What happens when the pipeline runs? (part I)
The data reduction (processing) consists of four stages: Loading, Editing, Calibration and Imaging. These will be familiar to radio astronomers, but possibly not to other users, so they are described below. For a detailed description of how Compact Array data is reduced, see either the Miriad Users Guide or the Guide to ATCA Reduction in the AIPS++ Cookbook. If there are any terms you are unfamiliar with, the AIPS++ Glossary may help.
Loading
Also called filling, this stage involves fetching the files from the archive and loading them into the AIPS++ native format (Measurement Sets). This is similar to loading the ATCA data files (RPFITS files) into a Miriad "uv" file using atlod. The filling process also involves:
- removing self generated birdies in 128 MHz data by deleting every other channel and the edge channels
- suppressing ringing of interference spikes by reweighting the lag spectrum for minimum spectral sidelobe response
These are the birdie and reweight options in the AIPS++ atcafiller (also the same in Miriad's atlod). For really detailed information about what is going on behind the scenes see the AIPS++ documentation for atcafiller. To learn how you would do filling using AIPS++ tools directly, see the (more friendly) cookbook documentation.
Editing
This is the process of removing bad data points (which could be caused by things such as weather, interference or malfunctioning hardware) that may affect the quality of the calibration and imaging. In radio astronomy this is commonly referred to as `flagging' and is often done interactively.
In the pipeline this is implemented using the AIPS++ autoflag Module. This provides a range of functions such as time-median filtering and frequency filtering that can be used to target different kinds of outliers in the data.
Prior to calibration, the primary and secondary calibrators are edited using time-median filtering to remove outliers that will affect the calibration. Several plots of the raw and editing data are produced for you to look at. These include Amplitude vs. Time, u-v coverage and Amplitude vs. u-v Distance and will allow you to assess the basic quality of the data.
After calibration, the target sources are edited, using a very conservative u-v binning method. This ensures only extreme outliers are removed, without affecting the genuine fluctuations in the visibilities that occur with extended sources. More plots are produced post-calibration so you can determine whether the calibration has been successful or not. These include Amplitude vs. Time, Phase vs. Time and Real vs. Imaginary components plots.
All the plots produced are discussed further in the interpreting the results section of this documentation.
|
