M87 VLBA Project Array Performance

R. C. Walker

Updated June 2025.

The table below is a summary of all of the observations related to the VLBA project to study the dynamics of the inner M87 jet at 43 GHz. The main publication for that project is Walker, R.C., Hardee, P.E., Davies, F.B., Ly, C., and Junior, W. 2018, Ap. J. 855, 128. There are also images and movies on the web on my personal web pages at NRAO. The table below gives basic data about each observation plus comments and a rough rating of the performance of each station.

The individual antenna performance ratings are based mainly on the tsm plots (displays of monitor data taken during the observations, mainly weather and flags) and first-look correlation results. The information is not complete for all epochs, especially for epochs not used in the 2018 paper. Also problems only identified during post-processing may not be shown. The table will likely evolve some if I do additional work on the project.

There are three main movies in the paper:

The movie that has attracted the most attention was based on BW088 which was observed in 2007 with an approximate 3 week cadence. The series of observations ran for a full year, but late in the sequence various issues complicated processing so the available movie only runs January through August. Data are available for the rest of the year that could probably be used to make reasonable images with enough effort, although all will suffer from the loss of SC which was being painted. Some were observed in a Multi Frequency Synthesis (MFS) mode where the bands were widely spread in frequency. Processing of those epochs has been somewhat problematic so far.

The second movie, based on BW090, used the faster cadence of about 5 days in 2008. Unfortunately the fast cadence made the use of dynamic scheduling to get good weather problematic. So this sequence suffered from reduced data quality compared to BW088.

The third movie was based on all observations spread over the 17 years for which we had data. For years in which multiple observations were made, stacked images were used in the movie.

Project BW082 was a pilot project to try to determine the best cadence. It sampled many intervals between observations. The 3 week cadence of BW088 was chosen based on these data, but unfortunately proved to be a bit too long, hence the shorter interval in BW090. A movie based on BW082 is available at the web site noted above.

The table includes some individual projects observed for other reasons that helped extend the time range to before the actual movie project was conceived.

There were a number of obervations made after the main movie observations that contributed to the long term movie and they are included. One of these projects, BW093 in 2010, had enough epochs to make a movie and that can be found on the web site mentioned above. Other projects experimented with frequent, short, observations to try to look for fast variations. Not much has come of that yet.

Finally a small number of observations made after the cutoff to make the 2018 paper are included here, but with incomplete data.

There are a number of observations whose data were not used in the 2018 paper. For some, the data quality was not considered adequate. Or key outer stations were missing leading to inadequate resolution for comparison with the other epochs. There were two projects that were attempting to get dense coverage with short observations, but where the short observations would require special processing that has not yet been developed (BW094 and BW115).

The Table of Array Performance for th 43 GHz VLBA Observations of M87

If the row is grey, the observation was not used in the 2018 paper.

Table columns:

Column 1: The VLBA project code.

Column 2: The year, month, and day of the observation. This is for the midpoint of the observation. Note that the dates in FITS and AIPS headers are for the start of the observations so they might not match. The dates and times are derived by AIPS runfile DSETDAT.ODW which contains the estimated offsets from the start to the center of each observation.

Column 3: The fractional MJD of the midpoint of the observations. This is a bit of an approximate estimate.

Column 4: The AIPS number under which processing was done on my computers. This helped with the bookkeeping of the very large number of AIPS data files generated. Since I was the only user on these machines, this did not create conficts. Any modern system would use subdirectories for this purpose.

Column 5: The number of resolutions used for CLEAN components in task IMAGR (NGAUSS). To find all the clean components, for example to use in a self calibration model (CALIB), you need this number of images with AIPS CLASS ICL001, ICL002, ICL003 etc. The first image (ICL001) has all the components restored and is the image used for analysis. But the attached CC file only has the point source components, not the larger gaussians.

Column 6: The number of iterations of the self calibration and imaging process used. The first few generally use phase-only self calibration. But most solve for both amplitude and phase. I generally kept going, changing parameters like the robustness nearly every iteration, until I just could not make more progress lowering the off-source RMS and I was within a few tens of percent of the thermal noise. A good data set using a good starting model (eg. previous epoch) would converge in roughly 10 iterations. If there were problems at some of the stations, convergence could be very slow and usually far from monotonic. But persistence can pay off. Part of the problem is that, with only 10 (or especially fewer) stations and significant large scale structure, plus the lack of short baselines to the outer stations, the self calibration/imaging process does not have much leverage to fix calibration issues. The situation is far different for the arrays with many more antennas such as the VLA and ALMA.

Column 7: The RMS off source in the image. The value was determined with the histogram fit option in task IMEAN using the whole field of view. The source is included as a tail in the histogram, but the off-source noise covers far more pixels and dominates the central part of the histogram over which the fit is being done. This method is less subject to direction dependent noise levels which are common in low declination sources (high noise north and south of the core). Of course, the off source noise measurement is different from, and lower than, the on-source errors in the image, which are hard to determine. Note that the dynamic range, measured as peak / off source noise, is typically a few thousand to about 10,000 for this project.

Column 8: The total flux density of the source. This is the sum of the CLEAN components. These match the values in Table 3 of the 2018 paper.

Column 9: The peak flux density. The peak in a version of the image convolved with an 0.43 x 0.21 mas beam in PA -16 deg. These also match the values in Table 3 of the 2018 paper.

Columns 10-19: An assessment of the data quality from each station. This is bases mainly on monitor data and quick look correlation data. If a cell is blank, such an assessment was not made. A presumption is made that anything with Tsys more than 40K above normal will be flagged. The concern with low level Tsys burbles is that they suggest unstable weather where coherence times might be short and phase referencing might be degraded. Color codes for the antenna information are:

Blank: Assessment not yet recorded. May be Ok in later epochs.

With 'Ok': No or minimal Tsys burbles. Less than 30 min lost to other causes.

Some Tsys burbles, but less than 20K. Less than 30 min lost.

Tsys burbles up to 40K. Less than 30 min lost to Tsys or other causes.

Between 30 min and 2 hr lost to Tsys or other causes

Between 2hr and half the observation lost to Tsys or other causes

Over half the observation lost to Tsys or other, but some data ok.

More than 90% of observation lost (usually all).

 

Column 20: A note about the status of the observation. These notes are incomplete and may not be current. I have not checked recently.

Column 21: This note relates to problems we had with the pulse calibration system.

Column 22: The date of the nearest VLA calibrator observations that could be used to assist with polarization calibration.