Updated EVN station positions: FR005 reanalysis
FR005 was a 6-cm phase-reference test experiment. It comprised ~1hr of interleaved observations of 3C345 & J1640+3946 (separation 0.5deg), then ~1hr of 3C345 & J1635+3808 (separation 2.25deg), both using a cycle time of 3.25min, and finally another ~1hr of 3C345 & J1640 at a shorter cycle time of 2min. Seven EVN telescopes participated: Ef, Mc, Nt, Wb_array, Jb2, On85, and Tr. Of these, only Ef, Mc, and Nt participate regularly in geodetic campaigns and hence had sub-cm positions in the original analysis.
FR005 was initially analyzed using the station positions in the SCHED station catalogue. Problems with the phase residuals and phase-reference maps were evident, and appeared consistent with station-position errors. We used the unrelated Network Monitoring Experiment C00C1 (12 hours on DA913) to estimate station-position adjustments based on residual delay-rates. Delay-rate solutions of course provide no information about the z-component of the station position, and with only one source there was no possibility of treating atmospheric effects in a satisfactory manner. Station-position offsets for Jb2, Wb_array, and Tr were estimated from the C00C1 delay-rate residuals. A position of the On85 antenna relative to the (geodetic) On60 antenna was provided by P. Charlot and G. Elgered. A reanalysis of the FR005 observations was performed using these updated positions, including the Wb axis offset that was missing in the SCHED database. Improvements in the phase residuals and phase-reference mapping were found, but there were still noticeable problems.
Patrick Charlot volunteered to conduct a 6cm test experiment using geodetic techniques in order to determine optimal positions for EVN stations that do not participate in geodetic campaigns (experiment TP001). He has now completed analysis of this experiment sufficiently to obtain positions determinations likely good to 10cm. We have done a further reanalysis of the FR005 observations with these TP001-derived positions, including the Jb2 axis offset that he discovered to be also missing. The improvements over the preliminary C00C1 reanalysis are evident, and are very significant compared to the original situation.
The figure below shows phase residuals on baselines from stations with updated positions to Ef for the pair 3C345 - J1640+3946. Click on the little image to download a full-size gzipped postscript file. On85 did not participate during the first hour; the source-swapping rate was too fast for Wb in the second. Three analyses were completed as above; one using the original station positions, one using the C00C1 position estimates, and one using the TP001 position determinations. In each analysis, the two sources were FRINGed separately, leading to a CL table "for" 3C345 and one "for" J1640. The plot shows four sets of points for post-FRING phases of J1640 for each baseline:
- purple: calibrated using its own CL table. These residual phases are ~0, and essentially the same regardless of the station-position adjustments used.
- grey-green: analysis using station positions as in the SCHED database; calibrated using 3C345's CL table. Clear non-zero trends are seen.
- light-blue: analysis using C00C1 station positions; calibrated using 3C345's CL table. The phase residuals approach 0, but there are still noticeable discrepancies, especially at Jb2.
- red: analysis using TP001 station positions; calibrated using 3C345's CL table. Jb2 phase-residuals essentially negligible, Tr improved.
The figure below corresponds to the previous one, but for the pair 3C345 - J1635+3808; all analysis/plotting details are the same. All stations participated. The slopes of the uncorrected phase residuals are more severe here, since the pair has a greater separation. The improvement in using the TP001 station-position adjustments are similarly more evident, especially at Wb.
The pair 3C345 - J1635 also best illustrates the improvements in the phase-reference maps themselves. We can compare the map of J1635+3808 made using its own CL table with those phase-referenced from 3C345 (CL tables applied cross-source during SPLIT). All maps were made by IMAGR with the same adverb settings for consistency in the comparisons (CELLSIZE=0.5mas, NITER=200, LOOPGAIN=0.1), no self-cal in the "own-CL" map. Better maps could be produced with more careful tinkering with the CLEAN-control parameters. The below table summarizes the features of the J1635 maps for each of the four cases:
Map Type | Peak FD | Total FD | SNR |
---|---|---|---|
[Jy/bm] | [Jy] | ||
own CL table | 2.15 | 2.42 | 320 |
PhRef, orig pos. | 0.49 | 1.88 | 12 |
PhRef, C00C1 pos. | 1.11 | 1.96 | 39 |
PhRef, TP001 pos. | 1.55 | 1.99 | 62 |
The figure below shows the maps of J1635+3808 for the four cases described in the table. The images were plotted with the same color map in each case, ranging from -0.1 to 0.9 Jy/bm. Click on the image to download the full-size gzipped postscript file. The map of J1635 made with its own CL table is in the upper left. This is followed, left to right and top to bottom, by the phase-reference maps using the original, C00C1, and TP001 station positions/axis offsets of On85, Tr, Wb, and Jb2. The problems with the phase-reference map using the original positions are evident, as is the improvement in focusing the flux into the source and the reduction of the off-source rms when the TP001-derived positions are used.
The phase-referencing is especially sensitive to the station-position improvements because 4 of the 7 stations initially had station positions that were 1-5 meters off. See the recipe for incorporating the TP001 station-position adjustments into already-correlated data for a more quantitative summary of the position adjustments.
In January 2002, the calculation of the offset between the single-dish and tied-array Wb positions was refined. Improvements were checked in the Network Monitoring experiment N01C3, which performed phase-referencing between DA193 and J0605+4030, 2.09deg apart, at 5GHz. The removal of a ~150deg drift in Ef-Wb, Mc-Wb, and Jb-Wb was evident. The Wb array position in the recipe includes this refined correction.
EVN webmaster (jive{at}jive.eu)