Calstis version 2.6 was installed in the pipeline on Feb 14, 2000 and is also being used by OTFC. Calstis 2.6 is available for download in the new STSDAS release Version 2.2 which can be found on the STSDAS Advisories page.

Changes to CALSTIS in this version include a number of bug fixes, enhancements, and a number of new user tools and scripts.

The following OPRs are closed by the release of CALSTIS 2.6:

OPR 39951: CALSTIS8 must normalize by the sum of exposure times
OPR 39952: CALSTIS11 must include gain when scaling science image
OPR 40019: CALSTIS should check for wild MOFFSET shifts
OPR 40030: CALSTIS should do BLEVCORR before error array initialization
OPR 40224: In calstis, use table interface to check for existence of files
OPR 40465: For echelle data, CALSTIS4 should find global offset
OPR 40466: CALSTIS12 should interpolate shifts over time

Below we give more details on changes that are likely to be noticed by users in their data.

Correct flux calibration of repeat-obs images: When Calstis8 is run on flux-calibrated spectroscopic data, the input image sets are now averaged (rather than summed), using the exposure times of the individual images as the weight. Earlier, the image sets were being averaged and the resultant image was being normalized by dividing by the exposure time of the last image set.

Taking CCD-GAIN correctly into account in scaling the images: For wavecal exposures taken with the external shutter open, Calstis11 subtracts the science image from the wavecal. Calstis11 was modified to get the CCD gain from the science and wavecal primary headers and include the ratio of gains in the scale factor. That is, the science image is now multiplied by this additional factor (science_image_gain / wavecal_gain) before subtracting it from the wavecal image.

Estimating MOFFSETs correctly: (1) MAMA data are intentionally offset by different amounts on a rotating schedule in order to distribute the charge extraction over a wider area of the detector. The offset can be up to 20 to 40 pixels in each axis. Calstis used to rely on header keywords MOFFSET1 and MOFFSET2 for the size of the shift in each axis. These keywords were occasionally incorrect, however. With echelle data, if MOFFSET2 was far enough off, calstis would assign the wrong spectral order number, which resulted in wavelengths that were off by of order six Angstroms. In order to work around this keyword problem, the wavecal processing for echelle data was rewritten so that it measures the combined offset (Mode Select Mechanism offset plus MOFFSET), and the SHIFTA1 and SHIFTA2 keywords now give this total offset. This will be much more reliable. (2) The second change is an enhancement for processing of GO wavecals. When more than one wavecal is stored in the same _wav.fits file, the SHIFTA1 and SHIFTA2 are determined for each wavecal observation. When updating the shift keywords in the science file (_flt or _crj), the current version interpolates the shifts linearly with time, while the previous version took the shifts from the wavecal that was nearest in time. This does not affect auto wavecals because there will be only one wavecal observation per file.

CALSTIS corrects the bias level BEFORE error array initialization In the basic2d portion of calstis, the error array was being computed before BLEVCORR is done, and the value of CCDBIAS from the CCDTAB was used when computing the errors. The result was that the error array values were too large. Calstis now does BLEVCORR before computing the error array values.

CALSTIS12 now interpolates wavelength shifts over time Calstis12 is a part of wavecal processing. Calstis4 determines the shift in each axis and records the values in the SHIFTA1 and SHIFTA2 keywords in the wavecal header. Calstis12 copies those shifts from the wavecal header to the science file header. In the case that the wavecal file contains more than one imset, calstis12 was selecting the imset that was closest in time to the science exposure. The option to use linear interpolation has been added, which is now the default.