3.6  Correction for geometric distortions, using VDS_ROTATE / NIS_ROTATE

Unfortunately, it is not possible to make in-flight adjustments to CDS to correct for, e.g. distortion during launch. This has resulted in a series of effects, that can only be partially corrected for during the data analysis.

First, it seems that the NIS instrument is not in focus. In fact, the line widths created by the 2^¢¢ and 4^¢¢ slits appear to be about the same, and much larger (by factors ~ 2) than in pre-flight measurements. No correction is possible.

Another optical factor to take into account in the data analysis is the fact that the centres of the various slits are not co-aligned, which has the consequence that images taken will not be aligned in the Y-direction. However, images taken with different slits by the author show that the 4^¢¢× 4^¢¢ and the 4^¢¢× 240^¢¢ slits are co-aligned, in accordance with results based on other work (D. Pike, priv. comm.), which have shown for the other slits misalignments of up to 4^¢¢.

3.6.1  Slant: rotation and tilt

Then, the spectra are slanted, in the sense that the dispersion direction is rotated relative to the rows of the VDS CCD, with different angles for NIS 1 and NIS 2, as shown in Figure 3.6.1. This is the main effect, and is caused by a slight misalignment between the grating and the detector.

#1 The NIS spectra, showing the effects of the slant of the spectra and the cosmic rays. First three spectra, from top to bottom: a NIS 1 de-biased spectrum; the same spectrum with the cosmic rays flagged as missing data, calibrated using VDS_CALIB (and NIS_CALIB), and rotated (using VDS_ROTATE); the previous spectrum, but with the unusable spatial areas (top and bottom rows) removed. The bottom three figures show the NIS 2 spectra. The particular exposure shown here was taken during a brightening in a spatially-localized region of the network. The bright stripe across the spectra is caused by enhanced emission in all the emission lines, plus by scattered light, and clearly shows the slant of the spectra, that is more pronounced in NIS 1. Note that a large portion of the spectra (along the slit) is lost after the correction for the slant is applied.

It implies that a row of CCD pixels covers different SolarY values at different wavelengths.

On top of this, the spectral lines are tilted relative to the dispersion direction (i.e. the angle they for with the dispersion direction is not 90 degrees, and varies). It is caused by a small misalignment between the grating and the slits. It implies that the position of a spectral line (measured in CCD x-pixels) will vary along the slit (SolarY direction) and would give a false velocity variation in the N-S direction.

Figure 10: [Figure from CDS user guide]

The slant of the spectra complicates the planning of a CDS `study' and data analysis.

A correction for the slant and tilt of the lines can be done in two ways, one during the data planning and one in the analysis (by the routine VDS_ROTATE).

During the planning of a CDS `study', the CDS routine MK_RASTER makes the following adjustment. If a study is designed with specific data extraction windows then the windows are positioned on the detector at different row positions to allow for the slant. Hence pixels along the slit of different data windows correspond (within one pixel) to the same solar location and the correction for the slant of the lines is not necessary, although the approximation is about one pixel in Y. Using this method, data extracted at any of the wavelength windows are co-spatial on the Sun to within one pixel in Y. [The usual extraction window height is 143 pixels, corresponding to 240^¢¢.]

For studies that use the full default extraction window the correction for slant has to be applied during data analysis, and it is important in the planning phase to extend the Y-size so that a sufficiently large area of the detector is extracted, to cover the full rotated spectrum. In the data analysis, VDS_ROTATE is used to rotate the entire NIS 1 and NIS 2 spectra to correct for the slant and tilt. In this category there are two further types of CDS `studies', one that extracts the full spatial positions along the slit (as done in the spectral atlases) and one with a shortened extraction along the slit (case shown in Figure 3.6.1).

In the case of spectral atlases, two windows of 1024x160 pixels are extracted. VDS_ROTATE pivots around the centre of the data. There is a difference in the height of the unrotated data for NIS1 and NIS2, where the first is 160 pixels in height, and the second is only 153 pixels. Therefore, NIS1 data pivots around vertical pixel 79.5, while NIS2 data pivots around pixel 76. Thus, there is a 3.5 pixel difference in the positions due to that effect. VDS_ROTATE then cuts the two spectra to 143 pixels, although two arrays of 1024x160 pixels are given in output. However, when the /BOTTOM keyword is used, then the bottom of the slit is aligned with the bottom of the array, and the data should be directly comparable.

In the other case, with shortened extraction along the slit, there is no difference in the height of the unrotated data for NIS1 and NIS2, and the spectra are rotated around the centre of the data.

In any case, the slant of the two channels is such that about 30^¢¢ along the slit direction are lost, if one is interested in having all the NIS spectral information at the same spatial position. Figure 3.6.1 shows the rotated NIS 1 and NIS 2 spectra.

The tilt of the lines can be partially corrected for by applying the routine VDS_ROTATE (or NIS_ROTATE) to the spectra, or by adjusting the wavelength calibration.

Note:

NIS_ROTATE, QLDS

Finally, note that the spectral line tilt may have changed post-recovery.