3.2 Spatial resolution, aperture bias and matching

One of the main tasks when assembling a multi-wavelength SED, indeed any catalog that contains more than one measurement, is to control whether what is measured in each band is actually physically the same. Due to either the seeing of the atmosphere or the diffraction of the telescope, the flux from a point-like source is re-distributed over the point spread function (PSF) of a width that typically depends on the time of observation and on the wavelength used. Moreover, galaxies are intrinsically extended and their morphology may depend on the wavelength in which they are observed.

One of the main problems in the process of matching is the size of the PSF. Typically, the PSF is narrowest at optical wavelengths, while UV and IR PSFs are broader. This can lead to situations in which there is more than one optical counterpart to the UV, IR or sub-mm source. The agnostic way to deal with this is to simply exclude such objects from the sample, however, this may introduce a bias if the multiple optical counterparts are actually physically associated. A more intricate, but also more uncertain, way is to redistribute the flux according to optical priors (Guillaume et al. 2006). Finally, the use of all available information, spatial as well as spectral, seems to provide a promising way forward for multiwavelength datasets (Roseboom et al. 2009).

Another problem related to the resolution of the telescope are aperture biases. A rather simple manifestation of this is that objects that are further away will be seen as smaller on the sky. Therefore, in order to construct comparable samples at different redshifts, one has to adapt the size of the extracted aperture to the same physical size. A more complicated problem is the definition of the “total light” from an object. Indeed, the surface brightness profiles (SBP) of galaxies usually extend much beyond the threshold observational surface brightness. In the case of specific objects, such as cD galaxies, these extended wings might contain a significant part of the total light from a galaxy (Oemler 1976Carter 1977). Different strategies have been developed to avoid these biases, such as either integrating over a full model for the SBP or simply using specific apertures to integrate the light only inside some physical radius, but each method has its own problems. One of these is also that galaxies have different intrinsic morphology in different bands, thus complicating the application of consistent procedures, even when using data with comparable angular resolution.

A particular concern with fibre spectroscopic surveys such as the SDSS is that the fibre apertures (3” in SDSS) only sample part of the object, with this fraction different at each redshift. The effect of this can be tested by comparing result of a fit to the photometry of the whole galaxy against the photometric fit corresponding to the area of the fibre only (e.g  Gómez et al. 2003Brinchmann et al. 2004).