Fig. 1: UVES night sky spectrum (10 times binned in wavelength for better clarity) and the quantum efficiency for the ST-10XME.
It is evident that the night sky is very dark in the blue and green parts of the spectrum. It gets very much brighter in the red. A luminance filter will block the flux beyond 700 nm, which is detected (see the red QE curve) by a clear filter.
Thus an exposure through a clear filter will have more sky noise, but it will also record more flux from the objects (in particular those which emit significantly beyond 700nm). To decide whether one of the filters has a clear advantage over the other, I took some exposures in the SBIG clear filter and in the Astronomik Luminance filter, in the same photometric night and on the same target.
| Luminance | Clear | |
| Exposure time | 5x600s | 5x600s |
| Background level (indiv. exposure) | 440 ADU | 1020 ADU |
| Sky noise (indiv. exposure) | 16.3 ADU | 28.2 ADU |
| Sky noise (coadded image) | 8.0 ADU | 11.6 ADU |
| Relative flux scale | 1.396 | 1.000 |
| Sky noise (rescaled, coadded image) | 11.1 ADU | 11.6 ADU |
In order to compare the noise in both images directly, they have to be rescaled such that the objects have on average the same brightness. By comparing object fluxes, the scaling factor for the Luminance image is determined to be 1.4. The sky noise in both images then turns out to be very similar. That is, the additional object flux transmitted by the Clear filter is cancelled by the enhanced sky noise.
The advantage of the Clear filter is that it allows you to image objects with a very red spectrum, such as quasars at a redshift of 4 to 5. In a luminance filter such objects remain invisible, as e.g. Johannes Schedler had to find out for CFHQS_J1641. The disadavantage is that the night sky emission lines in the red can vary rapidly on time scales of 10 minutes, and on angular scales of a few arcminutes. This can lead to gradients in the image background and is the very reason why I am not using clear filters anymore, unless I want to image very red objects. This effect is probably only visible under very good skies when it is not superimposed by light pollution.
A Luminance filter does not suffer from this. A further advantage of a Luminance filter is that it suppresses chromatic abberrations, since not all lenses in the optical path are fully corrected to 900nm or 1000nm. That could mean sharper images.
The statistics above were obtained under extremely dark and transparent skies, and for one target close to the galactic plane. Due to extinction the rescaling factor of 1.4 is likely higher than for a non-extinted object, hence the luminance filter will have an advantage over a clear filter for different objects. The numbers (in particular the flux scaling factor) are therefore only indicative, not representative. Light polluted skies, or targets with different galactic extinction will yield different results.
Fig. 2: The rescaled coadded images for the Luminance filter (left) and the Clear filter (right). Click for a full-res image.