Thursday

June 20, 2013

We present here an analysis of the effects of including VLBA RDV data in global SOLVE solutions on astrometric position estimation. This is a work in progress. In the context used here, MkIII data is defined as VLBI data correlated with a Mark III/IV correlator. The MkIII data covers the time range from 1979 to 2001 and is comprised of about 2.5 million observations. VLBA data is defined as that data obtained under the guise of the VLBA RDV proposals, correlated at the VLBA correlator in Socorro, NM and covers the time range from 1997 to 2000 (RDV01 to RDV24 inclusive). The VLBA data set is comprised of about 500,000 observations. All solutions in Sections I-III were made using a parameterization represented by this SOLVE BATCH control file. The only difference between different solutions is the different data used (eg. MkIII only, VLBA only, MkIII+VLBA, etc.).

I) First, here are the familiar NRD (normalized residual delay) versus declination plots which first suggested that there may be a problem (excess NRD and/or position variation) with the VLBA data for sources south of the celestial equator. The things to note in these plots are 1) there appears to be larger than normal NRD, i.e. greater than 1, for some sources but these sources are not necessarily all southern sources and 2) there is a seasonal effect, i.e. winter experiments have fewer sources with elevated NRD than summer experiments. Also note that additional editing of the data improves the results for some experiments.

* automatic outlier elimination has been used for additional editing.

II) Here are NRD versus declination plots for all VLBA RDV experiments combined in one solution, summer only VLBA RDV experiments, winter only VLBA RDV experiments, and for reference, all MkIII data. Finally, here is the NRD plot for all data combined (MkIII plus VLBA). Note that in the combined solutions, with the exception of a few discrepant points, the NRD "problem" appears to almost completely disappear (or at least is extremely less well pronounced).

III) Here we examine the differences in estimated positions for different sets of solutions. Weighted rms differences are listed at the end of the difference tables.

IV) Here we compare "arc" positions of selected sources derived from SOLVE Global solutions using VLBA RDV data and NEOS data. Two sets of solutions were made: 1) using only VLBA RDV data (RDV01 to RDV24, inclusive); and 2) using only NEOS data from the same time period as the VLBA RDV data (NA197 to NA397). A total of 64 sources were observed with sufficient frequency in the VLBA RDV data to estimate positions as arc parameters. Of these 64 sources, a total of 54 sources were observed with sufficient frequency in the NEOS data to estimate positions as arc parameters. Four solutions were made with each data set. In each of the first three solutions, one third of the sources were estimated as arc parameters (all other sources were treated as global). The fourth solution was used as a control, i.e. all sources in the fourth solution were estimated as global parameters. The SOLVE BATCH control files for the VLBA RDV data are available here (1, 2, 3, Control). The SOLVE BATCH control files for the NEOS data are available here (1, 2, 3, Control). Solution differences are minimal and are listed in the tables below.

Plots of "arc" position (offset from a mean position) versus epoch were generated for each source using the GSFC program ploser. These position time series plots are available in the table below for the VLBA RDV data.

Weighted root-mean-square (wrms) position residuals were also calculated for each source using the GSFC program ploser. Plots of wrms residuals were made and are available below.

• wrms residuals vs. RA and DC for the VLBA RDV data
• wrms residuals vs. RA and DC for the NEOS data
• wrms residuals from the VLBA RDV data vs. wrms residuals from the NEOS data