Australia Telescope National Facility

Next: Step 4: Postprocessing to Up: How Pieflag works Previous: Step 2: Rms-based flagging

Step 3: Postprocessing to find small clusters of bad points

If both amplitude- and rms-based flagging have been applied, Pieflag's internal tables contain information about the badness of some points as derived by amplitude-based flagging, and information about which parts of the data should be flagged as derived from rms-based flagging. To turn the badness values into flags, the running sum of the badness values in a 1min window is calculated in each channel and on each baseline. If the sum exceeds 1, the window is flagged. This procedure will tolerate single, moderate outliers (one value with badness 1 within one minute), but will flag the entire window if two moderate outliers, or one point with badness 2, occur within a minute. It also flags a small margin around the bad data. This algorithm is applied irrespective of pointings, because it is assumed that data are affected irrespective of changes of pointings, at least on such short timescales.