February 4, 2013
CALCOS RELEASE NOTES
About This Article
This describes changes from calcos version 2.18.5 to 2.19.6 (version 2.19.7 is the same as 2.19.6 except for the installation procedure).
|616||66158||Calcos should be modified to use Sphinx for documentation.|
|720||68550||Calcos should accept data taken with different cenwaves.|
|918||71634||G140L Segment B only: no SHIFT applied|
|919||71762||Calcos should read the TIME column as double.|
|924||71881||Calcos should use the HAS_LINES column in the LAMPTAB.|
|925||71949||Calcos should use the FPOFFSET column in the DISPTAB.|
|927||71997||Calcos should create the output directory.|
|928||71998||Calcos should use the Gehrels variance function.|
|930||72008||Calcos should use fp_pixel_shift for wavecals.|
|936||72063||Calcos should ignore difference in SPWCSTAB for a wavecal.|
|938||72065||Calcos should include error estimate in the weight.|
|939||72066||Change the sign of calcos variables mindopp and maxdopp.|
|940||72067||Calcos should broaden flagged regions in the DQ array.|
|943||72166||Change the tests in calcos for an empty table.|
|946||72494||Calcos should copy association file to output directory.|
|960||72794||Calcos should do the Doppler correction after csum.|
|963||72824||Bug in calcos normalization for average wavecal shifts.|
- If the user specified an output directory but that directory doesn't exist, calcos will create it.
- If the input is an association file and an output directory was specified, the input association file will be copied to the output directory, and the copy (rather than the original) will be updated. This allows running calcos on write-protected data.
- If the lamptab has a HAS_LINES column, the value in this column will be used to determine whether to copy wavecal information from segment A to segment B data. This was previously done only for G140L data. If the disptab has an FPOFFSET column, this will be used (in addition to other keywords) to select the appropriate row of the disptab.
- The Gehrels variance function is now used to compute the error estimates for the flt and counts files and for the ERROR column in the x1d and x1dsum tables.
- When averaging x1d spectra to an x1dsum, the flat field is included in the weight. The primary reason for this is to give less weight within the grid wire shadows.
- When flagging regions as bad in the DQ extension, the regions will be expanded by a quarter of a pixel in each direction. This reduces the number of cases where a pixel should have been flagged as bad but wasn't due to roundoff to integer pixel number.
- In the case that there is a non-tagflash exposure with no auto/GO wavecal at the same fpoffset, another wavecal in the association can be used, and the correction due to different fpoffset will be taken from the FP_PIXEL_SHIFT column in the lamptab.
- The Doppler correction and filtering by pulse height are now done after the call to write the csum file, and checking for FUV bursts is now done after wavecal correction. The Doppler correction uses the dispersion and 1-D extraction reference tables (DISPTAB and XTRACTAB), and there may be situations where these tables do not contain rows for some mode, but it is desired to do some calibration (e.g. GEOCORR) and write a csum file before stopping with an exception. It was possible to move Doppler correction because the csum file does not depend on it; however, the primary header keyword DOPPCORR will now be 'PERFORM' instead of 'COMPLETE' in csum files. Checking for bursts makes use of the Y locations of the source and background regions, and these will not be known until after wavecal correction, so moving the burst correction should improve the results, i.e. version 2.19.6 should be less likely to find bursts unless they are real.