Integral Field Spectroscopy vs Fabry-Perot Interferometry

Question

Why are Fabry-Perot Interferometers (FPIs) and Fourier Transform Spectrographs (FTSs less common than Integral Field Spectrographs (IFS) nowadays? My understanding is that:

• With FPIs and FTSs you have to scan your spectral range over time. This (A) Takes more telescope time, and (B) Is affected by seeing variations over time (i.e., different wavelengths end up having different PSFs).

• Limited spectral range. As a result, you have limited information about the observed object.

• I was also under the impression that reducing FPI or FTS data is more complex. However I do not have a reference to support this claim. It this true? If it is, can someone explain why?

• For some reason FPIs use Image Photon Counting System (IPCS) instead of CCDs. Why is that? Does this somehow make them "worse" than IFSs?

What other advantages does IFS have over FPI and FTS?

Answer 1

I think you've got it pretty well summarized. In particular, your first three points are dead-on.

(I'm going to use "IFU" = integral field unit instead of "IFS" = integral field spectrograph, but that's just because I'm more used to the former term; they're basically synonyms.)

Let me combine the "data reduction" question with one other key advantage of IFUs that you may have missed, which is that they are (almost) plug-in substitutions for traditional long-slit spectrographs, which means they can take direct advantage of existing spectrograph-design expertise and data-reduction software.

As an example, one common form of IFU is a closely packed array of small input apertures attached to fiber-optic cables. Away from the IFU, the cables can be re-arranged so that their output ends form a 1D row, very much like the output of a long slit in a traditional spectrograph. (And even more like a multi-object spectrograph design, where multiple individual input apertures are scattered across the field of view rather than being organized into a contiguous 2D array.) This means that the rest of the spectrograph design -- gratings or prisms, collimation optics, imaging onto a 2D CCD detector, etc. -- is pretty much the same as traditional spectrograph. There are instruments on some telescopes which allow you to switch between long-slit, multi-object, and IFU modes, with the same set of gratings, filters, and detector being used for all modes (for example, the GMOS instruments on the Gemini North and South Telescopes).

This carries over into ease of data reduction: the spatial+spectral image on the CCD detector (or near-IR detector array) produced by an IFU is very similar to that produced by a traditional long-slit spectrograph, so the same data-reduction techniques can (with a little modification) be used. In essence, you just chop up the big 2D data image you get from the CCD into a bunch of little sub-images, one for each fiber that came from the IFU, and you can treat each one as though it were a standard single-aperture or single-slit dataset. This contrasts quite strongly with the very different (unique) form of data and data reduction that F-P requires.