Next: A warning for users
Up: HOW TO COMBINE MULTI-CONFIGURATION
Previous: HOW TO COMBINE MULTI-CONFIGURATION
Before you attempt to use DBCON, I suggest that it would be useful to
read the DBCON EXPLAIN file. DBCON will concatenate two
single-source files. To combine more than two requires you to run
DBCON a number of times, always appending the new file to the output
of the previous concatenation.
DBCON can make some checks to make sure that you are not combining
data that it thinks do not merit concatenation. It can check
consistency of frequency and position. At this point, however, I am
only considering the combination of data for the same source at the same
frequency from different configurations and observing dates. Discussion
of how to combine data of different frequency is delayed until the
discussion of imaging (§ 15).
- The dopos adverb is used to control
consistency checks. Turn on the frequency check with
dopos(2,1)=1. If you set dopos(1,1)=1, DBCON will check that
the data bases have the same phase centre. If they don't, it will shift
the second set to match the first. This is so that observations of the
same field but with different pointing centres can be combined. Of
course, for this to be successful, the primary beam correction
difference must be negligible. If you try to combine different sources,
you will notice messages about phase shifts, but it will still do it and
you will be in a mess. Note that DBCON cannot phase shift compressed
data, so you should disable this check by setting dopos(1,1)=-1.
It will fail even if it doesn't really need to shift the data, but
the check is requested and the data are compressed. If you need DBCON\
to phase shift the data, you will have to uncompress it with
UVCMP.
- You must be careful with the adverb doarray, which allows data from
different configurations to retain their identity. It is very
important that this identity be retained, otherwise, the self-calibration
procedure, which must solve for the antenna gains as a function of time
(similarly to the normal point source calibration procedure), will get more
than confused; put doarray=-1. What really happens is that the input
data bases are separated by 5 days in time in the output file, and they are
treated as separate sub-arrays. This means that the output file will contain
as many antenna (AN) files as there are input data bases to be joined up. The
version number of the AN file is its sub-array number. You can often point at
separate sub-arrays with the subarray adverb that many AIPS tasks
recognize.
- Because of the change to the weighting scheme used by ATLOD (see
discussion of dparm(3) in § 4.4) it may be necessary
for you to make use of the reweight adverb. This enables you to
apply a scale factor to the weights of the input visibility files. For
example, consider two visibility files. Let us say the first was loaded
with ATLOD before March 03 1994 (the day ATLOD changed and the earth
stood still). Let us say the observations for the other file occurred
after February 22 1994 and in addition the data were loaded after March
03 1994. The first file will have all weights set to unity, the second
file will have weights equal to the integration time in units of
15 seconds (note that to discover what weights you have in your data,
you can use PRTUV to list a little bit of it, and the listing
will include the weights. ATLOD, in its more recent incarnations,
will also list and write into the history file the average weight).
Let us say the integration times were 20 and 30 seconds for
the former and latter files, respectively. Your goal is for them to
have weights of, say, 20/15 and 30/15 (it does not really matter
what the absolute values are, as long as they are consistent. The
second file will already have a weight of 30/15=2, so you simply
need to set reweight=4/3,1 to scale the weights of file one
appropriately and leave the weights of file two unchanged.
You can actually find the reweight factors a bit more
precisely than this and in a way that does not require you to understand
wgat weighting scheme ATLOD is using this week. To do this you have
to make a naturally weighted image (see § 15). The
imaging routines all give you a number, the ``sum of the gridding
weights''. Let us say that these numbers are and 200 for images
made from files one and two, respectively. If you wish them to
contribute equally to an image made from the concatenated data, then
you should set reweight=1,/200.
DBCON |
inname,inclass | First input file |
inseq,indisk | to combine |
in2name,in2class | Second input file |
in2seq,in2disk | to combine |
reweight=#,# | May need to set, otherwise leave 0 |
dopos(1,1)=1 | Check phase centres |
dopos(2,1)=1 | Check for frequency agreement |
doarray=-1 | Separate input files in time by 5 days |
The end result of all this should be one single-source file (for each
source and frequency) containing data from two or more observing runs.
If you are combining multi-channel data bases (e.g., for spectral-line
work), you can only combine data bases with identical frequency axes.
This means they should have the same number of channels and the header
items describing the axis should be the same. This is why CVEL\
does not change the frequency axis description.
Next: A warning for users
Up: HOW TO COMBINE MULTI-CONFIGURATION
Previous: HOW TO COMBINE MULTI-CONFIGURATION
nkilleen@atnf.csiro.au