14.6 Image Mode Geometric Distortion

The STIS cameras have significant geometric distortion that not only affects astrometry, but also in principle affects photometry (because the extended sources used to generate flat fields have an induced change in the apparent surface brightness). For the CCD the image distortions are less than one pixel across the whole detector, and can often be ignored. For the MAMA the distortions are larger, approaching 3 pixels at the corners of cameras.

Geometric distortion and plate scales for the STIS imaging configurations have been measured on-orbit by observing star fields shifted to different positions in the field, following a procedure similar to that used for WFPC-2 (Holtzman et al., PASP, 107, 156). The geometric distortion data also allowed a determination of the mean plate scale at the center of the field for each detector. These plate scales, for unfiltered CCD exposures and filtered MAMA exposures (see ISR STIS 2001-02), are given in Table 14.38. The quoted errors are formal random errors derived from the uncertainties in measuring the positions of the sources. Maíz-Apellániz has determined that the plate scale for unfiltered FUV-MAMA images (25MAMA) is 1.0031 times larger than for filtered images (F25QTZ and F25SRF2); the plate scale for unfiltered NUV-MAMA images (25MAMA) is 1.0008 times smaller than for filtered images (F25QTZ and F25CN182).


Table 14.38: Mean pixel scales for STIS detectors.

Detector

X scale
(arcsec/pix)

X error
(arcsec/pix)

Y scale
(arcsec/pix)

Y error
(arcsec/pix)

Corrected
mean scale

Unfiltered CCD

0.050716

0.000072

0.050735

0.000087

0.050725

Filtered FUV-MAMA

0.024328

0.000025

0.024608

0.000086

0.024468

Filtered NUV-MAMA

0.024526

0.000120

0.024829

0.000126

0.024677

Distortion solutions have been implemented in the IDCTAB reference files (see ACS ISR 2001-08). The equations which relate the true xi, yi positions of the stars with the observed positions x, y are as follows (STIS ISR 2004-01):

x_i - x_r = \sum^k_{i=0} \sum^i_{j=0}c_{x,i,j}(x-x_r)^j (y-y_r)^{i-j}
y_i - y_r = \sum^k_{i=0} \sum^i_{j=0}c_{y,i,j}(x-x_r)^j (y-y_r)^{i-j}



where xr and yr are the coordinates of the central pixel.

Maiz-Apellaniz and Ubeda discovered that the coefficients for the NUV-MAMA had been incorrectly implemented in the original IDCTAB reference file and performed a revised analysis (ISR STIS 2004-01). A similar re-analysis for the FUV-MAMA has recently been performed by Sohn (2017 August STAN). Both analyses employ a fourth-order transformation between raw and rectified coordinates, using the mean plate scales given in Table 14.38. The plate scales imposed on rectified images produced by calstis using the latest IDCTABs (column "SCALE") are given in Table 14.39.


Table 14.39: Plate scales of rectified and drizzled images.

Detector

IDCTAB

Plate Scale (arcsec/pixel)

Clear

Filtered

CCD

o8g1508do_idc

0.05072

0.05075

FUV-MAMA

1771408ro_idc

0.024395

0.024468

NUV-MAMA

o8g1508fo_idc

0.024604

0.024678