Change page style: 

Observing Conditions

Sky conditions can have a large effect on mid-IR observations. On this page we provide advice about choosing filters and some brief information about mid-IR seeing. This page will be updated and expanded as we gather statistics and information from actual Michelle and T-ReCS observations. A document showing some preliminary results is now available (Mason et al., submitted to SPIE).




Clouds and saturation

Thermal emission from clouds can saturate Michelle and T-ReCS' detectors, especially in imaging and low-resolution spectroscopic modes. The lengths of individual exposures are optimised for good conditions and are not adjusted on-the-fly to compensate for the increased and variable background present in cloudy conditions. The Observing Tool places no restriction on PIs' choice of cloud cover constraints for queue observations, but PIs should be aware of the likely degradation of data quality resulting from choosing CC70/90/Any for many thermal IR observations.




Atmospheric transmission and choice of filters

As is evident from plots of the atmospheric transmission at 7-25 µm , the transmissions of the different Michelle and T-ReCS filters are affected differently by water vapor, ozone, methane, CO2, and other atmospheric gases. Some of the filters are located in very clean portions of the transmission spectrum and others are not. While some science programs require the use of many filters, others need only one or two and are interested mainly in sensitivity. The transmissions of some filters vary strongly with airmass, while others do so only slightly.

We have put together a table of transmittances and relative signal-to-noise ratios for many of the Michelle and T-ReCS filters, as functions of airmass and water column (as given by ATRAN). The following general conclusions can be drawn.


  • The cleanest filters in the 10um window are Si-2, Si-5 and N'. Si-2 is the most sensitive filter for detecting sources whose mid-IR spectral energy distributions fall off rapidly with increasing wavelength (e.g., stellar photospheres). For objects with flatter spectra and for highly reddened objects, N' is preferred as it has nearly identical transmittance to Si-5 but has more than twice the bandwidth.
  • By far the cleanest 20um filter is Qa, although it doesn't compare to any of the better 10um filters.
  • Filters whose transmittances are highly dependent on water vapor are Si-1 and all Q filters. Even when the water column is low their transmittances are not high, even at the zenith. Thus, in general measurements using these filters should be made close to the meridian (i.e., near the lowest airmass) whenever possible and they should not be made on wet nights. Although there are a small number of very bright targets that can be observed in the Q-band even in relatively wet conditions, nearly all Q-band observations need to be done in conditions of low water vapour.
  • The Si-3 and Si-4 filters are affected strongly by ozone, which varies in an irregular manner, but only slightly by water vapor. Measurements with them should be made close to the meridian whenever possible.
  • The very wettest conditions can produce excess noise in the mid-IR - generally known as "sky noise" well above that produced under photometric conditions. It is likely that this "sky noise" is caused by fluctuating regions of condensation (i.e., very thin clouds) passing through the telescope beam. Thus, in general, "WV any" conditions should be avoided for mid-IR observing.

The integration time calculators can be used to investigate the effect of weather conditions on signal-to-noise ratios. PIs might also like to remind themselves of the observing condition percentile definitions.


Mid-IR seeing

The seeing at N and Q is usually much more stable than that at shorter wavelengths, being commonly around 0.4 arcseconds at N and 0.6 arcseconds at Q (the diffraction limits are ~0.3 arcsec at N and ~0.6 arcsec at Q). In conditions of good seeing one or two (and sometimes even three) diffraction rings around a point source may be seen. Typical strehls for good conditions are 0.6 for the narrowband filters in the N-band window and 0.9 for the Qa filter at longer wavelengths. Only occasionally is the seeing much worse than this at N-band or Q-band, although seeing of 1 arcsecond or more is not unknown in the N-band. When this happens the seeing tends to be extremely poor at shorter wavelengths.