As of Sat Jan 20 15:38:36 2018

ATLOD: Task which reads RPFITS format uv data into AIPS.


INTAPE            0.0       9.0    Tape drive number (0 -> 1)
                                   No REMOTE capability
NSKIP    *        0.0              Number of tape files to
                                   skip (0 -> 0)
NFILES   *        0.0    1000.0    Number of tape files to
                                   load (0 -> 1)
BCOUNT   *        0.0              First scan to read (0 -> 1)
NCOUNT   *        0.0              No. scans to read (0 -> all)
DATAIN                             Disk file name
OUTNAME                            Output uv file name (name)
                                     blank -> source name
OUTCLASS                           Output uv file name (class)
                                     blank -> UVDATA
OUTSEQ            0.0    9999.0    Output uv sequence number
                                     0 -> highest unique AND a
                                     new file for EACH pass
OUTDISK           0.0       9.0    Output disk drive number
OPTYPE                               'LOAD' load data (default)
                                     'LIST' scan summary
                                     'SUMM' file summary
                                     'SYSC' dump syscal info
                                       see CPARM(9)
SOURCES  *                         Source selection list
TIMERANG *                         Time range selection
FREQSEL  *                         Frequencies to select (MHz)
IFSEL    *                         IFs to select
CHANSEL           0.0     8192.    Channel selection array:
                                     (1,n) start channel group n
                                     (2,n) end channel group n
                                     (3,n) incr. channel group n
IFMAP                              True to map on-line IF chain
                                   to AIPS IF axis.
NIF               0.0        4.0   Length of IF axis,
                                   0 => ATLOD decides.
DOUVCOMP         -1.0       1.0    Use compressed data?
APARM                              1) If > 0, convert to Stokes
                                      if < 0, kludge RR,LL,RL,LR
                                      if = 0, don't change the
                                   2) If true, keep flagged data
                                   3) If true, retain
                                   4) Shadowing diameter (m)
                                   5) max NX time gap (min)
                                       0 -> 10 min
                                   6) max NX scan length (min)
                                       0 -> 60 min
                                   7) CL entry interval (min)
                                       0 -> 5 min
                                   8) Number of integrations
                                      over which to average Tsys
                                       0 -> 1  See HELP/EXPLAIN
                                   9) Number of integrations
                                      over which to average XY
                                      phase diff if APARM(1) = 1
                                       0 -> 1  See HELP/EXPLAIN
                                   10) If non-zero drop data on
                                      quality of XY phase diff.
                                      See HELP/EXPLAIN
BPARM                              Baselines to be rejected,
                                   (see HELP)
CPARM                              Further control parameters:
                                   1) System cal. info. print
                                      level. See HELP/EXPLAIN
                                   2) If true negate system XY
                                      cal phase difference
                                   3) If true, negate phases
                                   4) If true, use XYPHASE array
                                      instead of on-line values
                                   5) If true, apply XY phase to
                                      the Y gain
                                   6) If 1, report antennas for
                                      which sampler stats bad
                                      If 2, drop baselines with
                                      bad sampler statistics
                                      If 3, do 1 and 2
                                      If 4, correct data for
                                      bad statistics.
                                   7) Sampler tolerance
                                   8) If true, DATAIN will be
                                      read as lower case.  This
                                      is intended for reading
                                      from VAX disks mounted
                                      under unix via multinet.
                                   9) If OPTYPE='SYSC' then the
                                      3 text files contain:
                                     1 => XY phase, Tx & Ty
                                     2 => X sampler stats
                                     3 => Y sampler stats
                                     0 => 1
                                   10) If 1, Tsys from IF 1 is
                                     applied to IF 2, if
                                     APARM(8) is > 0
XYPHASE                            Enter XY phase difference for
                                   each antenna & FQID (degrees)
DPARM                              1)  If > 0, then the scan
                                     based source number of all
                                     sources is set to 1.
                                   2) If > 0, the spectra are
                                     3-point Hanning smoothed.
                                   3) If > 0 -> set weights to
                                     D(3) instead of integration
                                     time in units of 15s


Type:  Task
Use:   To read and load or list a summary of an RPFITS format uv
       data file into AIPS.  The RPFITS file can be on tape or
       disk.  If OUTNAME, OUTCLASS, and OUTSEQ specify an
       existing file the data will be appended to it.

INTAPE.....Tape drive number.  0 -> 1  Note that ATLOD does not
           function with the AIPS remote tape system.
NSKIP......Number of tape files to skip from current tape
           position before starting to load data.
NFILES.....The number of tape files to load (see OUTSEQ)
BCOUNT.....The first scan to read (same in each file).  0 -> 1
NCOUNT.....The number of scans to read (same in each file).
           0 -> read all scans from file
DATAIN.....Name of an RPFITS format uv data disk file to be
           loaded into AIPS.  If this name is blank, ATLOD
           assumes that input is to come from tape as
           specified by INTAPE, and NFILES. If DATAIN is not
           blank, INTAPE and NCOUNT are ignored.
OUTNAME....Output uv file name (name).  Standard behavior with
           default is source name.
OUTCLASS...Output uv file name (class).  Standard behavior
           with default is 'UVDATA'.
OUTSEQ.....Output uv file name (sequence number).
           0 -> highest unique.  If NFILES > 1, all the tape
           files go into one output file.  To make different
           tape files go into different  AIPS files, use
           successive applications of ATLOD and NFILES=1.
OUTDISK....Output uv file disk drive number.
OPTYPE.....'LOAD' to read the data into an AIPS file (default)
           'LIST' to list a summary of each scan
           'SUMM' to list a summary of each file (very brief)
           'SYSC' to write XY phases and Tsys values into text
                  files in the FITS area
SOURCES....List of sources to select.  ' ' -> all.  A "-" before
           ANY source name means ALL EXCEPT any source named.
TIMERANG...Time range selection, 0 -> all.
FREQSEL....Frequencies of data to be selected (MHz), 0 -> all.
IFSEL......IFs of data to be selected, 0 -> all.
CHANSEL....Channel selection array: 0 => Load all channels
           Up to 10 groups of channels may be selected.
           CHANSEL(1,n) selects the begin channel, CHANSEL(2,n)
           selects the end channel, and CHANSEL(3,n) selects the
           channel increment for group n.  Generally, unselected
           channels are loaded but flagged.   However, if only
           one group of channels is specified, channels are
           dropped rather than flagged.
IFMAP......If true, then the IF chain number defined by the on-line
           system is mapped to the AIPS IF axis value one-to-one
NIF........Length of IF axis.  0 -> ATLOD chooses
DOUVCOMP...If true (DOUVCOMP >= 0) the output data is written
           in compressed format which can result in a
           substantial reduction - about a factor of 3 - in disk
           space needed, but some tasks may not be able to
           handle this format.  Data are stored as 16-bit
           integers with a different scaling for each visibility
           spectrum.  ATLOD will warn you if there is a loss of
           accuracy due to some channels (interference?) being
           much higher than the average.
APARM......Control parameters:
           1) If > 0, convert XX,YY,XY,YX to Stokes IQUV.
              If < 0, ignore the actual polarization order and
                      type and call them RR,LL,RL,LR.
              If = 0, don't change the polarizations.
           2) If true, retain data flagged on-line as bad.
           3) If true, retain autocorrelations in the output
              uv file.
           4) Antenna diameter to adopt when rejecting data due
              to shadowing effects.  The physical diameter of
              the AT antennas is 22 meters, so setting APARM(7)
              smaller than 22 relaxes the shadowing constraint,
              while making it larger tightens it.
              The default is 22m.
           5) Start a new index entry if a gap longer than
              APARM(5) minutes occurs in the scan.
              The default is 10 min., mininum 1 min.
           6) Start a new index entry if an uninterrupted
              section of data on the same source longer than
              APARM(6) minutes occurs in the scan.
              The default is 60 min., mininum 1 min.
           7) A new CL (calibration) table will be created with
              empty entries every APARM(7) minutes.
              The default is 5 min., mininum 1 min.
           8) If = -1 Undo the on-line correction and redo with
                      nominal 50 K system
              If = 0  No additional corrections
              If > 0  Undo the on-line correction and redo with
                      a Tsys correction averaged over APARM(8)
           9) If APARM(1) = 1 then Stokes conversion is
              requested and the XY phase difference per antenna
              is used.
              If A(9) = 0 no averaging of  XY phases us done
              If A(9) > 1 average XY phases for A(9)
          10) If A(10) > 0, drop visibilities if one of the
              antennas has a bad XY phase differences.
              If A(10) < 0, apply additional absolute XY phase
              tolerance as well as usual clipping algorithm.
              See EXPLAIN for clear explanation!
BPARM......Baselines to be rejected, specified as 10*I + J
           where I and J are the antenna numbers, order
           being irrelevant unless the value is negative.
           If I is 0 all baselines containing antenna J will
           be dropped.  Examples
              12  Drop baselines 1-2 and 2-1
             -31  Drop baseline  3-1 (but not 1-3)
               4  Drop all baselines involving antenna 4
CPARM......Control parameters for experts only.
           1) If = 1 list system calibration information
              If = 2, you are told when the XY phases or Tsys
              values appear to have drifted or jumped.
              If = 3, then the information from 1 + 2 is printed
           2) If true negate the system XY phase difference
           3) If true, negate the visibilty phase as it is
              read in (to fix data taken in 1989).
           4) If true, use the XYPHASE array rather than the
              on-line values
           5) If true, apply the XY phases to the Y gains.
           6) If 1, report antennas for which the sampler
                statistics are out of range
              If 2, drop baselines involving antennas with
                bad sampler statistics
              If 3, do 1 & 2
              If 4, correct data for bad sampler statistics.
           7) If > 0, use this as the sampler statistics
              tolerance instead of the internal default
              (5 percent for C(6)=1,2,3  12 percent for C(6)=4)
           8) If true, DATAIN will be read as lower case.
              This is intended for reading from VAX disks
              mounted under unix via multinet.
           9) If OPTYPE='SYSC' then the text files contain:
               1 => XY phase, Tx & Ty
               2 => X sampler stats
               3 => Y sampler stats
               0 => 1
           10) Apply Tsys from IF 1 to IF 2 if (A8) > 0
XYPHASE....Array of XY phases (degrees) for each antenna
           and FREQID.  See EXPLAIN
DPARM......1) If > 0, all sources are given source number 1.
              This is for mislabelled single source per scan
              data taken in March 1992 following intial
              mosaicing experiments.
           2) If > 0, the spectra are 3-point Hanning smoothed
           3) By default, the weight of each visibility is set
              to the integration time in units of 15 sec.
              If DPARM(3)>0 then the weight for each visibility
              is set to DPARM(3)


ATLOD: Task to read RPFITS format uv data into AIPS.
DOCUMENTERS:  Mark Calabretta and Neil Killeen

   The uv file produced by ATLOD is in the multi-source format,
that is, it includes the SOURCE and FREQSEL random parameters,
and has SU, FQ, and NX extension files associated with it.  The
SU extension file contains information which associates each
source with a set of identifiers (name, ra, dec, etc.), and
other information.  The FQ extension file records information
for each IF used (frequency, bandwidth, sideband, etc).  The NX
file records the index of visibility records associated with
each scan.

BCOUNT,NCOUNT    Each RPFITS file has multiple scans.  These
   specify the first scan and number of scans to read.  If
   reading from tape, then these numbers are the same for each
   file.  Thus, they probably useful only for single file input.

DATAIN   If blank, ATLOD assumes that input is to come from
   tape as specified by INTAPE, and NFILES.  Use the AIPS verb
   AVFIL to skip files on the tape to reach the one of interest.
   If DATAIN is not blank, INTAPE and NCOUNT are ignored. The
   DATAIN, when used for disk file input, must be of two forms:

   1) 'AREA:FILENAME'  where AREA is an upper case environment
      variable (e.g.,  percentsetenv AREA /scratch/data) pointing
      to the location of the file called FILENAME. The file
      name must also be upper case.

   2) 'FITS:MYDIR/FILENAME'  FITS is a pre-defined environment
      variable and MYDIR is a directory (upper case) below the
      FITS area.  This can be shortened to 'MYDIR/FILENAME'
      and the FITS area will be assumed as the route.  Note
      if you use DATAIN='SCRATCH/FILENAME' then the files are
      assumed to reside in /scratch/aips which is symbolically
      linked to the directory FITS/SCRATCH.

   3) We recommend that you simply put your data in the FITS
      area, and then DATAIN='FILENAME' will work by default.

OUTNAME, OUTCLASS, OUTSEQ, OUTDISK  Specify the output file
   name.  If they specify a file that already exists, then the
   new data are appended to it.

   'LOAD' means load the specified data into an AIPS UV file
   'LIST' resticts ATLOD to reporting summaries of the contents
          of each scan
   'SUMM' restricts ATLOD to reporting summaries of each file
   'SYSC' causes ATLOD to write text files in the FITS area
          The contents depend upon what value you give
          CPARM(9).  By default, you get files containing
          the XY phases (degrees) for each antenna  (XYPHS_UID)
          and the system temperatures (TSYSX_UID) and TSYSY_UID)
          where UID is your AIPS user number in decimal.

          The first column in each file is the UT in days.
          The second column is the frequency.
          An integer follows each quantity for each antenna.
          A zero indicates that value is good, anything else
          indicates that that value was flagged by the on-line
          system (e.g. if the telescope was not on source).

          If you have multiple frequencies in the data, you can
          use FREQSEL and IFSEL to select only one, otherwise data
          from all frequencies will be put into the one file.
          Note however, that the utility program ``pltsys''
          provided at the ATNF for plotting the text files
          produced by the 'SYSC' option, has frequency selection,
          and it is quicker to load all frequencies with ATLOD and
          then select the ones you want to plot with ``pltsys''.

          Use SOURCES to select the desired sources (note that
          Tsys will vary from source to source).

          Use TIMERANG to select by time.

          All other selection parameters are inactive in this mode

          See CPARM(9) for more details of other possible
          output files.

FREQSEL   Allows you to select a number of frequencies. They
   should be specified in MHz.   This apples to simultaneous
   and time cycled frequencies. They are matched to the actual
   frequencies in the data with a 1 KHz tolerance.

   If you select frequencies that come from data with different
   numbers of channels or polarizations, they cannot all go in
   the one AIPS file. ATLOD will discard any frequencies that
   are inconsistent with the attributes of the AIPS uv file.
   You will be notified as this happens.  You will have to run
   ATLOD as many times as you need output files to deal with
   this sort of data.

IFSEL     Allows you to select a number of IFs. The combination of
   FREQSEL and IFSEL should offer sufficiently flexible selection criteria
   to deal with all cases.    The ATCA produces one or two IFs.
   Generally, spectral-line observers will use 1 IF, and continuum
   observers two.    It is possible, although not recommended, to
   observe both IFs at the same frequency.  If your correlator setup
   was such that you had different numbers of channels in the two
   IFs, then the two IFs would need to go into separate AIPS files.
   In this case, FREQSEL would be insufficient to discriminate between
   the two IFs and you would need IFSEL to pick out one or the other.

   It may also be useful if you are time switching frequencies
   but only want one IF to be loaded.  With just FREQSEL you
   would need to list all the desired frequencies, with IFSEL
   you just select the IF number, 1 or 2.

   Data are selected accroding to FREQSEL *and* IFSEL.

CHANSEL   Channel selection array: 0 => Load all channels. Up to
   10 groups of channels may be selected.  CHANSEL(1,n) selects
   the begin channel, CHANSEL(2,n)  selects the end channel, and
   CHANSEL(3,n) selects the  channel increment for group n.  If
   all of CHANSEL(1:3,n) are 0, then no more groups are accepted.
   Otherwise, the defaults are CHANSEL(1,n)=0 -> 1, CHANSEL(2,n)=0
   -> NCHAN, CHANSEL(3,n)=0 -> 1.

   Generally, unselected channels are LOADED but FLAGGED.

   An important case in point  is when you also set DOUVCOMP = 1
   which means the visibilties are stored as scaled 16 bit integers.
   If you have a strong birdie in a certain channel, you can deselect
   it with the CHANSEL array.  E.g., the bad channel is 10.  Set
   CHANSEL=1,9,1,11,33,1 to load channels 1-9  and 11-33. What this
   really means is that all channels are loaded but channel 10 is
   flagged.  In addition, channel 10 is not used to work out the scale
   factors for the visibilty on conversion to 16 bit integers, so that
   your dynamic range is not compromised in the good channels.

   Note that if only one group of channels is specified, channels
   are actually  dropped rather than flagged.  Thus, CHANSEL=7,29,2
   would load only the odd channels from 7 to 29.   In the FQ table,
   you would see that the channel increment has been doubled to
   reflect this selection. However, the total bandwidth would reflect
   only the actual channels that were loaded.  Really, there should
   be a channel separation AND a channel increment, but AIPS does
   not support this.   You will also note in the header (IMHEAD)
   that the frequency reference pixel will also have changed
   to reflect the new size of the frequency increment.

IFMAP     AT data may have up to four simultaneous frequencies, but
   generally you will see only two.  Let us call these ``IF chains''.
   Contemporaneous data is located within the AIPS uv file on the
   IF axis (if the number of channels and polarizations is equal).
   When IFMAP is true, (+1) then IF chain 1 will ALWAYS turn up on
   the IF axis in location 1, and IF chain 2 at location 2.  However,
   you might select, with FREQSEL,  only the IF chain 2 frequencies.
   In this case, you would want them to go to IF axis location 1,
   otherwise you would waste a lot of disk space by having an empty
   IF axis location 1,  followed by your data in IF axis location 2.
   When IFMAP is false (-1), then ATLOD will put the next frequecy
   it encounters in the next available IF axis location, regardless
   of the on-line IF chain that it came from.

   I suggest generally you leave IFMAP as false (-1).   See the
   example after the discussion of NIF below for an example
   where you might set it to true (+1).

NIF      Normally, ATLOD will work out the length of the IF axis
   based upon the information present in the first scan header that
   is consistent with the user's selection crteria.   However, it
   may be you know more than ATLOD and need to tell it the length
   of the IF axis.

   Consider the case where you have an observation such as

               IF chain 1       IF chain 2
   Scan 1:        4700              8300
   Scan 2:        4600              8000

   Let us say you set FREQSEL=4700,4600,8000, thus omitting
   the 8300 frequency on IF chain 2.  Now, you MUST set NIF=2
   to get all these frequencies loaded.  ATLOD notes that you
   want one frequency from the first scan, and that it would
   go to IF axis location 1.  But this is the only information
   it has when it reads the header for the first scan, which
   is when it must build the attributes for the output file.
   It doesn't know that the other specified frequencies are
   is succeeding scans and that they will need an IF axis of
   length 2.   Thus, you must set NIF=2.   IF axis location
   2 for the first scan will get filled with dummy visibilities
   of zeros.

   Note, that if you set FREQSEL=8300,4600,8000 instead, and
   IFMAP=-1, then the 8300 frequency would be put onto IF axis
   location 1.   However, if IFMAP=1 then it would go to
   location 2, and location 1 would be zero filled instead.

DOUVCOMP   For large data sets, you may wish to compress the
   data format.  This saves disk space, but loses dynamic range.
   The task UVCMP can switch between compressed and uncompressed
   formats.  Uncompressing compressed data does NOT recover
   that lost dynamic range.

APARM(1)  If > 0, convert XX,YY,XY,YX polarizations to Stokes
   IQUV.  The format of raw AT data is linear polarizations.
   At some stage this must be converted to the Stokes parameters
   IQUV for imaging.  Because AIPS does not deal very well with
   XX,YY,XY,YX data at present, and for other reasons too,
   we encourage you to use this option and convert to IQUV.

   Note that the XX and YY gains must be roughly equal (< 10 percent)
   for this to work properly if any of your sources are
   polarized.  Check this by loading an unpolarized and strong
   source (e.g. 1934-63) with no Stokes conversion and compare
   plots of XX and YY for individual baselines (to do this you
   need APARM(1)=-1.  See below.

   If < 0, ignore actual polarization order and type and call
   them RR,LL,RL,LR.  This is for cases where you actually want
   to load the data in their raw form, but in order to get many
   of the plotting functions in AIPS to work, we need to trick
   AIPS and call them something it knows how to recognize.
   Calibrating XX and YY independently assumes that the
   calibratirs are unpolarized.

   APARM(1)=0 means load the data as they are with no Stokes
   conversions or naming tricks.

   See also APARM(10) for a bit more fiddling about.

APARM(2)  If > 0 then retain all visibilites that the on-line
   system flagged as bad.  These usually occur for data taken
   when the system is not yet on source, or for errors detected
   by the correlator.  You would want to be desperate for
   visibilites to turn this one on.

APARM(3)  If > 0 then autocorrelations are retained.  At
   present, these contain nothing useful and there is no point
   to hanging on to them.  When they become useful, the default
   will be change to retain them.

APARM(4)  The shadowing criterion is currently fairly simple and
   is a geometric calculation for sources at the pointing
   centre.  For a given shadowed baseline, the shadowed antenna
   is found, and all visibilities involving that antenna
   dropped.  Note the EW nature of the ATCA is used for
   the shadowing calculation, so you must disable the shadowing
   check for LBA data (see below).

   If = 0 the shadowing is computed for 22 m diameter dishes

   If > 0 the shadowing is computed for A(4) m diameter dishes

   If < 0 the shadowing check is disabled

   Increasing A(4) above 22 provides as more conservative

APARM(5)  The index (NX) table contains information about
   where scans begin and end (times and visibility numbers).
   This allows rapid access to the data base.  A new index
   record is always started for a source change.

   If = 0 then begin a new index entry if a gap longer than
   10 minutes occurs in the data.

   If > 0 then then begin a new index entry if a gap longer than
   A(5) minutes occurs in the data.  Minimum allowed is 1

APARM(6)   If = 0 then start a new index entry if an
   uninterrupted section of data on the same source is longer
   than 60 minutes.

   If > 0 then start a new index entry if an uninterrupted
   section of data on the same source is longer than A(6)
   minutes.  Minimum is 1 minute.

APARM(7)   ATLOD will also create the pristine calibration (CL)
   table.  It consists of 1s (Real part of gain) and 0s
   (imaginary part of gain).  Later, as the calibration
   procedes, new CL tables will be created by applying
   corrections to CL table 1.  The frequency of CL table entries
   sets the fundamental time scale on which you can make gain
   corrections, when you work entirely within multi-source
   files.  That is, when doing self-calibration, it would be
   pointless having solution intervals smaller than the CL table
   entry interval.  If you split off single source files, the
   fundamental gain change time scale that you can add is the
   integration time (10 seconds).  Self-cal is often done on
   time scales of a couple of minutes or maybe less (depends on
   the S/N), so you should bear this in mind.

   If = 0 then a new CL table entry will be created every 3

   If > 0 then a new CL table entry will be created every
   APARM(7) minutes.  The minimum is 1 minute.

APARM(8)  The Tsys correction attempts to correct for gain
   changes that occur faster than you observe calibrators.  It
   is measured every integration by comparing the switched noise
   source of known power with the total power for each X and Y
   feed on each antenna.  It is this number that is used to
   normalize the signals before they are transformed into
   digital signals in the samplers -- the samplers require input
   signal levels that are steady.  The correlation function is
   the rescaled by Tsys.  Since Tsys is worked out from a single
   integration it is noisy.  It is preferable to apply a less
   noisy Tsys correction.

   If APARM(8) = -1, the the on-line correction, if any, is
   undone, and the Tsys correction is redone with a nominal
   50 K system.  This is the same as no on-line Tsys correction

   If APARM(8) = 0, no additional changes are made to the Tsys

   If APARM(8) > 1, then APARM(8) integrations will be used to
   determine the running Tsys mean.

   Some data are too old for the offline Tsys correction
   (earlier than 13-Aug-90), you will be warned if this is the
   case.  The algorithm attempts to deal with frequency and
   source changes (Tsys may jump if the source is bright enough)
   as well as unexpected drifts or jumps.

APARM(9)  If you have requested Stokes conversion (APARM(1)=1)
   then the phase difference between the X and Y feeds on each
   antenna is used.  Essentially, XX and YY must be summed in
   phase to form Stokes I.  These XY phases are noisy and
   APARM(9) allows an averaged value to be used.

   If APARM(9) = 0  then no averaging of the XY phases is done
                    and no checks for bad values are made
   If APARM(9) > 1  then the XY phases are averaged over APARM(9)

APARM(10)  The XY phase differences for each antenna are
   measured from a noise diode every integration.  Dropouts in
   normal AT visibilities generally correlate with dropouts in
   the XY phase difference for at least one of the antennas
   involved in the baseline.  If your source is very weak you
   cannot detect these dropouts by visual inspection.  However,
   the XY phase difference is always measured.

   If > 0 then visibilities are dropped when the XY phase
   difference is bad. The XY phase is deemed bad by comparing
   it with the running mean and the width of a stack of
   accumulated phases (see APARM(9)).

   If < 0, then in addition to first being marked as bad by the
   algorithm which compares the  current point with the stack
   mean and width (see above), the point is finally only called
   bad if it is greater than ABS(APARM(10)) DEGREES from the
   stack mean as well.   Note that if you are using the XYPHASE
   array (CPARM(4)=1), then the values in it are used as the
   stack mean instead of the on-line values.  This provides for
   a simple means of flagging on the basis of XY phases if you
   have gone to the trouble of plotting them up.

   If = 0 then no dropping of visibilities is done.  If you have
   also selected to convert to Stokes, ATLOD will replace what
   it thinks are bad XY phases with the stack average (see
   APARM(9) and below).  However, visibilities involving the
   antenna with the bad XY phase are likely to be bad.

CPARM(1)  If = 1 then lots of information can be printed that
   is used in the conversion to IQUV.  XY phase difference,
   Tsys and parallactic angles are listed.

   If = 2, then the user is told when the XY phases or Tsys
   values appear to have drifted since the last freq
   or source/freq combination, respectively, was observed.
   The accumulations of these quantities are reset when this
   happens.  Otherwise, the current accumulations would reflect
   the wrong values.

   If = 3, then the information from 1 + 2 is printed

CPARM(2)  If the conversion to IQUV does not look correct, try
   turning this on AND come and talk to me (NEBK).

CPARM(3)  If true, negate the visibilty phase as it is read in.
   There was a problem with early AT data sets which produced
   inverted images.  The phase had the wrong sign, so you can
   change it here.  Only data from 1989 is likely to be

CPARM(4) Use the XYPHASE array rather than the on-line values.

CPARM(5) When not converting to Stokes IQUV, you can optionally
   apply the XY phase to the Y gains.  Thus, the YY phases would
   be rotated to those of XX

                     ******* NB ********

   A trap here is that applying XYPHASEs of zero degrees, is different
   from NOT applying any XYPHASE.  Thus, let us say you use the 'SYSC'
   option to plot up your XYPHASEs, and you see that they are all
   close to zero.  However, because, internally, ATLOD adds pi to the
   XYPHASEs (to bring them to the same convention as Miriad), applying
   an XYPHASE of zero is different from  not applying them.   If you
   want to apply the XYPHASEs, you should ALWAYS do it, even if they
   are apparently close to zero.

CPARM(6) The sampler statistics for each antenna have well
   defined ideal values.   The correct values are 50 percent for the
   zero level statistic.   For the positive and negative level
   statistics, they should be 17.1 percent for data taken before
   21/8/93, and 17.3 percent thereafter.

   If the samplers are out of range, it indicates that the
   data will be scaled incorrectly, although the spectrum
   may look fine.  A tolerance of about 5 percent is about all that
   one should tolerate on these statistics.

   A particular case is that of interference, which may drive
   the samplers to their rails.  This would produce
   uncalibratable data but would be hard to flag otherwise.

   If = 1, then when the sampler statistics are out of
     range, a warning is issued to the user.  The default
     is that the samplers must be within 5 percent of the correct

   If = 2, then baselines involving antennas with out of range
     sampler statistics are dropped.  The default is that
     the samplers must be within 5 percent of the correct value.

   If = 3, then 1 and 2 above are done.

   If = 4, then the data are CORRECTED for the error in the
     sampler statistics when they are within the tolerance
     of the correct value.    The default is that the sampler
     statistics must be within 12 percent of the correct value in order
     that the correction can be made.   Any visibilities outside
     of this range will be dropped.   In addition, the correction
     can only be made of the zero statistic is within  5 percent of 50 percent
     and if the difference of the positive and negative statistic
     is within 5 percent.  Data outside of this range will again be
     dropped.   Note also that the sampler correction is for
     small correlation coefficients only.   Thus, autocorrelations,
     if loaded, are not corrected. Similarly, if you have visibility
     amplitudes of the order of 50 Jy, they should not be corrected

CPARM(7)  The internal setting for the acceptable tolerance in
   the sampler statistics is 5 percent if C(6)=1,2 or 3 and 12 percent if
   C(6) = 4.   If you want a different TOLERANCE,  set CPARM(7)
   to this value.

CPARM(8)  AIPS normally converts adverbs to upper-case, so
   reading files with lower-case names is impossible. Files on
   VAX disks, available to unix via multinet, are seen by
   default on a unix system as lower-case. If CPARM(8) is set
   true, ATLOD converts the string in DATAIN back to lower-case,
   before reading the file.

CPARM(9)  If you set OPTYPE='SYSC' then information from
   the SYSCAL group is written into three text files for
   you to examine it (3 is the maximum number of text files
   that can be open at one time in AIPS).These files go
   to the $FITS area.  Each file contains 14 columns, the
   first is the time, the second the frequency, and the
   next 12 are the quantity and a flag for each antenna.
   A flagf of 0 means the data are good, a flag > 0 means
   the on-line system flagged it bad (e.g. not on source
   for Tsys)

   If C(9) = 1  the files contain the XY phase differences
     (XYPHS_UID), the X (TSYSX_UID) and the Y (TSYSY_UID)
      feed system temperatures.

   If C(9) = 2  the files contain the X sampler statistics.
     The negative (XSSNEG_UID), zero (XSSZER_UID) and
     positive (XSSPOS_UID) statistics are each put into
     a separate file.

   If C(9) = 3 the files contain the Y sampler statistics

   Any other value => 1   UID is your AIPS user number

XYPHASE  The on-line XY phases for each antenna, are still quite
   often very poor, especially at 13 and 20 cm.  We believe that
   these phases shoudl be constant over the observation.   If
   you have a better estimate of the XY phase than the on-line
   value, you can put it in here, and set CPARM(5)=1.

   You might use OPTYP='SYSC' to write the XY Phases to a text
   file and examine them.  Then input into the XYPHASE array,
   the mean value for each antenna.   If you have more than one
   frequenciy in the file, then you can input 6 XYPHASEs for
   the first frequency and then the next 6 for the second

   We need to associate the correct XYPHASE with the correct.
   frequency. The order in which to input the XYPHASEs is the
   order in which they list in the FQ table.
   Thus,   first 6 phases for IF 1, FREQID 1, then IF 2 FREQID 1,
   and so on to IF n FREQID 1.  Then start on FREQID 2 and so on.
   Note that if IFMAP=-1 the location of a frequency on the IF
   axis does not necssarily reflect its on-line IF chain value.
   You should know what you are doing here in the more complicated
   cases.  See me if unsure (NEBK).

   You must always specify 6 values if you are going to specify
   values for more than one frequency. This is true even if
   antenna 6 is not in use for your observation.

   SPace for up to 6 frequencies is provided.  This would require
   thirty numbers

DPARM(2)  If > 0, then the spectra are three point Hanning smoothed.
   The real and imaginary spectra are smoothed separately.   The
   smoothing occurs before channel selection, so that you can
   smooth the data, and then drop every second channel say.  This
   is appropriate for spectral-line observations where adjacent
   channels in the spectrum are intiially independent, but the
   raw spectrum is still convolved by a SinC function which will
   cause sidelobes about narrow spectral features.

DPARM(3)  ATLOD has always written the weights for each visibility
   as 1.0.  However, as of ATLOD version 03/mar/94, and for data
   taken after 22/feb/04, it now by default sets the weights
   to the integration time divide by 15 sec.  This includes
   allowing for time losses such as the blank and hold times.

   If DPARM(3) > 0 then the weights will be set to DPARM(3).
   So if you wish the weights to still be unity, set DPARM(3)=1

   If you wish to concatenate with some old data with weights=1,
   you should use the REWEIGHT adverb of DBCON accordingly

SHOW and TELL are implemented for ATLOD.  Type


to see which adverbs you can change while ATLOD is running.
Set their values appropriately, and then type


to inform ATLOD of your intentions.  ATLOD receives TELL
information in two places.  The first is after each specified file
has been skipped (NSKIP).  Thus you could change NSKIP if
you wish.   The second is after each scan has been read into
AIPS.   ATLOD will inform you when it receives a TELL command.

Here are the adverbs you can change.

OPTELL           Operation ('CHAN', 'QUIT', 'ABOR')
NSKIP            Number of tape files to  skip
NFILES           Number of tape files to load
BCOUNT           First scan to read
NCOUNT           No. scans to read
SOURCES          Source selection list
TIMERANG         Time range selection
FREQSEL          Frequencies to select (MHz)
IFSEL            IFs to select

set OPTELL='CHAN'  to change them.
set OPTELL='QUIT'  to have ATLOD finish cleanly at the next
                   available opportunity.    VERY useful if
                   you suddenly find you asked ATLOD to read
                   many files from tape, but that you have not
                   selected any data from the last 10 and don't
                   want to wait while it finds this out.
set OPTELL='ABOR'  to have ATLOD abort

The only one that can run you into trouble is IFSEL.  Consider
loading data where you have set IFSEL=0 and IFMAP=-1.  Let
us say IF 1 goes to IF axis 1, and IF 2 goes to IF axis 2.

Now, with TELL, you change IFSEL=2,0   ATLOD then finds on the
next scan that you don't want IF 1 any more, but that you do want
IF 2.  It sees that IFMAP=-1, which means put the data on the
first available IF axis location.  Thus, IF 2 will now arrive
in IF axis 1.   Ths would be a bit of a mess.

However, if you set IFMAP=1 in this case, IF 2 would still go to
IF axis 2 and all would be well.