Preparing Your Data in AIPS

If you feel that you would rather load and flag your data within AIPS but calibrate within Miriad, a number of steps need to be performed:

1. The AIPS task to load RPFITS data is ATLOD. Copious information on this is given in Neil Killeen's `Analysis of Australia Telescope Compact Array Data'. However, before loading your data into AIPS, if you have measured all four polarisation correlations, it is best to do a preliminary run of ATLOD using the optype ='sysc' option. This run writes a text file (XYPHS_xx, where xx is your AIPS number) containing the XY phase for each antenna into the FITS area (/DATA/FITS). Text files of system temperature are also written. Although the various selection parameters of ATLOD are still active, you probably want to see all the data. Most of the other ATLOD parameters are unimportant for this.

   AIPS/ATLOD
   optype = 'sysc'	Load the data
   freqsel	Select all data
   ifsel
   source
   timer

   You should plot the phases (and the system temperatures) with the Unix program pltsys - which prompts you for the name of the text file to plot as well as other information. Examine these plots carefully to assess their quality. You should use these plots to choose your reference antenna (for calibration purposes). Choose the reference antenna to be the antenna having the cleanest, most stable XY phase measurements.

   Determine some mean value of the XY phase for each antenna from the plots. The command pltsys prints out both the average and median XY phase. As there are often outliers, the median is more likely to reflect the true XY phase value. Getting a good value is only important for the reference antenna. Do not be too concerned if there are large jumps in the XY phases on antennas other than the reference antenna.

2. No XY Phase Correction in ATLOD: You are now confronted with the decision of where to correct the XY phase of the reference antenna. Your choice will depend on taste, circumstances and the quality of the XY phase measurements. There are three main options:
   • Correct using the Miriad task atxy (described later). This is generally the best option. This allows you to correct for the variation of XY phase with time. This will be essential if there is significant time variability. To use this approach, you will need to keep a copy of your XY phase text file, XYPHS_xx.
   • Correct in AIPS ATLOD. This has been the approach once recommended, and is still useful if the XY phase of at least one antenna is quite constant with time. It also has the advantage of getting the XY correction step over and done with early. However the user input can be error prone and tedious if the observation switches frequency with time.
   • When only XX and YY correlations have been measured, absolute XY phase becomes irrelevant. Indeed it is not measured. In this case you do not need to apply any XY phase correction.
   There are a number of other possibilities, which will not be described here.

   If you are going to correct the XY phases in Miriad, or if you are not going to correct XY phase at all, then you should now load your data without applying any XY phase. It is probably worth your while to pretend that the polarisations are circular rather than linear with the usual fudges, as not all the AIPS software will recognise linears (most of the calibration software will). You must not convert to Stokes parameters. The appropriate ATLOD parameters are

   AIPS/ATLOD
   optype = 'load'	Load the data
   aparm(1) =-1	Label as circular
   cparm(5) =0	Do not apply any XY phases.

3. Correcting XY Phases with ATLOD: If you want to correct the XY phases with AIPS ATLOD, the XY phases on at least one antenna should be reasonably constant with time (vary by no more than a few degrees). In this case, give ATLOD the values of the XY phases that you determined from the plots discussed above. Input these values into ATLOD with the xyphase array. You must enter one value per antenna for each frequency. If you have more than one frequency, you must enter XY phases for all six antennas, even if you do not have antenna 6 in the array during the observation (the XY phase value is not important for this antenna, of course). Note that even if the values are close to zero, you still should apply them. Applying a value of zero is different from not applying anything. Again, you must not convert to Stokes parameters.

   AIPS/ATLOD
   optype = 'load'	Load the data
   aparm(1) =-1	Label as circular
   cparm(4) =1	Use xyphase array and
   	not the on-line values
   cparm(5) =1	Apply XY phase to Y gains
   xyphase	Assign the XY phases here

4. Now flag the data in the way you would normally do with the AIPS tasks SPFLG, TVFLG, IBLED, and UVFLG. Using SPFLG is highly recommended, particularly at 20 and 13 cm, to check for interference.

   For continuum work, to save disk space and to speed access to the data, you may consider averaging your channels together to form ``channel-0'' datasets, using task AVSPC. While this causes very little degradation for 3 cm observations, forming ``channel-0'' results in bandwidth smearing in 13 and 20 cm observations, and so is probably inadvisable for high dynamic range work there. For high dynamic range work at 6 cm, it is debatable whether averaging is detrimental. If in doubt, do not average. It is always possible to form a channel-0 dataset later anyway.

   One other consideration in determining whether or not to average is whether or not you are going to apply XY phase corrections with task atxy. For obscure reasons, if you used AIPS ATLOD, atxy needs to know the ``sideband indicator'' of the data. The sideband indicator, which is $\pm 1$, is copiously reported by ATLOD, both in its output to the terminal, and in the history file. The sideband indicator also happens to be the sign of the channel frequency increment. This is how atxy normally determines them. However, if you form a channel-0 dataset, the sign of the frequency increment is lost! So if you give atxy a channel-0 dataset, you will also have to tell it the sideband indicators. You must give it a sideband indicator for each IF. Provided the sideband indicator remains constant with time, this is little more than an annoyance. However if the sideband indicator varies with time, you are in some trouble. Overall it is best not to form channel-0 datasets if you used AIPS ATLOD and you are going to use atxy.

   After flagging (and possibly averaging), write your data as a FITS file using FITTP. It is probably most convenient to write out a multi-source file. At this stage you have no calibration - only flagging tables (which you can apply in Miriad later).

5. Read the data into Miriad using fits. Task fits does not apply AIPS flagging tables (FG tables). Instead you have to use another task to do this - fgflag. Tasks fits and fgflag are discussed in Chapters 8 and 10 respectively, although they are usually fairly straight forward. An exception is for spectral line observations, where the velocity system should be defined with fits - see Chapter 16.

   FITS
   in=MULTI.FITS	FITS multi-source file to be loaded into Miriad
   op=uvin	Read uv data in
   out=multi.uv	The output Miriad dataset.

   FGFLAG
   vis=multi.uv	Apply AIPS flagging table to the data.

   At this stage it is worth running uvindex. This produces a summary of your dataset, which you should probably save in a log file. Inspect this summary carefully, particularly the frequencies (especially in fits complained about inconsistent frequency definitions). If the frequency information looks incorrect, read Chapter 8 more carefully and/or seek help.

   UVINDEX
   vis=multi.uv	Dataset to summarise.
   log=multi.log	Output log file.

6. Skip this step if you have only measured two polarisation products. Otherwise now is the time to apply your XY phases to the data if you have not already done so with AIPS ATLOD. As mentioned above, the task to do this is atxy. If you used AIPS ATLOD, you should also have XY phase text file that it produced. Let us discuss the various input parameters:
   • vis: The name of the input dataset. Generally this will be a multi-source dataset.
   • xyphase: You need to give the name of the XY phase text file that AIPS ATLOD produced. Generally this will be of the form XYPHS_xx, where xx is your AIPS user number. Task atxy can also be used to correct the XY phases of data loaded with Miriad atlod where options=xycorr was not used. In this case, the on-line XY phase measurements are contained within the dataset (as the variable xyphase), and you do not require a input text file.
   • refant: It is best to correct for a time varying XY phase on only one antenna. All other antennas are assumed to have an XY phase of zero (in the AIPS XY phase convention - see below). The antenna which is corrected for a time varying XY phase should be the antenna with the cleanest, most constant XY phase (as determined by the pltsys or varplt plots). This will be the antenna you will use as the reference antenna for calibration purposes. It is not necessarily the same as the antenna used as the reference during the observation - although it will often happen that it is the same. You give this antenna via the refant keyword. Note that if you have broken a dataset up into sub-files before using atxy (e.g. break it up into a calibrator dataset and a program source dataset), then you must correct the same antennas in all datasets.
   • interval: This gives one or two numbers, both in minutes, which determine the length of a solution interval (the time interval over which an XY phase is solved for; should be comparable to the time scale on which the XY phases are constant over - use varplt). The first number gives the maximum length of a solution interval, whereas the second gives the maximum gap within a solution interval. A new solution interval is started when either the maximum time length is exceeded, or a gap larger than the maximum gap is encountered. The default maximum length is 30 minutes. The default maximum gap is the same as the maximum length.
   • break: If significant steps in the XY phase occurs during the observation (usually caused by resetting the delays), then it is best to prevent a solution interval spanning this time. To do this, you list the times where there was a break in the XY phase. The times are given in the normal Miriad time format (i.e. either hh:mm:ss or yymmmdd:hh:mm:ss, such as 93oct18:19:21:00, for 7:21 pm on 18 October, 1993).
   • sideband: As noted above, atxy needs to know the sideband indicator for for data loaded with AIPS ATLOD. For multi-channel data, this will be the sign of the frequency increment. However this sign will be lost if you form a channel-0 dataset. In this is what you have done, you will need to tell atxy the sideband indicators for each IF band. Note that atxy cannot cope with a channel-0 dataset if the sideband indicators change with time.
   • out: The name of the output dataset. Apart from application of the XY phase, this will be a copy of the input dataset.
   Typical inputs to atxy are given below

   ATXY
   vis=multi.uv	The input dataset.
   xyphase=XYPHS_56	The input AIPS ATLOD XY
   	phase text file, or
   xyphase	leave unset if Miriad atlod was used.
   refant=3	Correct antenna 3 in time varying way.
   interval=#	Solution interval. Default is 30 min
   sideband	Sideband indicator. Leave unset for
   	multi-channel data.
   out=multi.uvxy	Output XY phase corrected data.