Data Processing: Bad Pixels

No attempt is made to interpolate over or correct damaged or contaminated detector pixels. Detector defects across the physical unbinned pixel array are instead flagged within the Current Calibration File (CCF), released by the XMM-Newton project. Image quality is ultimately propagated through the source detection process and attached to localized and flux-calibrated sources within the catalogue columns FILTER_QUALITY_FLAG, where FILTER is one of 'UVW2', 'UVM2', 'UVW1', 'U', 'B', or 'V'. Sources extracted partially or fully from known bad pixels will have additional unquantified uncertainties associated with their brightness and location. It is left to the investigator's discretion whether to include or exclude such sources from their data sample.

Science images may be both windowed (reduced in size) and binned. Associating science image pixels with cosmetically bad areas of the detector would be trivial except for an on-board "shift-and-add" process which is performed in-situ on the data. Shift-and-add is performed in order to correct for spacecraft drift during an exposure and subsequently optimize the time-averaged point spread function. Each photon is time-tagged as it arrives at the detector. On-board software identifies suitable tracking stars from the field, centroiding them at regular time intervals, and any positional offsets over time are used to adjust the arrival positions of every incident photon in discrete pixel units. This means that bad pixels tend to shuffle locally across a recorded image (the typical rms drift during an exposure is 1", or 2 unbinned pixels). The shuffling is accounted for on the ground when judging which image pixels are coupled to bad detector pixels using the tracking history, which is telemetered to the ground with the image data. One bad detector pixel will therefore generally affect several image pixels.

SAS tasks: omprep, omcosflag
CCF product: OM_BADPIX_000n.CCF