[ Basic Info | User Guide ]
Basic Information on gpcal
Purpose: Gain/phase/polarization calibration of dual feed data.
Gpcal is a MIRIAD task that determines calibration corrections
(both antenna gains and instrumental polarisation
characteristics) for an array with dual feeds, from an
observation of a point source. The source can be polarised,
with unknown polarisation characteristics. Though the source
may be strongly polarized, the instrumental polarisation errors
are assumed to be small (of order at most a few percent).
Normally GPCAL writes the solutions as a gains table (item
'gains') and a polarization leakage table (item 'leakage').
With nfbin>1, GPCAL writes additional tables 'gainsf' and
'leakagef' with solutions for frequency binned data.
Not many other tasks currently read these tables, but
they will be applied in preference to the single frequency
versions if present.
GPCAL can handle either dual linear or dual circular feeds.
However by default it expects dual linears -- you must use
options=circular so switch it to circular mode. Also the
terminology in this document is aimed at linears (e.g. we talk
of "XY phase" -- for dual circulars this is really "RL phase").
Note that the user specifies which parameters are to be solved
for. In the case of leakages and xyphases, GPCAL will check for
the existence of items "leakage" and "gains" in the input data-
set. If present, these will be used as the initial estimates of
these parameters. If you are not solving for these parameters,
they will be held at their initial value through the solution
process. After converging on a solution, and if the xyphase
offsets or leakage parameters have been modified, GPCAL will
write out their current values to the appropriate items.
Conventions: Unfortunately there has been a number of changes in
the "sign conventions" used within Miriad. For a discussion of
the conventions, past sign errors and how they affect you, see
the memo "The Sign of Stokes-V, etc." by Bob Sault.
Input visibility data file. The data should be either raw linear
or raw circular polarisations. No default. The visibility data
must be in time order.
Standard uv selection. Default is all data.
Standard line-type specification. Multiple channels can be
given. Generally it is better to give multiple channels, rather
than averaging them into a "channel-0". The default is all the
channel data (or all the wide data, if there is no channel
The values of the I,Q,U,V Stokes parameters. If no values are
given, and it is a source known to GPCAL, GPCAL uses its known
flux as the default. If GPCAL does not know the source, the
flux is determined by assuming that the rms gain amplitude is 1.
If the option "qusolve" is used, the given fluxes for Q and U
are used as the initial estimates. Also see the oldflux
The reference antenna. Default is 3. The reference antenna
must be present throughout the observation. Any solution
intervals where the reference antenna is missing are discarded.
The minimum number of antenna that must be present before a
solution is attempted. Default is 2.
This gives one or two numbers, both given in minutes, both being
used to determine the extents of an amplitude calibration
solution interval. The first gives the max length of a solution
interval. The second gives the max gap size in a solution
interval. A new solution interval is started when either the
max time length is exceeded, or a gap larger than the max gap is
encountered. The default is max length is 5 minutes, and the
max gap size is the same as the max length. The polarisation
characteristics are assumed to be constant over the observation.
The number of frequency bins. The default is 1. Use nfbin>1 to
solve for variation across the band in the gain and leakage
Works best for uv files with a single spectral window, i.e.,
Error tolerance. The default is 0.001 which should be adequate.
Generally the use of this parameter has been superceded.
Initial estimate of the XY phase of each antenna. The default is
0 for all antennas. If the XY phase has not been applied to the
data, then it is important that this parameter is set correctly,
particularly for the reference antenna.
These options determine what GPCAL solves for. There are many
permutations, the more obscure or useless of which are not
supported. The option values are used to turn on or off some of
the solvers. Several options can be given, separated by commas.
Minimum match is used. Some combinations of these options are
xyvary The XY phase varies with time. By default the XY
phase is assumed to remain constant.
qusolve Solve for Q and U fluxes. Good parallactic
angle coverage is required for this.
oldflux This causes GPCAL to use a pre-August 1994 ATCA
flux density scale. See the help on oldflux for
nopol Do not solve for the instrumental polarisation
leakage characteristics. The default is to solve
for the polarisation leakages on all feeds except
the X feed of the reference antenna.
noxy Do not solve for any XY phase offset. The default
is to solve for the XY phase offset on all antennas
except for the reference antenna.
nopass Do not apply bandpass correction. The default is
to apply bandpass correction if possible. This is
rarely useful. Use with caution.
noamphase Do not solve for the amplitude and phase. The
default is to solve for amplitude and phase. This
option is rarely useful.
linear Expect/handle data from feeds that are linearly
polarised. This is the default.
circular Expect/handle data from feeds that are circularly
The following are options for advanced users, and should be used
with forethought and caution.
reset If calibration tables (leakage and gains) are
present within the input dataset, GPCAL usually
uses these in determining default XY phases,
leakages and flux scale. This is usually the
desired behaviour. The "reset" option suppresses
this behaviour, and starts GPCAL with a clean
xyref Solve for the XY phase of the reference antenna.
To do this, the source must be linearly polarized
and you must have reasonable parallactic angle
coverage. This option can be used with "noxy", in
which case GPCAL solves for the offset of the
polref Solve for the instrumental polarization leakage of
the X feed on the reference antenna. This can be
combined with "nopol", in which case GPCAL solves
for X feed of the reference antenna only.
vsolve Solve for the Stokes-V of the source. This is only
possible for linear feeds and a preliminary leakage
solution for the observation already exists. This
preliminary solution must be formed from a
calibrator with known Stokes-V.
Revision: 1.15, 2013/01/29 00:55:15 UTC
Generated by firstname.lastname@example.org on 29 Jan 2013