Input fields: ============= -A guide to what's happened to all these targets, and what to do at the telescope! Stage 1: -------- The input fields (sds files) that will be delivered to the telescope have been through the tiled configuration process (GBD's code). This outputs a list of fibre and target pairs which are read into a version of the AAO configure program in Oxford. Stage 2: -------- A list of candidates for merging objects is automatically generated by the code that writes the original field files, as is a list of input positions for the DSS getimage program for each field. All the images are extracted to small postage stamp images and stored in one directory per field. These images are visually inspected using the visual schnauzer option to xv, and a list of bad objects (screwy fiducial stars, noise objects and blank fields) is generated by the inspector. --This process provides for the detection of really screwy objects like 10th mag stars which we don't want getting under the fibres! Stage 3: -------- Images for objects in the list of merger candidates are automatically analysed by a seperate program that generates a tweaked input position, calculates the new object position in J2000 from the DSS image headers and converts this position back to 1950 using the FK5-FK4 conversion routine provided in iraf (skycalc). (I think this uses slalib routines). The new position is then inserted into the .fld file by a perl script which also picks up the list of bad objects and strips them out of both the fld file and the import file described above. The UTDATE entry in the .fld file is also changed at this point to coincide with the middle observing date of the current run. This procedure also renames the objects to remove underscores (Matthew) and changes the object priorities so that the priorities in the final .fld file are as follows (This is IMPORTANT!): Fiducial 9 Object allocated to a fibre by GBD's configure 6,7,8,9 depending on type Object missed completely by GBD's configure code 5 Object allocated to a fibre on a different field 4 There are now a few more stages that need to be applied before the field can be observed: Stage 4: -------- The field is imported into AAO configure (gbd's version in Oxford) by reading in the .fld file and the import file and then allocating fibres to targets by names. HOWEVER, the set of input fibre positions and plate coefficients used in the original configuration passes will, in general, be different to that in use at the telescope on the night of observation. These seem to change each time the system is taken apart, so I just used the set that was in force at the time I started the configuration run. When the field is imported into configure a check is run and errors are reported. I then run a re-allocation routine to try and catch the targets that were lost as a result of these errors. This will generally leave some free fibres, so a second allocation pass is run to try and allocate any extra objects, with an argument to tell the routine how many fibres to leave free for sky positions. This will attempt to allocate objects by priority, and so may pick up extra targets that were missed by the original configure pass, or may just allocate a few fibres to priority 4 targets to give us some duplicate observations. Finally the system will generate a radial grid of sky positions and try to allocate the remaining fibres to sky, although there may be a few left over because of acccessibility problems so I will add in a few extra fibres or objects depending on what's available. Stage 5: -------- The list of allocated sky positions can be output from configure (in Oxford) as an input file for getimage (note that these are J2000 coords, not B1950!) These should be extracted using getimage and eyeballed using the visual schnauzer in xv just to make sure we didn't hit any bright stars! Stage 6: -------- Finally the field should be checked for the relevant range of hour angles (+-4h), and this may generate a few errors which should be addressed by hand. --it's usually possible to recover the objects that fail by rearranging the fibres to introduce a few extra crossings, but the following rules of thumb should be applied: 1) always deallocate a priority 4 target or sky fibre if this causes a collision:- priority 4 targets are duplicates anyway, and sky fibres can be re-placed by hand (move the cursor to a position and press CTRL-middle-mouse-button to allocate the current fibre to this sky position) 2) any target higher than priority 5 should be preserved if at all possible. -You can usually do this by rearranging fibres, or by replacing the fibre allocated to this target with one allocated to a nearby sky position/priority 4 target and then placing the original fibre on sky or another target. 3) sometimes there will be a collision between an object fibre and a fiducial fibre. It's not easy to move the fiducial fibre as there aren't that many targets for them (big red circles), but they've all been eyeballed and passed as OK, so there's no problem with using a different star! Operational Considerations: =========================== Stages 1--5 should take place in Oxford prior to observation. Stage 5 can be done either in Oxford or at the telescope. --I will leave some more detailed instructions in Oxford for Will or someone else to do this in the event that I'm not around prior to a run. HOWEVER: it's very important that the Oxford copies of the input files for the positioner are up to date in Oxford before importing the data, otherwise chaos will ensue at the telescope and we'll degrade the quality of the final survey by losing objects unecessarily. --I may be able to change the import routine to make this a little more flexible, and I will investigate this before the January run. Stage 6 can be done at the telescope by the conscientious observer, but be advised that to run through a set of 34 fields doing the +-4h checks in a way that I consider conscientious took the best part of two full working days, so DON'T leave this till the last minute!!! I will also investigate a way of making the postage stam images available on the web, as they're useful when you find there's no signal down a particular fibre due to a plate screw up that might have been missed in the eyeballing pass! --I would distribute these to all comers, but they add up to about 4Gb for the sgp alone!