The following procedures are specific to the reduction of point astrometry data.


Initialize the dispersion corrections. In order for calibrated delays to be processed by average, both dispersion and metrology corrections have to be initialized or computed. Initialize (zero) the dispersion corrections for the specified stars. Initialization for all stars is the default. A class (ALL, POINT, SCAN) can be specified using keyword class; the default is all classes. For scan data, initialization means setting the dispersion corrected delays equal to the averaged FDL data.
Flag the visibility phase for the specified channels and beams. If none are specified, do channels 11 and 21 to 32 in all beams. If beams are not specified, but channels are, assume this selection is valid for all beams. If channels is a linear array and beams are specified, use the same channel selection for the specified beams. If channels is a two-dimensional array, the first index refers to the beam, and the second index to the channels to be flagged for the beam specified in the second parameter.
Enforce the closure relations. Since the fringe delays computed for the coherent integration of the complex visibilities are baseline based, they do not necessarily close, i.e. add up to zero. This procedure makes sure they do, and adds the correction to the phases.
Unwrap the visibility phase as a function of wavelength.
Compute the dispersion correction. Place the results in common block.
Initialize (zero) the metrology delay corrections. They can be recomputed using metrocorr or astrocorr.
Compute delay corrections from $(x,y,z)$ solutions provided by INCHWORM and stored in variables parx, pary, and parz. Interpolate to the epochs of the point data. Results are stored in variable metropos.
Compute delay corrections by interpolation of the INCHWORM solutions already converted to delays and stored in variable metropath to the observed point data times. The results are stored in metropos. Note that metrocorr and astrocorr store their results in the same variable, since they should give very similar if not identical results.