Observing Techniques

This page lists a few things that you might want to keep in mind when planning your IFS observations (writing your Phase I proposal), over and above the usual issues you have to consider for any other observing plan.

Proposal and Observation Planning

Acquisition and pointing

  • To acquire a faint diffuse region (that may not show up in a finding chart or via the normal acquisition procedures), you could offset from bright star. Bear in mind that if you want to be 100% certain of your pointing while you are observing, you will need to reconstruct the image of your field; for that you will need to have a visualisation tool handy (these not usually being provided on the computers in the control room of a telescope).
  • Be aware that the pointing accuracy of Gemini GMOS can be up to 5" off, even after offsetting from a bright star.
  • The IFU FoV will change with airmass and wavelength range due to differential atmospheric refraction (DAR). Be aware of this if you worked out your pointings in white light/a filter not in your wavelength range. More on DAR here.
  • If spatially dither your target you probably should dither your flux standard with the same pattern.
  • Changes in PSF with position in FoV as well as DAR effects will affect how well you can combine spatially dithered images.
  • If observing yourself, be aware that, because of the above effects your FoV may not be the same after changing gratings.

ETC (exposure time calculator)/ITC (integration time calculator) issues

  • If there is no option to input emission lines fluxes, one work-around is to specify an extended source, set the surface brightness to the flux of the emission line, and use a model power-law spectrum of index 0 (i.e. straight line). The S/N calculated will then only be relevant at the wavelength of the emission line.
  • Check whether the ETC returns the S/N per spaxel or per point source/PSF — these are not the same unless the PSF size is 1 spaxel size. To convert between the two you need to know the number of spaxels per PSF "element".

Overheads

  • IFS observations can incur large overheads — up to 30-40% in some cases (particularly with AO).
  • ESO-VLT imposes a limit to queue-scheduled observing blocks of 1hr. This can be quite restrictive (even 2 OBs in a row will repeat the set-up block and thus you will lose time on-target), but there is a possibility to request a waiver that would allow you to increase this limit.

Other

  • For echelle observations (true for any echelle not just IFU), if there are no arc with lines in your wavelength region-of-interest, you can use the solar spectrum from twilight flats (remembering that as the resolution and sampling change, the wavelengths of the lines change!).
  • If you are looking at a field containing both bright and faint sorces (e.g. whole galaxy), you can adaptively bin the spatial dimensions to maintain e.g. a constant S/N in each element. More on this here.
  • When defining the Phase II for VLT/FLAMES-IFU/ARGUS, you will need to create an FPOSS/FLD file for the fibre positioner. Help can be found here.

Sky subtraction

If you anticipate needing to subtract the sky background emission from your observations you will need to include observations of a region of blank sky in your programme. Is this available in your field of view or will you need sky offset fields?

Corresponding data reduction section

Nod & Shuffle (for the IR and beyond)

During your observations the telescope or the (light) beam direction is offset-onset-offset-onset… by an amount equal to the spatial separation between two fields, A and B. The result is that in exposure 1, field A samples the object while field B samples the background (actually on the opposite side of the object from the first background field sampled) and in exposure 2, field B samples the object and A the background, and so on back and forth. By forming linear combinations of the signal obtained from each element in exposures 1 and 2, the true object signal minus the sky background emission (and spectrum) can be obtained. Since the same physical elements have been used to sample both object and sky, the background signal can be completely eliminated. Since the background field is smaller than the object field, the field over which true beam-switching can be done is only as large as the background field. Sky-subtraction will still be possible over the rest of the field, although the precision may not be as great. This is something one can consider doing for IR and sub-mm observations.

Herschel, launched in May 2009, has an IFS as part of the PACS instrument. It will use a more complex version of nod/shuffle: nodding the telescope between two fields will be done together with high frequency, light-path chopping. This focal plane chopping is done to help eliminate the effect of the very fast modulation you get in the sky background in the sub-mm, with a variation time-scale too rapid to allow for full telescope nodding. However, when chopping you are using different parts of the telescope mirror and thus any temperature gradient across the mirror will lead to a ("sky") background temperature gradient. To allow this to be removed, nodding the telescope at a lower frequency than the chopping is done:
nod A - chop on : sky + source + telescope (beam1)
nod A - chop off : sky + telescope (beam2)
nod B - chop on : sky + source + telescope (beam2)
nod B - chop off : sky + telescope (beam1)
(nod A - chop on ) - (nod A - chop off) + (nod B - chop on) - ( nod B - chop off) = 2* source

For more information we suggest you read up on the instruments that use these techniques, such as GMOS (e.g. here). There is also an Astrophysics and Space Sciences Library book on Scientific Detectors for Astronomy 2005 which includes a chapter on IFS and the nod & shuffle technique (Roth et al.).

Splitting individual exposures

It is common to split observations into multiple exposures to aid removal of cosmic-ray hits during reduction. However, there are a number of further advantages to splitting up your exposures. Dithering means offsetting (either spatially or spectrally) by small amounts between exposures.

Spatial dithering

This can be done for the following reasons:

  • to compensate for dead fibres in the array
  • to fill in the gaps between spaxels where flux can be lost (especially important for fibre systems without lenslets)
  • to improve the spatial resolution (i.e. drizzling)
  • (and similarly to above) to improve the sampling of the PSF (e.g. if the PSF and spaxel size are not well matched). Ideally, your spaxel size will sample the spatial PSF well. If you expect a seeing of 1", then 1"-sized spaxels will not give you the ideal spatial profile for your sources (especially if point sources) and will also result in a worse flux calibration (because of flux falling between spaxels, and in a different way as the FoV changes with time and spectral range, during which the source will fall in a changing arrangement of gap/spaxel).

You will have the choice to dither by integer spaxel or fractional spaxel quantities. Clearly dithering by integer amounts makes reduction easier. Fractional dithers require sophisticated algorithms to properly combine the exposures; if you are using dithering to improve the PSF you will have to move by fractional amounts. Spatial dithers are usually executed in a set sequence, e.g. box or spiral pattern.

Dithering usually adds a considerable extra overhead to your plan, so make sure you really need it.

Corresponding data reduction section

Spectral dithering

This can be done for the following reasons:

  • to cover gaps in the CCD array (e.g. the Gemini GMOS detector is composed of 3 CCDs joined together with small gaps in between). If it is essential to have continuous spectral coverage, or a region of interest is predicted to fall on a bad section of the CCD, you might want to spectrally dither.
  • to help improve spectral resolution

Mosaicking

Mosaicking multiple fields together increases the field-of-view (FoV), and is useful for observing a source larger than the IFU size. You should consider the amount of overlap you want between the individual FoVs so that the images can be later properly combined. If there are strong variations to the instrument PSF over the detector you may need to be especially careful with this, so that your final mosaic is not more messy than any individual input dataset was.

Corresponding data reduction section

Flux standards

With IFU-based integral field spectroscopy it is possible to do flux calibration, so if you need calibrated fluxes then you must make flux standard observations. These should preferably close in time to your astronomical observations and at a close position in the sky (at a similar airmass anyway). It doesn't matter where the flux standard is on the IFU field-of-view, but better in the centre and/or somewhere where there are no dead pixels or spaxels.

Corresponding data reduction section

PSF (point spread function) variations

Almost all IFSs (in fact almost all instruments) have some level of variation (spatial and spectral) of the PSF over the detector. Thus it may be that there are slight variations in the intrinsic instrument profile across the detector and thus the FoV. For most IFS these variations are insignificant for most applications (otherwise it would constitute a serious design flaw!); however, if your programme requires very high spectral (and spatial) fidelity, this is something you may want to check out. All good instruments will have been designed to have an instrumental profile as close to Gaussian as possible across the entire FoV.

Corresponding data reduction section

AO coupling

Active optics (tip-tilt) correction will improve the effective PSF of your observations resulting in crisper images and better spectra. On the bigger telescopes this is automatically performed (using a nearby guide star)

Adaptive optics correction is now starting to be offered at a number of telescopes for IR observations with the latest IFS instruments (e.g. VLT SINFONI, Gemini NIFS). You may be able to use natural guide stars (NGSs) or need to supplement these with a laser guide star (LGS). Gemini, VLT, Keck, and the WHT now have LGS systems installed.

Fabry-Pérot

Some things to keep in mind if you are planning Fabry-Pérot observations:

  • The main difference to IFUs in the result and application is the large possible field of view (e.g. 5 arcmin @ 0.3 arcmin/pixel) at the cost of the small spectral coverage, namely a single line, mostly H-alpha in emission to study the ionised-ISM kinematics of galaxies.
  • One thing to be always kept in mind is the cyclic nature in the wavelength direction:
      • Light with wavelengths that match the neighbouring interference orders will also be transmitted, therefore a filter that matches the redshift of the object is needed.
      • Since these filters often are wide enough to allow several possible p, this means that the continuum (or other emission, e.g. the NII-lines beside H-alpha) from these will be "folded" into the measured wave-length range.
      • This also means that the absolute values of the measured velocities may be off by one FSR (free spectral range). A reasonably well determined systemic velocity for the object is therefore needed.

Corresponding data reduction section

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License