AIPS NRAO AIPS HELP file for FILLM in 31DEC18



As of Sat Jun 23 2:15:02 2018


FILLM: Task to read a VLA archive tape

INPUTS

INTAPE                             Tape drive number
DATAIN                             Non-blank: disk file name
NFILES                             Number of files to advance
                                   Also applies to ON-LINE!
BAND                               Freq. band:
                                     (4,P,L,S,C,X,U,K,A,Q)
QUAL             -1.0    32767.0   Source qualifier -1=>all
CALCODE                            Calibrator code
VLAOBS                             VLA obs. pgm. name ('AZ99')
VLAMODE                            VLA obs. mode ('  '=any)
REFDATE                            Reference date. 'yyyymmdd'
TIMERANG *                         Timerange selected
BCHAN             0.0      512.0   First spectral channel
ECHAN             0.0      512.0   Highest spectral channel
OUTNAME                            Output UV file name (name)
OUTSEQ           -1.0     9999.0   Output UV file name (seq. #)
OUTDISK           0.0        9.0   Output UV file disk unit #.
DOUVCOMP         -1.0        1.0   1 (T) => compressed data
DOCONCAT         -1.0        1.0   1 (T) => append data to old
                                   files
APARM    ?                         1 0 : append files
         ?                           1 : each scan => new file
         ?                           2 : close current files,
         ?                               create new ones.
         ?                           3 : immediate stop.
NCOUNT                             Number of files to read
DOWEIGHT         -1.0       10.0   Use nominal sensitivity to
                                   scale weights?
                                   = 10, use Memo 108 weights
DOACOR           -1.0        1.0   Load autocorrelation data?
CPARM                              User options.
                                   1 => Avg. time (seconds)
                                   2 => bitmap    (see help)
                                        = 16 for planets & Sun
                                   3 => Max. OK IF status
                                        <=3 => 3
                                   4 => Control dropping of
                                        shadowed data
                                        < 0 => no shadow check
                                        0 => system bit when ok
                                             else 25m
                                        > 0 => Shadow limit is
                                              CPARM(4) in meters
                                   5 = channel code (see HELP)
                                   6 => Subarray number-see HELP
                                   7 => FQ entry tolerance (kHz)
                                        (see HELP)
                                   8 => CL table time incr. min
                                   9 => TY table time incr. min
                                   10=> Calibrator avg. time
DPARM                              Selection by Frequency
                                   1+2 => A Frequency (Hz)
                                   3+4 => B Frequency (Hz)
                                   5+6 => C Frequency (Hz)
                                   7+8 => D Frequency (Hz)
                                   9 => tolerance for A&C
                                   10 => tolerance for B&D
BPARM                              Opacity and Gain curve
                                   control (see help)
CALIN                              Antenna gains file

HELP SECTION

FILLM
Task:  This task will create AIPS uv data file(s) and fill it by
       reading selected data from a VLA archive tape or disk files.
       Multiple output files will be written if necessary; continuum,
       "channel 0" and line data will be written into separate files.
       Also, if several spectra are present with incompatible
       polarizations (e.g., RR and LL at different frequencies), they
       are written into separate files.  Up to 8 simultaneous output
       files may be written.
          Data from several files on a tape or disk and from several
       tapes may be written into the same output file(s).
          This program reads VLA tapes written after January 88.
       Older tapes may be read by the obsolescent FILLR; the
       alternative will be to have older tapes transcribed to the new
       format.  This has been done with the entire VLA archive.

       **************************************************************
       This version of FILLM writes times centered in the on-line
       integration.  Previous versions wrote times at the end of the
       integrations.  Do not use UVFIXPRM(13) any longer in UVFIX for
       data loaded using this version of FILLM (31DEC07, 6 July 2007)
       **************************************************************

       The default opacity model is now based on an algorithm of Josh
       Marvel.  It uses the older model which mixes surface weather
       data with a seasonal model but always evaluates it at K Band.
       Then a large table is used to relate the K Band opacity to that
       at other frequencies.  This allows different opacities for each
       IF, while the old method used the same opacity for any IFs
       within the same band.  The old method is available by setting
       BPARM(1) to 100.  See EVLA Memo # 143 "Improving the frequency
       resolution of the default atmospheric opacity model" by Josh
       Marvil (NRAO), 04/06/2010.

          If you are at the Array Operations Center in Socorro, you
       may run a special version of FILLM to load your data as they
       are being observed.  Set VERSION='ONLINE' and mount an
       "ON-LINE" tape device for INTAPE.  In the new release of this
       system (after 20-Aug-2003), FILLM will begin reading data from
       the current UT day's disk archive.  Thus, all data taken for
       you during the current day (or even previous days) may be
       loaded, not just those data taken after you started the on-line
       FILLM.  See NFILES below for details.  This new version of
       FILLM will not stop automatically after your run.  Set a stop
       time (carefully) or remember to kill the on-line FILLM after
       your run has finished.  There are several procedures which TELL
       on-line FILLM to take certain actions.  These are QUIT (stop at
       end of scan), FEW (make as few files as possible), MANY (make a
       new file for each scan), BREAK (create new files now, then
       append to them as appropriate), STOP (stop now).  Accessing
       files that are currently being written by on-line FILLM has
       certain risks; BREAK guarantees that the previous files will
       not encounter these collisions.  The multiple files will have
       to be put back together later with DBCON.  On-line FILLM is to
       be used only to load data for an experiment on which you are
       an official investigator.  Any other use has the potential to
       cause serious harm to your scientific career.
Adverbs:
  INTAPE.....Tape drive from which to read data.
  DATAIN.....If non-blank, the data are already on disk and are to be
             read from disk file DATAIN (with the file number
             concatenated).  Thus,
                  DATAIN='MYAREA:VLAtapes/file_
                  NFILE = 4
             will cause files from directory $MYAREA/VLAtapes named
             file_5, file_6, file_7, etc. to be read.  (Note that
             DATAIN= is the only command on the line and there is no
             close '.)     See EXPLAIN for a discussion.
             To use the VLA archive directly, contact NRAO at
                analysts@nrao.edu.
             The web has tools to review the contents of the archive
                http://www.vla.nrao.edu/astro/archive/vladb/
             and to request data from the archive
                http://www.vla.nrao.edu/astro/archive/archreq/
  NFILES.....Number of tape files to advance before reading:
             For tape, 0 is the start of the current file, 1 the next
             file, -1 the previous file.  For off-line disk, see
             DATAIN above.
             For ON-LINE, -1 is the start of yesterday's disk file,
             0 is the start of today's file, and 1 is the current end
             of today's file (i.e. right now).
  BAND.......Name of desired frequency band i.e., '4','P','L','S','C',
             'X','U','K','A', or 'Q'.  1st char. => A-C IF, 2nd char.
             => B-D IF.  If 2nd character is blank, it is set to the
             1st character.  A resulting blank means fill any band
             found.  Bands are defined as:
                4   0.0  -  0.15   GHz      X   8.0 - 12.0   GHz
                P   0.15 -  0.7             U  12.0 - 18.0
                L   0.7  -  2.0             K  18.0 - 26.5
                S   2.0  -  4.0             A  26.5 - 40.0
                C   4.0  -  8.0             Q  40.0 - 56.0
             Note that some receivers in the EVLA can observe
             frequencies in more than one of these official bands.
             Beginning in Feb, 2008, the VLA data format includes a
             code to tell which receiver was used.  FILLM will use
             that when available, to select data by band to to name
             data sets.
  QUAL.......Source qualifier number, -1 => all.
  VLAOBS.....The name of the observing program on the VLA, e.g.,
             'AZ99'.  blank => all found in the selected time range.
             It is a very good idea to set this parameter.  Note, if
             this parameter is set, real-time FILLM will stop after
             some data are read and then data some another program is
             encountered.
  VLAMODE....The VLA observing mode (2 chars.).  Normal interferometer
             mode is '  ' and all modes are accepted when VLAMODE is
             blank.  Solar mode is 'S*' where *=any character.  In solar
             mode, data with the specified VLAMODE and data with blank
             mode are kept.
  REFDATE....Reference date 'yyyymmdd' ('dd/mm/yy' accepted but
             deprecated),  ' ' => first found
             Better get it right!!! ... or use ' '.
             Task FXTIM can be used if you end up with negative times.
  TIMERANG...Range of IAT times to be read wrt REFDATE. 0=>all.  See
             EXPLAIN FILLM for details.
  BCHAN......Lowest 1-rel spectral channel number to be passed.
             Continuum and channel 0 data will always be written. 1 is
             the first channel of the spectrum.  0 => 1.
             In Z-mode, where the on-line system also writes the 0-lag
             channel, BCHAN and ECHAN are ignored and full spectrum
             containing that 0-lag channel are      written.
  ECHAN......Highest 1-rel channel number to be read.
             0 => highest on tape.
  OUTNAME....Output UV file name (name).  Standard behavior with
             default 'ref. date'.  The class for continuum data will
             be 'm BAND' where m is the band code.  Line data will be
             broken up into 'channel 0' and proper spectral data with
             classes 'CH 0  ' and 'LINE  '.
                If the output file already exists then the new data is
             appended to the end of the existing file if DOCONCAT is
             true.
  OUTSEQ.....Output UV file name (seq. #).  0 => highest unique.
  OUTDISK....Disk drive # of output UV file.  0 => highest disk with
             space for the file.
  DOUVCOMP...If true (> 0) the output data is written in compressed
             format.  This can result in a substantial reduction in
             disk space needed.  WARNING: in compressed form, a single
             weight is kept for all spectral channels, IFs, and
             polarizations of the sample time and baseline.  If the
             weights are substantially different, this may cause a
             serious loss of sensitivity in imaging; see the DOWEIGHT
             option.
  DOCONCAT...If true (>0) then data is to be appended to an existing
             file if possible.  Possible output files can be selected
             by OUTNAME, OUTDISK and OUTSEQ.
  APARM......Used in SHOW/TELL only and only make sense when running
             FILLM in real time
             (1) =1 => new file for each scan,
                 =2 => create new files now, append following scans
                       into them
                 =3 => stop now
                 else => concatenate files read in if similar enough.
  NCOUNT.....Number of files to attempt to read, writing to a single set
             of output files.   0 => all.  Ignored by ON-LINE version.
  DOWEIGHT...> 0 => use nominal sensitivity (plus integration time) to
             set data weights.  Equivalent to mod(CPARM(2),8) = 0.
             Note that weights are not fully retained if the data are
             stored in compressed form (spectral channels,
             polarizations, and IFs are merged in weight at each
             time/baseline sample).  Using "correct" weights for the
             different IFs and antennas allows poorer receivers to be
             downweighted which can make a considerable difference in
             the sensitivity of the final images.  Weights set in this
             way must be calibrated along with the visibilities
             (DOCAL=2 does this in all calibration tasks).
             AIPS Memo 108 by Bryan Butler suggests that the weights
             should be 2.3667 (for continuum before 7/30/98 and all
             line data) and 2.9417 (continuum after 7/30/98) times
             larger than normally computed by FILLM.  To have this
             correction applied, set DOWEIGHT=10.
  DOACOR.....> 0 => load autocorrelation data for spectral-line
             observations.  In continuum modes, the autocorrelation
             data are never loaded.
  CPARM......User options.
             (1)  Averaging time in seconds.  If 0, no averaging
                  beyond that done on-line will be performed.
                  Averaging of a sample will be terminated if the
                  accumulated integration time is >= that requested or
                  the elapsed time from first sample to current sample
                  is >= CPARM(1).
                  Note a special, hidden use: when CPARM >= 100000, it
                  will use CPARM-100000 as averaging time and will
                  produce an OT table (over the top table).
             (2)  u * 2048 + t*1024 + s*512 + r*256 + q*64 + p*32 +
                    n*16 + m*8 + k*4 + j*2 + i
                      where i, j, k, m, n, p, q, r, s, t, u are
                      0 (no) or 1 (yes)
                  i: ignore on-line "Tsys fluctuating" flags.  Other
                     on-line flags are summarized in the IF/antenna
                     status and controlled under CPARM(3).
                  j: NOTE change: a TY table is now always written.
                     j = 0 => fill backend Tsys for P- and 4-band, and
                     front-end Tsys for all other bands. j=1 does the
                     opposite.  Back-end Tsys are used for EVLA
                     antennas since they do not have a front-end Tsys.
                  k: use all correlator pairs in holography mode.
                  m: use nominal sensitivity to set weights
                     DOWEIGHT > 0 also selects this.
                  n: planet/solar mode - do not change source name
                     when position changes.
                  p: include pointing mode data
                  q: output the on-line "channel-0" rather than
                     re-compute it from the current data
                     (on-line sum done before Hanning -> unreliable,
                     but there are modes in which it will be of much
                     wider bandwidth and hence good for continuum)
                  r: ignore changing qualifiers for common source name
                  s: don't load visibilities for which one of the
                     antennas had failed its last attempted reference
                     pointing
                  t: include TIP data
                  u: output correlation coefficients (i.e. not scaled
                     by the measured nominal sensitivity).  Note that
                     Solar data are always written as correlation
                     coefficients.
             (3)  Data with VLA on-line system flag levels>ABS(CPARM(3))
                  are flagged in the output file.  Since flag levels < 4
                  in the archive are merely informative, CPARM(3) is
                  set to 3 when the input is 0 <= CPARM(3) <= 3 and
                  set to -3 when -3 <= CPARM(3) < 0.  This instructs
                  FILLM to flag all data with significant on-line
                  flags.  Setting CPARM(3) negative instructs FILLM to
                  discard all IFs whenever any one or more IF is
                  flagged in the archive.  So, for the most
                  conservative flagging, set -3 <= CPARM(3) < 0.  For
                  typical flagging, set 0 <= CPARM(3) < 3.  For no
                  flagging, set CPARM(3) >= 16 (since the maximum
                  on-line flag level is 15).  Note that this is only
                  antenna/IF level flagging - for antenna level
                  flagging control based on shadowing and failed
                  reference pointing, see CPARM(2), bit s, and
                  CPARM(4).
             (4)  Control shadow checking
                  < 0       => don't drop shadowed data
                  = 0       => drop data using system bit (knows about
                               other science subarrays) in versions in
                               which the bit is reliable, otherwise 25m
                  > 0       => drop data when the baseline <= CPARM(4) m
                               (the antenna that is "behind" is
                               flagged).  Note = 25.0 now causes FILLM
                               to compute shadowing on its own.
             (5)  If spectral-line data for a given source/band
                  with more than one "number of channels" is
                  present in the data, one may be specified by
                  CPARM(5).  The value to be entered is an
                  encryption of the number of channels in each
                  IF.  Values can be determined from task PRTTP
                  or as described in EXPLAIN FILLM.
                  0 => any.
             (6)  Subarray number desired.  Default behavior
                  (CPARM(6)=0) depends on VLAOBS. If VLAOBS is
                  specified, the first subarray with data belonging to
                  code VLAOBS becomes the current subarray.  If VLAOBS
                  is blank, the subarray number defaults to 1.  If the
                  bit 2**3 (8) is on, then the Van Vleck correction is
                  applied to continuum data.  This is needed for very
                  strong sources where the non-linearity of the
                  correction begins to matter.  This is still in an
                  experimental phase and should be used with extreme
                  caution.
             (7)  FQ entry tolerance for deciding what FQ number
                  to assign to a frequency.
                  Continuum:
                  > 0 => tolerance in kHz.
                  <=0 => 100 kHz
                  Spectral line **** CHANGED JULY 2, 2001 ****
                  > 0 => tolerance in kHz.
                  < 0 => all data are given the same FQ number.
                  = 0 => a new number is assigned to data that
                         differ from any previous frequency by
                         an amount greater than the frequency
                         equivalent of 30 km/sec.  (i.e., half
                         maximum Doppler shift between sources
                         180 degrees apart.
             (8)  Time increment (min.) for CL table entries.
                  0 => 5 min.
             (9)  Time increment (min.) for TY table entries
                  0 => input data increment
             (10) Averaging time for calibrators (sec).
                  0 => same as CPARM(1).
  DPARM......Selection by frequency.  If data needs to be
             selected by frequency more precisely than by
             observing band then the desired frequencies and
             tolerances allowed are specified by DPARM.
             1+2 = the frequency for if A (Hz)
             3+4 = the frequency for if B (Hz)
             5+6 = the frequency for if C (Hz)  0 => 1+2.
             7+8 = the frequency for if D (Hz)  0 => 3+4.
             9 = tolerance for A and C ifs.  0 => all.
             10 = tolerance for B and D ifs.  0 => all.
  BPARM......Opacity and gain curve control.
             Opacity is controlled by BPARM(1) and BPARM(10):
                BPARM(1) <  0  ->  no opacity correction.
                BPARM(1) =  0  ->  set BPARM(1) = 20
                0 < BPARM(1) <= 10  ->  opacity correction is done,
                    with zenith opacity given by BPARM(1).
                BPARM(1) > 10  ->  opacity correction is done, with
                    zenith opacity taken as weighted average from
                    surface weather and seasonal model, with the weight
                    of the surface weather portion given in BPARM(10),
                    and the weight of the seasonal model portion given
                    by 1 - BPARM(10).
                BPARM(1) > 100 -> same as above but using the old
                    model where opacities depend only on observing
                    band rather than being the new, IF-dependent
                    values.  BPARM(1) > 100 may be appropriate for
                    backwards compatibility, but the new opacities are
                    thought to be rather more accurate.
                BPARM(10) - weight of surface weather model.  0 -> 0.5
                    If you want 0 set it < 0.
             Gain curve is controlled by BPARM(2):
                BPARM(2) <  0     ->  no gain curve correction.
                0 <= BPARM(2) < 2  ->  gain curve correction, with
                    coefficients which vary as a function of band and
                    antenna.  These coefficients are read from a file.
                2 <= BPARM(2) < 3 ->  gain curve correction, with
                    coefficients which vary only as a function of band
                    only.  These coefficients are read from a file.
                BPARM(2) >= 3     ->  gain curve correction, with
                    coefficients specified in BPARM(3), BPARM(4),
                    BPARM(5), and BPARM(6) for all bands and
                    antennas.
  CALIN......The name of a file to provide the antenna gains used to
             populate the initial CL table.  If CALIN is left blank,
             it defaults to a system file located at:
             'AIPSIONS:VLA.GAINS'.  The file is in free format with
             fields: a 1-character band code, antenna number, begin
             date in form YYYYMMDD, end date YYYYMMDD, and 4 gain
             curve coefficients.

EXPLAIN SECTION

FILLM:  Task to import data into AIPS from a VLA archive data tape.
DOCUMENTERS: W.D. Cotton, G.C. Hunt, B.J. Butler, NRAO.
RELATED PROGRAMS: PRTTP, All AIPS uv programs

                          PURPOSE

    FILLM reads a VLA archive format visibility tape, creates
and fills an AIPS uv data base.  Since such a tape may contain
data for different observing programs and bands, the user must
specify the data that he/she desires.
    An output file may contain either continuum, 'channel 0',
or line data for multiple sources observed in the same observing
band.  AC and BD data will be written as IFs 1 and 2 if
possible.  The two exceptions are: 1) continuum data in split
band mode (e.g., P and L band) and 2) Spectral modes in which two
or more different spectra are observed such that a) they are at
different bands, b) they have a different number of channels, or
c) the channel widths are different.  In these two cases, the
data is split into separate files.
    Output file(s) will be automatically expanded as necessary
and unused space released when FILLM terminates or opens another
output file.  If the file expansion fails for some reason FILLM
will quit saving the data already read.

COMMENTS
     The tape must be properly mounted and the correct density
specified.  This is most easily done using the MOUNT command in
AIPS.  When through reading the tape and no other tape
processing is anticipated, use DISMOUNT to free the tape drive
for other users.  The contents of a tape may be determined using
task PRTTP.

DATAIN
     It is now (as of 2002-07-22) possible to read the data from disk
files instead of magnetic tape.  All files to be read in a single run
of FILLM must reside in the same disk directory and have the same name
except for the last character or characters (no leading zeros) which
define the file number.  Thus, you could copy tape files with dd to
disk and then read them with FILLM or you could ftp them from some
archive to your local disk and then read them.  Note that remote tape
does not work for FILLM (from disk).  The analysts at the AOC in
Socorro will have access to a disk archive and can prepare data files
suitable for loading with FILLM for you to download via anonymous ftp.
For example, when I ask them to find the data for program AB1024, they
find that tape XH02045 data files 1 through 10 contain some of my data.
I then copy those files to a disk directory which I call
          $VLADATA/AB1024
using file names file_1, file_2, ... File_10.  Then in FILLM I set
          DATAIN='VLADATA:AB1024/file_
          NFILES = 0; NCOUNT=10
to read all of these files.  Note that I put the DATAIN= command last
on its line and did not follow it with a close quite (').  This allows
the lower case letters to be retained.

SOLAR MODE
     Solar mode observing is indicated by either a VLA observing code
of 'S*' or by a value of VLAMODE='S*' where '*' may be any character.
In solar mode and when the CALCODE on the tape is ' ' (assumed to be
the sun) then the visibility data written to disk are divided by the
"nominal sensitivity" and the "nominal sensitivity" is written into
the TY table as the system temperature.  When the CALCODE on tape is
non-blank then the visibility data are not modified.  Values of the
system temperatures in the TY table may be plotted with SNPLT.

HOLOGRAPY MODE
     In "holography" data from the VLA, reference antennas are
pointed towards the nominal source position, as normal.  All
other antennas are pointed to an offset position controlled by
the observer.  The angular offset from the nominal position, for
each antenna, is written to tape in the same variables that are
normally used for u,v,w.  In this mode, only data on a baseline
between a reference antenna and an offset (non-reference) is
considered useful.  The correlations observed on these baselines
is stored in the multi-source file in the normal way.  Data is
calibrated as usual and then extracted into complex maps to give
the antenna beam patterns.  These data can now be Fourier
transformed to give aperture distributions.

NFILES:
     AIPS counts tape label, header and trailer labels as
separate files so they should be included in the number of files
to advance.  FILLM will quit after examining one file containing
VLA data if it has not encountered any of the desired data.  If
data has been found it will continue until it either reaches the
end of the tape or reads the maximum number of files specified
in CPARM(9).  Use PRTTP to determine what data is in which file.

TIMERANG:
     The desired time range is expressed as the time will be
described in AIPS; i.e., the time includes the number of days
since the reference date.  The reference date is kept in the
catalogue header as the "Observing Date".  Unless REFDATE is
specified the reference date is undefined (the day number is 0)
until selected data has been found.  The reference day then
becomes the current day if the output file has just been created
or the previous reference day if the file previously existed.
     If REFDATE is specified then its value is used as the
reference value and TIMERANG is with respect to this date.

OUTDISK:
     Make sure there is enough space for the source data on
the specified OUTDISK (see below).  This is especially
important if many large data bases will be written onto the
disk.  A value of 0 allows the task to try all disks to fit the
output files.  In this later case, the output files may reside
on different disks.

DOCONCAT:
     Data from several tapes may be written into the same AIPS
uv data file if DOCONCAT is true (>0.0).  FILLM will search the
catalog for multi source files whose OUTNAME, OUTDISK and OUTSEQ
match.  If DOCONCAT=true then the data will be appended to any
existing file whose attributes match.  Data must be appended to
the file in the correct time order or UVSRT and INDXR must be
run on the file to get it into correct time order.

CPARM:
CPARM(1): The data are averaged using all unflagged data with time tags
within the specified period after the first record in a given
integration.  The time tags, u, v, w, etc. for each visibility written
will be the average for the integration regardless of the flagging;
that is the u, v, w may not correspond to the true weighted average.

CPARM(2): bit map: bit0 + 2*bit1 + 4*bit2 + ... + 256*bit8
     bit0 - The VLA on-line system sets correlator-based flags for
correlators that it determines to be bad.  In continuum, these flags
apply to individual correlation pairs; in spectral line, these apply
to individual spectral channels.  If bit0=1, then these flags from
the on-line system are ignored, in which case the data need to be
carefully edited. (This feature is not yet implemented in the on-line
system though!)
     bit1 - The TY table is now always written.  It has, for each
polarization a Tsys column and a Tant column.  The Tsys column contains
what is called the 'nominal sensitivity', which is the conversion
factor from correlation coefficient to pseudo-Janskys.  These values
are not calibrated and thus do not give true Janskys when used for the
conversion, but are a useful function of time, ignoring the absolute
values.  See also the discussion involving the weighting scheme which
uses the nominal sensitivity values.  Since May 1, 1996 the on-line
system has also written system temperatures to the archive tapes, which
FILLM puts in the Tant columns.  These should be correct antenna
temperatures in Kelvins within the noise tube calibration and total
power fluctuation accuracies (reported to be about 20 percent uncertain).
The default (bit1=0) is to use backend Tsys for P- and 4-band data
and front-end Tsys for all other bands in the Tant columns.  The
reverse (bit1=1) may be done but is not generally recommended.  EVLA
antennas do not have a front-end Tsys and so back-end is always used
for them.
     bit2 - In holography mode, usually only correlations involving
the reference antenna are passed on.  If bit2=1, ALL correlations
involving ALL antennas are passed on.
     bit3 - The nominal sensitivity (see discussion in bit1 above)
can also be used to "correctly" set the data weights by setting
bit3=1.  In principle, you always want to do this whether you plan to
use natural or uniform or any other kind of weighting scheme.  In
practice, if the weights so generated are incorrect, things can go
wrong in an apparently mysterious way.  If PRTUV shows really strange
weights, you can run UVMOD to reset the weights to (t_int/[10 sec]),
the traditional weights value.  If you have set weights "correctly"
(bit3=1), then you should set DOCALIB = 2 in all AIPS tasks allowing
calibration.  This applies the calibration to the weights as well as
to the visibilities.  If you have had to set the weights back to the
old values (or used bit3=0 [and DOWEIGHT<=0]), then use DOCALIB = 1
which calibrates only the visibilities.  DOWEIGHT > 0 also selects
this behavior.
     bit4 - Whenever a source changes position, it is assumed that
this is truly a different source, and that a new entry in the SU
table should be written (with subsequent visibilities at that new
position pointing to that new source).  This is useful for, e.g.,
mosaicing observations (if no source qualifiers are used to
distinguish between the pointing centers).  If, however, your source
truly moves during the observation (the Sun and planets, e.g.), then
you won't necessarily want new entries in the SU table.  Setting
bit4=1 overrides the default behavior, and does not create these
new source table entries.  Note that you will almost always want to
set bit4=1 for solar system observations.
     bit5 - Normally, FILLM will not enter pointing mode data into
the created datasets.  If bit5=1, then pointing mode data will be
included.
     bit6 - Because of demonstrated problems with the on-line version
of the "channel-0" dataset in some instances, FILLM now recomputes
channel-0 on-the-fly from the input line dataset by default.  Setting
bit6=1 overrides this behavior, and takes the channel-0 directly from
the archive.  This may be useful if data selection has been used to
reduce the number of channels written to the archive, so that the
on-line channel-0 is of wider bandwidth than that created by FILLM.
     bit7 - unsupported.
     bit8 - Whenever a source qualifier changes, FILLM normally makes
a new entry in the SU table (with subsequent visibilities with that
qualifier pointing to that new source).  Setting bit8=1 overrides this
behavior and essentially ignores the source qualifiers, using only the
source names to determine when new SU entries should be made.

CPARM(3): The VLA on-line system sets IF-based flags for IFs that it
determines to be bad from monitor information.  Flags are given the
values 0-15.  In general, flag values less than 4 are informative and
do not reflect bad data quality.  CPARM(3) defines the level at which
the data is considered good: the higher the value, the more data is
accepted.  If CPARM(3) is set above 3, then the data need to be
carefully edited.

CPARM(4): Data for which one antenna shadows another antenna is
normally rejected since in crosstalk and interference can pose serious
problems at very short baselines.  CPARM(4) specifies the shortest
baseline in meters the user considers to be unaffected by shadowing.
Any visibility with a baseline shorter than CPARM(4) will be flagged.
CPARM(4) = 0 defaults to the obvious case where the baseline equals
one antenna diameter D.  Sometimes a more conservative treatment may
be necessary, and CPARM(4) values larger than 25 meters may be used.
Note that the actual shadowing in meters s and CPARM(4) are related
by: s = D - CPARM(4).  In the default case (CPARM(4) = 0 or 25.0),
the shadowing is determined from the archive tape.  In all other
cases, the shadowing is determined by FILLM itself.  In this case,
there may be shadowed antennas that are not found by FILLM, because
it knows nothing about antennas which are in different subarrays
from the one it is checking for shadowing.  The default case does
not suffer from this problem, as the on-line check for shadowing
includes all antennas, regardless of subarray (including even the
master pad!).

CPARM(5): VLA archive tapes may contain data for a given band with
different numbers of spectral channels at different times.  Such data
must be kept in separate files.  CPARM(5) is used to specify which set
of data is desired by specifying the number of channels.  If CPARM(5)
<= 0 then the first set encountered will be used.  The default will
have the desired action in almost all practical situations.
     The value of CPARM(5) to select a given configuration of
channels is given by the following:
channel code = 4096*log2(NCHA) + 256*log2(NCHB) + 16*log2(NCHC)
                                                   + log2(NCHD).
Where NCHA = number of channels in IF A if less than 256, 0 if =
256 or 2 if = 512.  If the number of channels is larger than or
equal to 256 the entire value is to be made negative.  NCHB,
NCHC and NCHD are the number of channels in IFs B, C and D.
This is how the value is stored on the tape.

     Since only Ken Sowinski can predict the values for unused
IFs, it is best not to try to guess it, but instead to run task
PRTTP which will give the values on the tape.

CPARM(7): FQ numbers are assigned to data whose frequency
characteristics differ from previous data, i.e., bandwidths are
different, a different sideband is recorded, or the frequency itself
differs by an amount greater than the maximum Doppler shift possible
(i.e., a difference greater than that caused by a velocity change of
30km/s).  It is also useful for the user to have some control over
this selection so CPARM(7) allows the definition of a frequency
tolerance.

     In continuum, setting CPARM(7) <= 0.0 will force the
assumption of a 100 kHz difference.  This means that, by
default, all different frequencies will have different FQ
numbers.  If CPARM(7) > 0 then the value is taken to be the
desired frequency tolerance in kHz.

     In spectral line, CPARM(7) < 0 will result in all data having
the same FQ number.  If CPARM(7) = 0 then the frequency tolerance will
be half of the maximum frequency difference due to the Earth's
rotation for sources 180 degrees apart (i.e., 10**-4 * observing
frequency).  If CPARM(7) > 0 then the value is taken to be the desired
frequency tolerance in kHz.

     An example: if the user had observed the 1665 and 1667 MHz
line of OH and wanted just to separate the two so he/she could
process all 1665 MHz data in one pass, then CPARM(7) should be
set to 500.0, i.e., 500kHz; this allows a window of 1.0 MHz, and
anything that lies inside the first frequency found +/- 1 MHz
would be assigned the same FQ number.

CPARM(8): A CL table will be written with a time interval between
entries of CPARM(8) min.  0 defaults to 5 min.

CPARM(9): VLA archive tapes may contain several files.  Using CPARM(9)
the number of files read can be controlled.  The default (0) is to
keep reading the entire tape; this may produce unpredictable results
if there are other types of files following the true archive records.

CPARM(10): The averaging time for calibrators may be specified
differently than for non calibrator sources.  Any source with a
non-blank calibrator code is considered a calibrator for this
purpose.  It may be desirable for calibrator averaging times to be
shorter that for program sources if the averaging time of the program
sources is relatively long and the calibrator scans are short.  This
allows better editing of the calibrator data.

DPARM:
    If the desired frequencies need to be specified more
accurately than observing band, say for frequency switching in a
given band, the frequencies of the A, B, C, and D "IFs" of the
VLA can be specified in DPARM.  Since only single precision
values are passed to FILLM by AIPS, two words are provided to
specify the frequency.  The frequency used will be the sum of
the two, e.g., if DPARM(1) = 1665.0E6 and DPARM(2)=0.33333E6
then the requested frequency for the A "IF" is 1665.33333 MHz.
C and D frequencies default to the A and B values.
    The maximum allowable deviation from the specified values are
given by DPARM(9 & 10).  Example:
    DPARM= 1420e6,0,1670e6,0,0,0,0,0,1e6,1e6
will cause the selection of data when the AC IF-pair has a
frequency between 1419 and 1421 MHz AND the BD IF-pair has a
frequency between 1669 and 1671 MHz.  All other data will be
ignored.

BPARM:
   The BPARM adverb array is used to specify corrections made to the
initial CL table attached to the data.  These corrections currently
include that for opacity and antenna gain curve, both as a function
of elevation and time.  The defaults (BPARM(n)=0) will get you our
current best estimate for these two corrections.  To turn these
corrections off completely, set BPARM(1)=-1; BPARM(2)=-1.  This will
get you behaviour similar to FILLM pre-31DEC01.  Note that if you
forgot to turn these corrections off, and don't really want them,
there is no need to run FILLM again - simply use EXTDEST to get rid
of CL table version 1 (that created by FILLM), and run INDXR on the
dataset.

BPARM(1): Determines if an opacity correction is written into
the CL table or not.  There are 3 possibilities for this:
 BPARM(1) < 0       ->  do not do any opacity correction in the
                        CL table.  This is what FILLM used to do
                        pre-31DEC01.
 BPARM(1) = 0       ->  equivalent to setting BPARM(1)=20, BPARM(10)=.5
                        (see description below).
 0 < BPARM(1) <= 10 ->  do an opacity correction, and take
                        BPARM(1) as the zenith opacity.
 BPARM(1) > 10      ->  do an opacity correction, and determine the
                        zenith opacity as a function of time from a
                        combination of the day of year (a seasonal
                        model) and the surface weather data on the
                        archive tape (a weather model).  In this
                        case, the zenith opacity is determined by:
                          tau = 0.0       - P-, and 4-bands
                          tau = 0.01      - L-, C-, and X-bands
                          tau = ww * fw(Ts,Td) + ws * fs(doy)
                                          - U-, K-, and Q-bands
                        where fw is the value of the weather model
                        opacity (given surface ambient temperature Ts,
                        and surface dewpoint temperature Td), and fs is
                        the value of the seasonal model opacity (given
                        day of year doy).  ww and ws are the relative
                        weights for the two opacity models, and sum to
                        1.  See VLA Scientific Memo XXX (in preparation,
                        B. Butler) for a description of fw(Ts,Td) and
                        fs(doy).  When BPARM(1)>10, the value of ww is
                        given in BPARM(10) (and ws = 1-ww).  To get
                        an opacity determined only by surface weather
                        measurements, set BPARM(10) = 1.0.  To get
                        an opacity determined only by season/day of
                        year, set BPARM(10) = 0.0.  To get a straight
                        average (the default behaviour), set
                        BPARM(10) = 0.5 (or, equivalently, set
                        BPARM(1) = 0).
   Note that in the case where 0 < BPARM(1) < 10 (user specified
opacity), this opacity value is used for *all* data, regardless of
frequency/band.  There is currently no mechanism for specifying
different opacities for different bands.  If you have multiple
frequencies in the same subarray, it is strongly recommended that you
either let the opacity be calculated (by setting BPARM(1) > 10, or
BPARM(1) = 0), or run FILLM in multiple passes, using the BAND adverb
to select the proper band, and setting BPARM(1) to the proper value for
each separate band.

BPARM(2): Determines if a gain curve correction is written into
the CL table or not.  There are 4 possibilities for this:
 BPARM(2) <= 0      ->  do not do any gain curve correction in the
                        CL table.  This is what FILLM used to do
                        pre-31DEC01.
 0 < BPARM(2) < 2   ->  do a gain curve correction, with coefficients
                        which vary as a function of band and antenna.
                        These coefficients are read from the file
                        specified in CALIN.
 2 <= BPARM(2) < 3  ->  do a gain curve correction, with coefficients
                        which vary only as a function of band.  These
                        coefficients are read from the file specified
                        in CALIN.
 BPARM(2) >= 3      ->  do a gain curve correction, with coefficients
                        specified in BPARM(3) through BPARM(6) for all
                        bands and antennas.
The default (BPARM(2)=0) is to have coefficients which vary as a
function of both band and antenna.
   The gain curve looks like:
       G = c0 + c1 * ZA + c2 * ZA*ZA + c3 * ZA*ZA*ZA
where ZA is the zenith angle, and c0, c1, c2, and c3 are the
coefficients.  So, in the case where you want to specify the
coefficients in BPARM explicitly, c0 = BPARM(3), c1 = BPARM(4),
c2 = BPARM(5), and c3 = BPARM(6).  Note that what is written into
the CL table is the gain curve correction, which is 1/G.
   Note that caution should be exercised when applying gain curve
corrections to data taken prior to 2001.  This is because no careful
record was kept of how the gain curves changed over time (e.g., as a
result of re-setting the panels or optics).  Gain curves were
determined approximately once per year during the period 1995-2001.

CALIN:
The name of a file to provide the antenna gains used to populate the
initial CL table.  If CALIN is left blank, it defaults to a system
file located at: 'AIPSIONS:VLA.GAINS'.  The file is in free format
with fields: a 1-character band code, antenna number, begin date in
form YYYYMMDD, end date YYYYMMDD, and 4 gain curve coefficients.
Look at the default system file for guidance.

SOME EXPLANATION ABOUT TIME STAMPS

Time stamps on the MODCOMP tapes mark the end of each integration
interval. When these tapes are read by FILLM, time stamps remain
unchanged for CPARM(1)=0 (no off-line integration). When the online
integration time is t, the time stamps look as follows:

     |_______|_______|_______|_______|

     0       t       2t      3t      4t

This simplicity is gone when CPARM(1) is some multiple of the online
integration time.  For time stamps in AIPS uv data, the following
definition holds:

             ts = tc + 0.5t                                     (1)

ts = time stamp assigned by FILLM

tc = time at the center (midway) of the integration. In the example
     above, tc would be 0.5t, 1.5t, 2.5t, etc.

This definition coincides with the end-of-integration definition when
FILLM does not apply off-line integration. The origin and the
justification of this definition lies back in the gray past. One
advantage of this definition is that the new time stamp after
averaging is the average of the old time stamps (which is not the case
if the definition of 'end of integration' were used).

If we assign an off-line integration in FILLM twice the online
integration time, the picture above looks like:

     |_______________|_______________|

     0               2t              4t

tc now becomes:      tc = t, 3t, ...

and ts:              ts = 1.5t, 3.5t, ...

Which does not follow the end-of-integration definition anymore (in
which case ts obviously would have been 2t and 4t). Note that these
time stamps are purely fictitious: they are not used for calculating
anything; after all, the u's, v's, and w's are read from tape
independently.

One case in which these times become important is when using UVFIX. In
UVFIX the user can specify the clock correction (to be subtracted from
each time stamp to obtain the true center time) in UVFIXPRM(13). For
VLA data, this correction is one-half the online MODCOMP integration
time t.


EXECUTION TIMES:
     On an otherwise unloaded (?) Convex C1 with 120 IPS tape
drive, reading 125,000 vis with 30 second averaging (4 hours of
continuum data) took 290 cpu sec and 600 real sec.  The data
was split between two tapes and 10 sec integrations were
recorded on the tape.


DISK SPACE:
     The size of a data set loaded via FILLM with DOUVCOMP=-1 is
approximately:

 # vis. * (6 + (# poln[1,2 or 4]) * (# freq) * (# IF) * 3) / 256

in units of 512 AIPS-byte (1024 byte on most systems) blocks.
The granularity of the disk may increase this somewhat.  Also
CL, SU and NX tables will require additional disk space.

     If DOUVCOMP=1 then the disk space needed for the main data
file becomes:

 # vis. * (8 + (# poln[1,2 or 4]) * (# freq) * (# IF)) / 256

This represents a disk savings of a factor of 1.9 for 2 IF
continuum data or nearly a factor of 3 for large line data sets.
The sizes of the tables are unaffected by DOUVCOMP.

AIPS