These scripts provide for imaging of LOFAR Radio Sky Monitor (RSM) data on the Lisa cluster.

• The Offline and LofarFT packages from the LOFAR software repository;
• The "special" version of awimager from r22783 of the LOFAR-Task3482-imager branch of the LOFAR repository (until this branch is merged to trunk).
• pyrap;
• NumPy.

Each unit of work consists of a pair of observations: a calibrator and a target. The calibrator observation consists of a number N of subbands directed towards a known calibrator source. This is immediately followed by N * M target subbands, tiling out a large area of the sky around the zenith with M beams. Each of the calibrator subbands has a 1-to-M relationship with the target subbands at the same frequency.

In recent observations N = 40 and M = 6, but this is potentially variable in future.

A skymodel is required for both the calibrator and for each beam of the target (M + 1 skymodels).

As part of the processing, our N * M target subbands are aggregated into groups of X subbands. Currently, X = 10, but this is potentially variable.

The required outputs are:

• Calibrated data for each calibrator subband (N MeasurementSets);
• Calibrated data for each target subband group (N * M / X MeasurementSets);
• Image data for each target subband group (N * M / X images).
• Logs from each of the processing steps.

The processing of each unit of work is carried out by the script imaging-multibeam.py. It takes a single command line argument: a parset which contains all the configuration information it needs. It performs the following workflow:

1. All input data is copied from shared storage to appropriate scratch areas for processing.
2. A separate BBS process (calibrate-stand-alone, from the LOFAR imaging repository) is invoked for each subband of the calibrator observation.
3. The resulting instrument databases are clipped using the script edit_parmdb.py.
4. The clipped instrument databases are transferred from each calibrator subband to each of its M corresponding target subbands using parmexportcal and calibrate-stand-alone (both from the LOFAR imaging repository).
5. The target subbands are combined in groups of size X using NDPPP (LOFAR imaging repository).
6. For each group:
   1. Phase-only calibration is performed using calibrate-stand-alone.
   2. Bad stations are identified and using asciistats.py and statsplot.py from the LOFAR repository and then stripped from the data.
   3. A limit is set on the length of the longest baseline to be included when imaging.
   4. A temporary, "dirty" image is constructed using awimager (LOFAR imaging repository trunk) and used to calculate the threshold to be used for cleaning the final image.
   5. A "mask" is constructed, based on the contents of the appropriate skymodel, using the msss_mask.py script included in this repository.
   6. The final image is constructed using awimager (from LOFAR repository branch LOFAR-Task3482-imager).
   7. Required metadata is added to the outpit image using addImagingInfo from the LOFAR imaging repository.

For any given RSM run, a large number of work units will be produced (typically 96 in a 24 hour period). Each of those is processed completely independently according to the above procedure. The script generate.py takes observation IDs for the calibrator and the target field, and the file name of a template configuration parameterset, as command line arguments, and generates a job suitable for submitting to the Lisa queue which will process one work unit.

A small number of sources are eligible for use as calibrators, and the skymodels for these have all been pre-calculated. However, each of the M beams in the target requires a skymodel specific to its observation direction. This skymodel can be generated on LOFAR CEP using the script gsm.py. The script skymodel.py, included in this repository, takes a list of directories containing target observations as command line arguments, and prints to standard output the gsm.py invocation required to generate an appropriate skymodel.