A large body of evidence has demonstrated that the global rest-frame optical and IR colours of galaxies correlate well with each other, as well as with other galactic properties such as surface brightness and morphology. However the processes that lead to the observed correlations are contrary; the stellar light that contributes to the optical is readily absorbed by dust which emits in the IR. Thus on small scales we expect these correlations to break down. We examine seven nearby galaxies ranging from early- to late-types, on a pixel-by-pixel basis and we demonstrate that there is disconnect between the optical and IR when normalized to the near-IR (H-band). We can decompose this disconnect into two distinct components through a Principal Component Analysis of the H-band normalized SED of the pixels: one mainly correlated with variations in the IR, the other correlated with variations in the optical. By mapping these two components it is clear they arise from distinct spatial regions. The IR dominated component is strongly associated with the specific star-formation rate, while the optical-dominated component is broadly associated with the stellar mass density. However, when the pixels of all galaxies are compared, the well known optical-IR colour correlations return, demonstrating that the variance observed within galaxies is around a mean which follows the well-known trend. We also examine the extremely strong correlations between the IRAC-NIR colours and demonstrate that they are tight enough to use a single IRAC-NIR colour (i.e. 8mum-H) to determine the fluxes in the other IRAC bands. These correlations arise from the differing contribution of stellar light and dust to the IRAC bands, enabling us to determine pure stellar colours for these bands, but still demonstrating the need for dust (or stellar) corrections in these bands when being used as stellar (dust) tracers.
Download