before adding any
        extra comments.
        If anyone thinks of extra comments please let me know, so that
        I can add them to the list so that they remain consistent
        between different people.

"orig"  Removes the zooming and replots on original scale.

"zoom"  Allows you to zoom in on a specific part of the spectrum.
        Place the cursor at the mininum required (x and y) and click,
        then place the cursor at the maximum required (x and y) and
        click. (Slightly altered from Lance's version.)

"smth"  Carry out simple smoothing.  This does a boxcar smoothing of
        the spectrum.  One click will result in a spectrum where each
        pixel is the sum of 2 original pixels.  A second click will
        sum 3 pixels together, a third will sum 4 pixels together, and
        a forth will revert back to the orignal binning.
        Treat this smoothing with some caution, smoothed sky lines
        etc. can look like broad lines if sufficiently smoothed.

"back"  Go back one spectrum, in the process reseting the ID of the 
        object back to the autoz2 values.

"fit"   Toggle on/off the autoz2 style fitting.

"CaH"   Identify a line as CaH absorption in an absorption line galaxy.
        The ID is automatically set to "gal".

"5007,Ha,Hb,Hy...etc" 
        Click on these buttons and then the line you wish to identify
        with that particular feature.  repeat until you get one that
        works.  NEW:  now when one of these buttons is hit (and the ID
        is a QSO or NELG) a local autoz2 fit will be carried out around the
        line to get the "best fit" redshift (this function can be switched
        off/on with the "fit" button).  If a fit is done then a
        value of chisq will be displayed in the terminal window.  If
        this is zero, then no fit could be found - you need to check
        the line position by eye.

8) when the last spectrum has been "okay"ed then the program will exit
and write files:

d08q_030407.checkedresults   - IDs and redshifts after checking
d08q_030407q_check.ps        - spectra plotted with checked IDs

###############################################################################

IDing NOTES:

Here are some notes to make the eyeballing of objects consistent
between different people.

1) stellar classifications.  Autoz2 gives an approximate stellar
   classification (for both WDs and MS stars):

  i)   Unless the autoz2 classification is clearly wrong (e.g. DF, but
       is obviously G/F) keep the autoz2 classification.  Just remove or
       add "?"s as required.
  ii)  If the autoz2 classification is clearly wrong then use the
       star/WD classifications.
  iii) If you are not sure whether the object is a WD or just a MS
       star, classify it as "star" and put "DA?" etc in notes.

so for example:
 ID  q notes
A3         : A certain star with approx correct autoz type.
F7   ?     : A possible star, with approx correct autoz type. 
DA         : A certain WD with approx correct autoz type.
DB   ?     : A possible ID, which if correct is probably a DB.
star       : A certain star, wrongly classified by autoz2 (either wrong
             spectral type or wrong ID all together).
star ?     : A possible star, wrongly classified by autoz2 (either wrong
             spectral type or wrong ID all together). 
WD         : A certain WD wrongly classified by autoz2.
WD   ?     : A possible ID, which if positive is probably a WD.
star   DA? : A certain star, that is possibly a DA WD.

The bottom line is that if the object is certainly a star of some sort
then in should NOT have a "?" quality flag (this is a little different
from the previous suggestions but I think it makes more sense).

2) NELG classifications.  Autoz gives the classifications
   NELG/LINER/NLQ.  

  i)  Do not change a LINER/NLQ ID to NELG.  Even if they are not
      strictly correct they do give us extra information as to the line
      ratios in these objects. 
  ii) Any object which is an NELG/LINER/NLQ but isn't IDed as such by
      autoz2 should be classified as "NELG".


3) BAL QSOs.  For a certain BAL QSO use the "BAL" button.  If you are
   not certain of the BAL classification then classify the object as
   "QSO" and put "BAL?" in the notes section.

4) ***REMEMBER*** to check the quality flags before hitting "okay".
   If the object is a certain ID then you should remove the ?s.

###############################################################################

    10.  Obsplan Tool

    ~observer/obsplan.tcl will bring up the Obsplan tcl/tk tool, which is very useful for planning observations. Just enter the necessary information at the top of the form (UT/sidereal twilight times, etc), then field information (name, ra, dec), then you see what airmass the fields are at, etc.

    11.  End of Night Summary

    Please write a summary at the end of the night, giving a list of the fields observed, reduced, and redshifted; notes on any problems which significantly affect the data; and the locations of the various data files. Use the file observing_notes.html (Sept 2004) as a template: available from the UQ website, or from /epping/rdc/rdc/2dflrg/.

    Last Revision: RDC, 07 Feb 2005