Offline IFU Data Reduction

Offline IFU Data Reduction

Lead Author: Watson Varricatt

NOTE: For any of this to work you must install the Starlink software, available from

NOTE: Additional tips are available from the UIST/IFU data-reduction web pages:

See bottom of page for further reading…

Creation of the Standard star datacube

This documentation assumes that you have already pre-processed your IFU data, using ORACDR.

ORACDR does the following:

  • flat fielding and masks bad pixels.
  • Arranges all the spectra from the slitlets of the IFU
  • Wavelength calibrates by cross-correlating with an observed arc
  • Scrunches the data, aligns the data from different slit-lets one above the other with the same wavelength along a column.

This is done for both object and standard. then:

  • Blackbody corrects the standard
  • Divides the object by the standard and flux calibrates

What it does not do is – interpolate across photospheric lines in your standard star. So you have to re-reduce the standard star datacube and interpolate across the absorption lines, and then divide the object by the standard. the required steps – collapsing the standard star cube to a 1-D slit spectrum, editing out the photospheric lines and then reconstructing the cube – are documented here.

First look at the reduced data directory there will be a file std_???_cub.sdf which is the blackbody corrected standard IFU spectrum.

Display the corresponding _im.sdf (white light image) using gaia and see within what range of X and Y axes you have reasonable counts on the standard star.

Producing a 1 D spectrum from the IFU CUBE

In the description given below —

  • x is the ‘short’ spatial axis of the IFU (length 14 units representing 14 usable ‘slit-lets’ of IFU. Each unit=0.24″ on sky or 2 pixel of UIST)
  • y is the ‘long’ spatial axis of IFU (length 54 units along each slit. Each unit=0.12″ or 1 pixel of UIST)
  • t is the wavelength/dispersion axis (length 1024 – Wavelength coverage depends on the grism selected.)

Step 1. Use figaro to collapse certain ‘constant-y’ planes look at the groupnum_im.sdf which is the white light image and identify between what ranges of x and y you have useful signal of the star.

Collapse (coadd) certain Y planes of std_???_cub.sdf and extract them. This will give you a thin sheet of thickness 1, width xmax, and depth 1024.

ystart = 0.5 (min) or give the start Y of the star image
yend = 53.5 (max) or give the end Y of the star image

This will give us an XT Plane

Step 2. From the thin sheet above, coadd the number of xt lines (‘constant-x’ lines) of interest to make a 1D spectrum of the standard star similar to what you extract from a single slit spectrum.

xstart=0.5 (or start X of the star image)
xend =13.5 (or end X of the star image)

The resulting spectrum can be plotted using linplot, splot, splat or splut

Step 3. Interpolate out the stellar absorption lines from the 1D spectrum

This can be done using the figaro task isedit

isedit will display your spectrum.

  • To interpolate a particular absorption line, first bring the mouse cursor to the left of the line, type “i”. Again bring the cursor again to the right of the line and press “i”. Now you can see that the linear interpolation is already done and plotted across.
  • press “r” to refresh the display and show the interpolated spectrum
  • Similarly, interpolate across all the photospheric lines that you want to remove.
  • press “q” to stop and exit. The interpolated spectrum is stored in the output file name that you provided at the command prompt while running isedit.

Next Step –

  • use the figaro utility bbody to create a blackbody spectrum at a temperature appropriate for the spectral type of the standard star, with the same wavelength coverage as the 1D spectrum that you created above, normalized at the central wavelength of the band.
  • divide the 1D spectrum created at the end of Step 3 by the blackbody spectrum.

Producing a CUBE (XYT) from the one dimensional spectrum

Now that you have removed the photospheric absorption lines from the standard and blackbody corrected it, you have to divide the object by the standard. But, now your standard spectrum is just a 1D spectrum and your object is a datacube. So, you have to re-grow your 1D standard spectrum to a cube and then, divide your object datacube by the re-constructed standard star data cube.

Step 1. expand the above 1D spectrum along the x axis

new true
ystart 1
yend 54

Step 2. grow it along the YT direction to make a cube

NEW true

The resulting cube will have the same standard star spectrum repeated throughout the whole spatial dimensions. Now you can divide the object data cube by the standard data cube created at the end of Step 2. After division by the standard, you can multiply the resulting data cube by the appropriate constants, taking care of exposure times of the object and the standard, known magnitude of the standard star etc. to flux calibrate it. From the flux calibrated cube (say of a cluster or an extended region) you can extract the spectrum of a particular star in the IFU field or a particular feature in your extended field, using the Steps 1-3 of the first stage described above.


OLD Documentation (may no longer work..)