Recipe for Incorporating Improved EVN Station Position into AIPS 24 July 2002 (last updated 14 July 2004) The procedure below will allow you to incorporate the improved EVN station position determinations into your already-correlated data. This procedure uses the task CLCOR to include the effects of an updated station position into CL tables. 0. DIFFERENCES FROM PREVIOUS RECIPES You may notice that this recipe differs in format and content from previous versions. In the beginning, we could provide a single set of station-position corrections that would apply to everybody; however, things are more complicated now. Contributing factors include: *) In an effort to correlate at the best-determined station positions at the time of correlation, we've gone through a couple iterations of incorporating preliminary station-position improvements as they became available. This introduces a correlation-date dependence into the required correction. *) The CLCOR program in the stable release of the 31Dec01 AIPS differs from previous versions, in that it now updates the station positions in the AN table when you apply corrections. (This is also covered in EXPLAIN CLCOR in the 31Dec01 AIPS, under the 'ANTP' description on page 2 of the EXPLAIN output.) *) We have recently also incorporated the effects of station velocities (i.e., plate motion) into the correlation. For earlier experiments, this introduces an observation-date dependence into the required correction. Rather than attempting to cover all possibilities, below I'll illustrate the steps required to incorporate the improved station positions assuming that the experiment was correlated before we made any station-position improvements at the correlator and that you have made no previous corrections with CLCOR. I'll draw attention to the implications of the above points where applicable. Individual e-mails have gone out (~03May2002) to PIs of all phase-reference experiments observed up through 2001 that have been correlated at JIVE, with specific details pertaining to their experiments. Subsequent phase-reference experiments will need none of the corrections described in this recipe (of course, future improvements in the determination of site positions/velocities are always possible). A similar set of e-mails was sent out to PIs of phase- referencing experiments correlated elsewhere (~23July2002). PIs of other experiments (i.e., who have not received such email) can contact Bob Campbell (mail icon below) for details appropriate to their experiment(s), if desired. 1. DO I NEED TO DO ANYTHING? In this sample illustration, we'll ignore the effects of station velocity (see notes in 8.b below). Use PRTAN or TBOUT to view the AN table. If you see the values in Table 1 for the station coordinates (column2 = STABXYZ) for ON, TR, WB, JB, CM, MH, and AR (as appropriate for your experiment -- make sure you check the right antenna for stations ON, WB, and JB), then your data has been correlated with the most-recent improved station positions and you don't need to do anything further (if they differ by just a few cm in all cases, it's likely effects of plate motion, which aren't included here). If not, continue on to step 2 to improve your phase-referencing results. --| Table 1: Updated Station Positions (in a Left-Handed Frame): --| X [m] Y [m] Z [m] --| On85 3370966.126 -711465.954 5349664.023 --| Tr 3638558.51 -1221969.72 5077036.76 --| Wb7 3828651.29 -443447.48 5064921.57 --| Wb_arr 3828445.659 -445223.600 5064921.568 --| Jb2 3822846.76 153802.28 5086285.90 --| Jb1 3822626.04 154105.65 5086486.04 --| Cm 3920356.15 -2542.02 5014284.42 --| Mh 2892583.77 -1311715.59 5512630.07 --| Ar 2390486.90 -5564731.44 1994720.45 a) The above coordinates are in a left-handed frame, to match what would be displayed in the usual case by AIPS. The conversion from right- to left-handed frames is accomplished simply by changing the sign of Y. All further values in this recipe will be in a left-handed frame, for direct entry into the CLCORPRM parameter in the task CLCOR. Note, however, that catalogues of station positions (locations.dat in SCHED, the ITRF from the IERS, etc.) generally use a right-handed frame. 2. SO I NEED TO UPDATE MY STATION POSITIONS As mentioned above, updating station positions in already-correlated data really means adding corrections into the CL table (via the task CLCOR); these account for the delay/rate/phase shifts resulting from moving to station positions different from those used to compute the model during correlation. This must be done before any sort of FRINGing occurs (see EXPLAIN CLCOR), i.e., on the CL table resulting from INDXR or that resulting from incorporating the SN table from ANTAB/APCAL via CLCAL (see note [e]). Let's call this initial CL table to update CL2 (INDXR --> CL1; APCAL --> CL2). (a) Note for PIs of experiments not correlated at JIVE who received e-mail based on their PRTAN output: if you have run CLCOR to modify station positions prior to the PRTAN run whose output you e-mailed to JIVE, the provided corrections are with respect to that post-CLCOR AN table. This in turn would imply that you have a higher CL-table version that should be used instead of the CL2 assumed here (for you, the symbol "CL2" should be replaced by the post-CLCOR, pre-FRING CL-table version). If you used a 31Dec00 or earlier version of CLCOR, things could get trickier -- see para. 3.b below. 3. SAFEGUARDING THE CL & AN TABLES a) Use TACOP to copy the CL table CL2 to a higher version number. Work from this new version. CLCOR will update the CL table version specified by GAINVER in-place; it does not create a higher-version CL table. The original CL2 will therefore remain as a starting point if you need it. b) Use TACOP to copy the AN table to a higher version number. CLCOR will update the AN table version=1 (as well as the CL table version=GAINVER) as follows: 31Dec01 version: for all station position and antenna axis-offset changes =<31Dec00 version: for all antenna axis-offset changes only In this case of the AN table, the higher-version AN table becomes the "original" backup. However, in all other tasks requiring specific AN-table selection, remember to use AN version=1 rather than the default "highest version". (The AN-table numbering can become more complicated when sub-arrays are used; this concern is not relevant to any experiments correlated at JIVE prior to the station-position improvements.) Use of the 31Dec01 or later version is highly recommended, since it will provide you with a record of accumulated station-position corrections that is internal to the data and easy to review. Regardless of how many iterations of CLCOR runs you have made, you will be able to determine immediately whether any further corrections are needed simply by subtracting PRTAN/TBOUT output from the correct station positions. You can always check the action of CLCOR if in doubt by making an obvious change to one station (say, 50m in X), and noting whether the corresponding value in the AN table indeed has changed. You can easily 'un-do' this test by running CLCOR again and making the opposite change (i.e., -50m in X). Step 5.b has the details for making station-position changes with CLCOR. 4. CORRECTING THE STATION AXIS OFFSETS Use PRTAN or TBOUT to check the axis offsets for Wb, Jb2, or Cm in the AN table (column 6 = STAXOF). The correct values should be: Wb = 4.95m Jb2 = 0.458m Cm = 1.503m If the listed STAXOF for any of these stations is different, run CLCOR as many times as appropriate to update them (only one station may be updated per run). You should correct all data/time (BIF=EIF=0, SOURCES=' ', TIMER=0, etc.). GAINVER should be the version of the CL table resulting from Step 3.a. OPCODE='ANAX' for axis-offset corrections. The CLCORPRM parameter should be the difference (correct value) - (current AN-table value). Assuming that the current AN table shows all 0s for these stations, the station-specific adverb settings are: Wb (7 or array) ANTENNA = [Wb]; CLCORPRM = 4.95, 0 Jb2 ANTENNA = [Jb]; CLCORPRM = 0.458, 0 Cm ANTENNA = [Cm]; CLCORPRM = 1.503, 0 Here, the [Station] symbol means insert the appropriate antenna number. If you received individual e-mail for your experiment(s), use the values in that e-mail in place of these. Some care is required using CLCOR with OPCODE='ANAX': it would seem from the help that you are replacing the effects of the old axis offset with the CLCORPRM value. However, the behavior of CLCOR/'ANAX' is consistent with changing the axis offset by the amount of CLCORPRM. For example, if you run CLCOR twice for Wb, you will see an axis offset of 9.9m in the AN table (version=1) and delay/rate/phase shifts in the CL table (version=GAINVER) that are twice too large. To un-do a correction, run CLCOR again, using the opposite sign in the CLCORPRM value. 5. CORRECTING THE STATION POSITIONS a) Again, for the sake of this illustration, we assume your experiment was correlated before we had improved station-position determinations. In this case, your PRTAN/TBOUT output would be as in Table 2: --| Table 2: Station Positions from SCHED (in a Left-Handed Frame): --| X [m] Y [m] Z [m] --| On85 3370968.181 -711464.917 5349664.113 --| Tr 3638558.0 -1221967.0 5077041.0 --| Wb7 3828647.18 -443450.02 5064923.08 --| Wb_arr 3828440.64 -445226.03 5064923.08 --| Jb2 3822842.66 153800.13 5086287.22 --| Jb1 3822626.497 154105.5889 5086486.2618 --| Cm 3920354.8 -2545.7 5014285.0 --| Mh 2892579.9681 -1311719.0699 5512640.6897 --| Ar 2390453.85 -5564816.46 1994663.40 ***If your Y-components have the wrong sign for all stations, you have a FITS file made at JIVE before 24aug2000. Use of this recipe will not improve your phase-referencing results. Contact your support scientist at JIVE to arrange for delivery of a new FITS file that can incorporate these station position updates. b) Run CLCOR as many times as appropriate to incorporate the changes for the following stations (again, only one station can be updated per run). If you have multiple frequency IDs, you will have to run CLCOR once for each FREQID for each station. You should correct all data/time (BIF=EIF=0, SOURCES=' ', TIMER=0, etc.). GAINVER should be the version of the CL table resulting from Step 3.a. OPCODE='ANTP' for station-position adjustments. Again, the CLCORPRM values should be the difference (correct value) - (current AN-table value). In the case of this illustration, the station-specific adverb settings are (if you received individual e-mail for your experiment(s), use the values in that e-mail in place of these): On85 ANTENNA = [On]; CLCORPRM = -2.055, -1.037, -0.090, 0 Tr ANTENNA = [Tr]; CLCORPRM = +0.51, -2.72, -4.24, 0 Wb_arr ANTENNA = [Wb]; CLCORPRM = +5.019, +2.43, -1.512, 0 Wb7 ANTENNA = [Wb]; CLCORPRM = +4.11, +2.54, -1.51, 0 Jb2 ANTENNA = [Jb]; CLCORPRM = +4.10, +2.15, -1.32, 0 Jb1 ANTENNA = [Jb]; CLCORPRM = -0.457, +0.0611, -0.2218, 0 Cm ANTENNA = [Cm]; CLCORPRM = +1.35, +3.68, -0.58, 0 Mh ANTENNA = [Mh]; CLCORPRM = +3.8019, +3.4799, -10.6197, 0 Ar ANTENNA = [Ar]; CLCORPRM = +33.05, +85.02, +57.05, 0 Here, the [Station] symbol means insert the appropriate antenna number. These position offsets to incorporate via CLCOR are appropriate for the left-handed frame used in AIPS for VLBI data. To un-do a correction, run CLCOR again, using the opposite sign in the CLCORPRM values. Again (not to beat a dead horse), CLCOR run with OPCODE='ANTP' does not change the AN table in AIPS versions 31Dec00 and before; other than the modified delay/rate/phase values in the CL table version=GAINVER, there is no record that you've made the the necessary changes, so keep good notes. AIPS version 31Dec01 does change the AN table for the station- position adjustments entered with 'ANTP'. 6. MY AN TABLE DOESN'T LOOK LIKE TABLE 2 This is quite possible, for among other reasons, because your experiment was correlated with improved station positions or because you have already run CLCOR (in AIPS version 31Dec01) with OPTYPE='antp'. In such cases, you can easily determine your own CLCORPRM values using the rule CLCORPRM = (correct value) - (current AN-table value), remembering to work in a left-handed frame. Run CLCOR as above in step 5.b with your own values for CLCORPRM. It is in this situation that some of the complications come to the fore: a) If you ever ran CLCOR from an AIPS version 31Dec00 or earlier, any station-position corrections you made won't be reflected in the AN table. You will need to rely on separately-kept notes for the "(current AN-table value)" term in computing your CLCORPRM values. b) Table 1 doesn't include station-velocity effects. If you want the (correct value) term to include them, you'll need to include them yourself. The individual e-mail to PIs of phase-reference experiments discussed in step 0 does include these terms. We have an automatic program for computing the CLCORPRM values for any experiment correlated at JIVE, including the effects of station velocities, relative to what was correlated (i.e., assuming the user has yet to make any CLCOR corrections). There is a similar program using PRTAN output to compute CLCORPRM values relative to the AN table, for experiments not correlated at JIVE. Contact Bob Campbell (mail icon below) for more info. 7. STATION POSITIONS/AXIS OFFSETS ARE NOW UPDATED... If not already done, proceed with the ANTAB/APCAL steps to incorporate system temperature information into the CL table now updated for the station position and axis-offset improvements. Subsequent steps in AIPS proceed as usual (i.e., FRINGing for manual phase-cal phases, full FRINGing on the reference source, etc.). 8. NOTES a) Updated station positions are in ITRF2000 (epoch 1997.0). The SCHED database generally uses USNO 1998-6 (epoch MJD 50449 = 0UT 1jan97). The difference between these two frames for geodetic stations on the European continent is 2.0 cm or less in any coordinate. Since this is below the estimated ~5cm uncertainties of the new station position determinations, I've made no effort yet to calculate a transformation matrix between ITRF2000 and USNO 1998-6. b) As mentioned above, the reference epoch for most stations in locations.dat in the current release of SCHED is 1997.0. The approximate net effect of the catalogued station velocities over the 5-year period to 2002 would be on the order of: Abs. site displacement Abs. baseline length European EVN stations 15cm 3cm Shanghai 20cm 20cm US (not including Mk) 10cm 15cm where baseline lengths refer to Effelsberg-* baselines. The baseline length can be smaller than the contributing site displacements because of nearly parallel station velocities; in such a case there'd be an associated apparent change in HA and declination of a source at a given epoch. Effects on experiments observed prior to 2002 will scale linearly with (t-1997). Station motions should be added to the values in Table 1 (keeping in a left-handed frame). The catalogued station velocities have been either derived through geodetic observations or calculated from the no-net-rotation NUVEL-1A plate motion model. c) The celestial pole offsets, which are observed corrections to the IAU1980 Nutation model, are not currently included in the JIVE correlator model. The net effect of this omission can exceed the cumulative effects of plate motion discussed above, but is much less than any individual non-geodetic station's position correction. Contact Bob Campbell (mail icon below) for more information about your experiment if you think this will be a concern. d) The Wb axis offset was incorporated into the SCHED database first, hence there are experiments that have been correlated at JIVE with its correct value, but not with that for Jb2 or Cm (see Step 4). There was also a period in which Wb_arr was correlated at a preliminary improved position, which appears neither in Table 1 nor 2, thus Wb in your AN table may not match either if your experiment was correlated at JIVE between Aug2001 and Jan2002. e) There is a characteristic of EVN observations up through 2001 related to applying ANTAB/APCAL solutions that may also affect the application of station-position offsets for phase-reference observations. For continuous tape motion, most EVN stations typically made a Tsys measurement only at the start of each pass. If each pass begins with the same source (say SRC_1) and you use INTERP='SELF' in CLCAL to apply the APCAL SN table, then the effect of the station position offsets would be prevented from being applied to SRC_2. Using INTERP='2PT' in CLCAL would avoid this specific problem, but at the cost of applying source-flux dependent Tsys biases incorrectly to SRC_2 (more important the more the flux densities of SRC_1 and SRC_2 differ). f) Also under this item in the EVN Users' Guide web page, there is a summary of the reanalysis of the phase-reference test observations FR005 using these updated station positions and illustrations of the improvement in phase residuals and the phase-reference maps themselves.