Thursday, May 17, 2007

WBR Step 1. Loading in Your Data

Ok, I'm going to start writing an introduction and overview of how to reduce data in AIPS for our REU students this summer. I think the AIPS cookbook overview is vague and opaque, so maybe this will help fill in some gaps. Starting with...FILLM!

From the archive, you'll probably have several files with the observing code, an underscore, and a number. For example: AS755_1.
Move all of the files for your relevant observing program into a convenient directory. Start aips from there.

Now it is time to load these data into AIPS, using the task FILLM. The most important parameters to input here are:
INFILE: pwd will tell it to look for the files in the directory you just started aips from. give it the observing code and an underscore as a prefix.
VLAOBS: give it your observing code
OUTNAME: you can give your data some cute name when it is loaded into aips
DOWEIGHT = 1: you want to load in the data weights!! important!
most of the other parameters here i still don't really understand. I bet some of the time-averaging parameters under CPARM are quite useful, but I'm not sure which are most important to think about. The defaults are probably fine if you're just beginning.

AIPS 1: FILLM Task to read a VLA archive tape
AIPS 1: Adverbs Values Comments
AIPS 1: ----------------------------------------------------------------
AIPS 1: INTAPE 1 Tape drive number
AIPS 1: INFILE 'PWD:AS755_' Non-blank: disk file name
AIPS 1: NFILES 0 Number of files to advance
AIPS 1: Also applies to ON-LINE!
AIPS 1: BAND ' ' Freq. band (4,P,L,C,X,U,K,Q)
AIPS 1: QUAL -1 Source qualifier -1=>all
AIPS 1: CALCODE ' ' Calibrator code
AIPS 1: VLAOBS 'AS755' VLA obs. pgm. name ('AZ99')
AIPS 1: VLAMODE ' ' VLA obs. mode (' '=any)
AIPS 1: REFDATE ' ' Reference date. 'yyyymmdd'
AIPS 1: TIMERANG *all 0 Timerange selected
AIPS 1: BCHAN 1 First spectral channel
AIPS 1: ECHAN 0 Highest spectral channel
AIPS 1: OUTNAME ' ' Output UV file name (name)
AIPS 1: OUTSEQ 0 Output UV file name (seq. #)
AIPS 1: OUTDISK 1 Output UV file disk unit #.
AIPS 1: DOUVCOMP 1 1 (T) => compressed data
AIPS 1: DOALL -1 1 (T) => write all data.
AIPS 1: DOCONCAT -1 1 (T) => append data to old
AIPS 1: files
AIPS 1: NCOUNT 0 Number of files to read
AIPS 1: DOWEIGHT 1 Use nominal sensitivity to
AIPS 1: scale weights?
AIPS 1: = 10, use Memo 108 weights
AIPS 1: DOACOR -1 Load autocorrelation data?
AIPS 1: CPARM 0 0 User options.
AIPS 1: 0 0 1 => Avg. time (seconds)
AIPS 1: 0 0 2 => bitmap (see help)
AIPS 1: 0 1 = 16 for planets & Sun
AIPS 1: 1 0 3 => Max. OK IF status
AIPS 1: <=3 => 3
AIPS 1: 4 => Control dropping of
AIPS 1: shadowed data
AIPS 1: < 0 => no shadow check
AIPS 1: 0 => 25 m limit
AIPS 1: > 0 => Shadow limit is
AIPS 1: CPARM(4) in meters
AIPS 1: 5 = channel code (see HELP)
AIPS 1: 6 => Subarray number-see HELP
AIPS 1: 7 => FQ entry tolerance (kHz)
AIPS 1: (see HELP)
AIPS 1: 8 => CL table time incr. min
AIPS 1: 9 => TY table time incr. min
AIPS 1: 10=> Calibrator avg. time
AIPS 1: DPARM *all 0 Selection by Frequency
AIPS 1: 1+2 => A Frequency (Hz)
AIPS 1: 3+4 => B Frequency (Hz)
AIPS 1: 5+6 => C Frequency (Hz)
AIPS 1: 7+8 => D Frequency (Hz)
AIPS 1: 9 => tolerance for A&C
AIPS 1: 10 => tolerance for B&D
AIPS 1: BPARM *all 0 Opacity and Gain curve
AIPS 1: control (see help)
AIPS 1: IN2FILE *all ' ' Antenna gains file

Watch the output and make sure AIPS read in the data from all of your files. After FILLM successfully runs, if you type "uc", you should see two files: one CH 0 and one LINE. The CH 0 data is basically the line data with the channels averaged together (well, the central 75% of channels). CH 0 data may be useful for calibration.


amanda said...

A couple of comments:

I generally set douvcomp false because otherwise a limited set of gains is kept (from the AIPS documentation: "in compressed form, a single weight is kept for all spectral channels, IFs, and polarizations of the sample time and baseline). Don't know if this will matter much if you aren't doing polarization. I use doweight=10, but I don't think that matters too much. I usually also set cparm(8) = 0.1, so that I don't have to reindex if I want shorter solution intervals.

These are more style things than anything else though.

katie d. said...

I'm with amanda on douvcomp -1.

I set cparm(8) = my integration time length, which is usually 10 seconds. Among other things, this makes it easier to flag the first integration of every scan, which in VLA data is often bad.

For my spectral line data, I set cparm(7) = -1 if there is a range of freqencies that are all using the same calibrators. Different FQ numbers will cause calibration problems, so unless you actually really want to distinguish your data by FQ numbers and have proper calibration observations for the different frequencies, don't let fillm give different FQ numbers to your different sources.

Laura said...

Yeah, good call. It seems particularly important during the EVLA upgrade to get as good of weights as possible, and therefore you definitely want to set douvcomp -1. Thanks for the tip!