rgsbadpix combines the calibration bad pixel data from the CCF with any hot pixels found by analyzing the raw telemetry for one CCD from an RGS exposure. The result is stored in the BADPIXn extension tables of the input dataset, or in a separate dataset if requested (withbadpixset=yes). When the same bad pixel is identified by both the CCF and the telemetry analysis, the output bad pixel table mentions only the CCF version. CCF bad pixels that lie outside the telemetry window of the exposure are not included. The columns of the BADPIXn tables describe these properties: location, type and source. Location is specified in chip-oriented coordinates, and describes a column segment of similar bad pixels in terms of its minimum y-axis coordinate and upward extent. The type codes are described in the CAL documentation. The source codes are as follows:
A bad pixel obtained from the CCF, and listed there as having been uplinked to the on-board data preprocessor (DPP). Such pixels should never appear in the telemetry, and should therefore never be identified by the telemetry analysis. There is no option to exclude these from the output bad pixel tables; to do so would corrupt the exposure map.
A bad pixel obtained from the CCF, but not listed as having been
uplinked to the DPP. If marked as hot ("H") in the CCF, these are regarded
as advisory, and may be
excluded from the output bad pixel tables (withadvisory=no). The
default is to include them.One would hope that any of these listed as hot would also be
identified by the telemetry analysis.
In addition defective columns, showing often a larger CTI and therefore
appearing as "cooler" than they should be, can be included in the list of bad
pixels (keepcool=yes) as well. They are listed in the corresponding
CCF (RGS[1][2]_COOLPIX_xxxx) as "D".
The default (so far) is not to treat them as bad columns.
A bad pixel identified by analysis of the telemetry. The current algorithm only attempts to find isolated hot pixels and columns. This process is optional and may be disabled (withfoundhot=no).
Four parameters control the hot pixel finding algorithm, which is an implementation of John Peterson's memo, RGS-COL-CAL-00015. First the pixels are analyzed individually for excessive activity and then the same is done for whole columns. pixnoiselimit and colnoiselimit respectively are the minimum uncalibrated energies for considering single-pixel events in the first and second phases. pixsharpness and colsharpness respectively control the degree of excessive activity required to flag a pixel or column as hot. Refer to section 8 for the precise meaning of the sharpness parameters. Frames marked with IN_BAD_FRAME in the FLAG column of the EXPOSURE table are considered bad and their pixels are discarded.
In High Time Resolution (HTR) mode the entire cross-dispersion
dimension is collapsed into one row; hence it is not possible to
consider individual bad pixels. If hot pixel finding is enabled the
hot column finder works just the same way as in Spectroscopy mode but
the initial search for individual hot pixels is skipped. Throughout
this document a 
marks items that do not apply to HTR mode data
and a 
marks items that apply only to HTR mode data.
