We examine here the way to determine the baseline vectors from calibration observations done for instance after an antenna move. Assume the relevant observations are gathered into the current index. One should make a plot of phase versus both Hour Angle and Declination (SET X HOUR_ANGLE DECLINATION and SET Y PHASE, for each sideband separately.
The command SOLVE BASELINE will then do a least square fit and output the baseline vectors. A linear method is used which works only if the starting values (that is the baselines values in the header) are within half a wavelength in any direction of the true values. Different starting values may be given with the /OFFSET baseline dx dy dz option. In that case the phases are first corrected for the given baseline offsets before the fit.
A blind search within a given range may be done. For this enter the option /SEARCH range where range is in meters. All offsets within the given range (in three directions, stepping by half a wavelength) are tried, and the best rms value is kept. A blind search in a 0.01 meter range will take about 4 seconds on the microVAX 3400. This the default if only SOLVE BASELINE /SEARCH is given. Rms values on the range 10-20 degrees are expected, if the data was scan averaged, and observing conditions were acceptable.
The program gives for each baseline two series of offsets, and the fitted baseline for record. The first offsets given (dx, dy, dy) are relative to the baseline coordinates used while observing, while the second set of offsets (DX, DY, DZ) are relative to the ``standard values'' of the antenna positions. After all available baselines have been fitted, the command PRINT BASELINE will produce a command procedure CLIC-BASELINE.OBS. This command procedure may be used in OBS, in the CONFIGURE mode, to update the antenna position offsets.
If SET ANTENNA 2 3 has been used instead of SET BASELINE, then the antenna position offsets of antennas 2 and 3 are directly determined (the position offset of antenna 1 are by definition zero, like antenna phases for this antenna in the SET ANTENNA mode. This mode is preferable since baseline closured is ensured.
Residuals may be examined by entering RESIDUALS BASELINE; phases should be independent of hour angle and declination.
Phase data can later modified in previous data to reflect this baseline change (if a previous value proved to be wrong) with MODIFY BASELINE FITTED.
If SET X HOUR DECL TIME is specified, then plots of phase as a function of time are obtained; RESIDUALS BASELINE will also display the variation with time of the residuals. This is useful to detect a time drift of the phases, especially during long observations (to be avoided in any case!). These drifts may be fitted simultaneously with the option /POLYNOMIAL degree. In that case a time polynomial (degree up to 3) is added to the fitted phase function.