9 Calibration files

The calibration files described here are static calibrations which describe the arrangement of the IFU slice images on the array and the spectral lines used to wavelength calibrate spectra from each grism. Standard versions of these files are provided with Orac-dr and you should not need to make any measurements of these calibrations or make any alterations to these files.

9.1 The IFU profile

The lengths and separations of the images of IFU slices are assumed to be constant. The shifts which are required to align all the spectra in the dispersion direction ($x$-axis) to an accuracy of $\sim 1$ pixel and to align the slices in the final datacube in the $y$-axis are also assumed to remain constant. The file ifu_profile.dat in the $ORAC_DATA_CAL directory contains all of these values.

These values should not change (except after physical adjustments of the IFU), though refinements in measurement may well be possible.

9.1.1 File format

The file contains one line for each slice. Each line contains four numerical fields separated by white space (any number of spaces or tabs). The first two give the $y$ positions of the top and bottom of the slice, the other two give the $x$ and $y$ shift needed to align the slices. All of these should be integers except for the $y$-shift. Blank lines and any text following a # will be ignored. The slices should be listed in the order in which they are arranged in the two dimensional field of view, not the order in which they are arranged on the raw IFU image. This is important because the slices will be put into the final datacube in the order in which they are listed here.

9.1.2 Current values

After commissioning the file ifu_profile.dat contained the following values.

9.1.3 Measuring the calibration

The positions of the top and bottom of the slices can be measured in pixel coordinates from any IFU observation. There is a different vertical offset for each grism but you do not need to worry about this because the offset between the positions given and the actual spectra is measured each time a flat field frame is reduced.

The $x$-shift values can easily be measured from an arc spectrum. Values should be chosen to get the arc lines in the _ext file produced by the _EXTRACT_SLICES_ primitive as close to being straight as possible. Integer values should be used because the fine alignment will happen automatically when all the rows of the spectrum are scrunched to a common wavelength scale. Inaccurate values will shift the spectral lines making it difficult for iarc to wavelength calibrate all the rows. Using non-integer values will cause the spectrum to be resampled twice in the dispersion direction, which is neither necessary or desirable. Positive values will move the spectrum in the positive $x$ direction.

The $y$-shift value is the most difficult to measure. This shifts each slice along its length which is necessary for accurate image reconstruction. This can be measured by using telescope offsets to scan a star across the field of view of the IFU. The star should be stepped across the field in steps of approximately one slice width (0.25 arcsec) then returned to the initial position to check for systematic drifts. The integrations should be long enough to smooth the seeing effects. The datacubes or extracted images can be added together to produce a horizontal line across the field of view. The offset from one slice to another can be measured and offsets applied. A positive value of the $y$-offset will move the slice in the positive $y$ direction. Note that the plate scale varies slightly from one slice to another (maximum variation $\sim 3\%$, corresponding to $\sim 1.5$ pixels over the length of the slice) so the offsets should be measured at the centre of the slices along their length. Alternatively they can be measured at two positions equal distances from the centre of the slices in the $y$ direction and the average offset used.

9.2 Wavelength calibration

Automatic wavelength calibration is significantly easier with a grism based spectrometer such as UIST than it would be with a grating based spectrometer such as CGS4. This is because the wavelength calibration of each grism should be constant to within a pixel or two, whereas the central wavelength of a grating can be varied continuously over a wide range.

For each grism there is at least one file in $ORAC_DATA_CAL/grisms/ listing arc lines (there may be more than one if the grism can sensibly be calibrated with either lamp). These files have the extension .lis and are in the format of the arlines.lis file produced by the Figaro arc task. The appropriate file is found by the calibration system and copied to $ORAC_DATA_OUT/arlines.lis where it is used by iarc to generate a wavelength calibration for each row of the image.

These arc line lists will need to be re-measured after any change which will shift the spectra on the array. This includes changes in the optical alignment of UIST (removing and replacing the grism wheels, slit wheel, camera wheel or the array). It may also be necessary to generate new calibrations if the $x$-shift values in ifu_profile.dat are changed significantly.

9.2.1 File format

You are unlikely to need to worry about the format of the arc lines files. These files are produced by arc and used by iarc. For details of the format please see the Figaro documentation.

A typical file looks like this (the arc lines for the short-$J$ grism with the argon lamp):

The only part of this file to be directly used by any of the recipes (rather than by a Figaro task) is the final line, giving the polynomial wavelength fit, which is used to calculate the wavelength range to which the spectra should be scrunched.

9.2.2 Measuring the calibration

To generate a new arc lines list file you will first need to reduce a flat-field frame as normal, then reduce an arc using the EXTRACT_SLICES recipe. This will extract the slices from the array, approximately align them in wavelength and apply the flat-field. The frame is now in the state that arc frames will be in when they are wavelength calibrated.

Extract a spectrum from the central slice, (or the one before if there are an even number of slices), so for UIST extract a spectrum from the 7th slice from the bottom. This is where iarc starts its fitting. Copy the spectrum to $ORAC_DATA_CAL/grisms/arcs/ and give it a meaningful name (e.g. short_K_Ar.sdf). This is not essential because the spectrum is not used for calibration but it will include all the headers describing the instrument configuration at the time, which may be useful for future reference.

Use Figaro arc to wavelength calibrate the spectrum. The wavelength calibration should be in Angstroms. A 2nd or 3rd order fit is likely to be most appropriate. arc will generate a file called arlines.lis. This is the arc lines file that you want. Give it a more useful name (e.g. short_K_Ar.lis) and put it in $ORAC_DATA_CAL/grisms/.

In order for the calibration system to know about the file you will need to add it to the index.arlines file which you will find in the $ORAC_DATA_CAL directory.