Rerunning the JVLA pipeline
-
(Recommended) You may download a template in our github repo. Be sure to define 'myvis' to the dataset you are working on.
-
Alternatively, you may download the 'cals' package from the NRAO data archive (the green icon below), which is about 1 GB for each JACKS SB. Unpacking it, you will find a CASA script called 'casa_pipescript.py'. This is the script we need. (Other files in the unpacked tarball include the weblog, calibration tables, and things needed for restoring the calibrated MS.)
-
Alternatively, you may create your own script from scratch, see this. The same script should work for any VLA data, with a correct definition of the 'mySDM' variable.
You need to download the raw data in the SDM format from the NRAO data archive.
-
Search for the project code '22B-193' in the search bar. You will see a list of observations like this:
-
Among the observations, only those with some 73 scans and sizes of 500 to 600 GB are successful scientific data. You can click the leftmost icon and select one or more observations, and then click the 'Download' button.
-
In the next interface, you may choose what data format to download. Since we want to rerun the pipeline from the beginning, here we may choose the raw (uncalibrated) data in SDM-BDF format:
-
You will receive an email once the data processing is done. Follow the link in the email to download the raw data.
The proper CASA version to use can be found in the weblog (either through corresponding github issues, or within the 'cals' tarball you download). In the example below, the version is CASA 6.4.1.12 with pipeline version 2022.2.0.64.
In the example above, we have:
- skipped Hanning smoothing (hifv_hanning), because we do not want to sacrifice spectral resolution. Later we may bin the channels if necessary.
- skipped RFI flagging (hifv_checkflag), because we do not want to unintentionally flag out lines (e.g., H2O masers).
- skipped adjusting data weights (hifv_statwt), because we do not give lower weights to spectral line data that have higher rms.
In the template in our github repo, the stages have been properly commented out.
Current VLA imaging pipeline only produces continuum images of single pointings (see VLA imaging pipeline). Therefore, we have to creat our own imaging script.
In the template script for continuum imaging, we take only the broad SPWs (128 MHz) and use the mtmtf algorithm to produce a continuum image. However the flagging is tricky, and is specific to each SB. We may simply comment out all the 'flagdata' lines to produce a first-look image.
The script makes use of the analysisutils package, which can be downloaded from here.
TBD