Astron. Astrophys. 351, 869-882 (1999)

6. Discussion

NIR growth curves of the magnitude as a function of the aperture have been built from the Catalog of Infrared Observations (Gezari et al. 1997) and optical catalogs. Using these, we have been able to compute total NIR apparent and absolute magnitudes, NIR and optical-to-NIR colors, and to estimate their uncertainties for a large sample. A statistical analysis of the colors as a function of type, inclination and luminosity, using estimators taking into account the uncertainties in the variables and the intrinsic scatter in the colors, highlights the interest of the NIR and notably of the comparison of the optical to the NIR.

Optical-to-NIR colors show a well defined sequence with type: the mean irregular is 1.35 magnitude bluer in than the mean elliptical. The intrinsic scatter is higher for star-forming galaxies () than for ellipticals or lenticulars (-0.2).

Because of the small extinction in the NIR, the optical-to-NIR colors of spiral galaxies redden considerably with increasing inclination, putting thereby constraints on the amount of dust and the respective distributions of dust and stars. The colors of S0 do not depend on the inclination. Rounder ellipticals are redder than more elongated ones.

The color-absolute magnitude relation is much steeper and tighter when the NIR is used as the reference band rather than the optical. Examination of the color vs. B-absolute magnitude relation for each type suggests that this is due to the extinction which dims and reddens the galaxies and counteracts the color- relation. A color-magnitude relation exists in each type, with brighter galaxies in the NIR being redder. The slope is steeper for Sa-Sc than for early-type galaxies or Sd-Im. The relation we obtain for ellipticals is shallower than in other studies. A likely explanation is that this is due to the effect of the color gradient in small apertures, a lesser, i.e. redder, fraction of the galaxy being observed in large bright ellipticals than in small faint ones - a problem we do not have using our total (or we expect so) colors. The slope for S0 galaxies is closer to that of spirals than to that of ellipticals. The color at a given absolute magnitude becomes bluer with increasing type, indicating that both the mass and the type must be used to characterize the colors and to describe the star formation history of galaxies. Once corrected for the inclination and the magnitude effect, the intrinsic scatter in the colors of spiral types drops significantly, although a few outliers with very blue colors typical of starbursting galaxies remain.

However satisfying these results are, some problems still exist. NIR observations are obtained in small apertures and the extrapolation to total magnitudes and colors suffers from significant uncertainties. Nevertheless, we do not believe that our results strongly depend on this extrapolation since similar slopes are obtained when we use effective colors - which are interpolated or only slightly extrapolated - instead of total colors.

The selection of galaxies is a more crucial problem. The only criterion used here has been to take whatever was available in the NIR. For this reason, the sample, especially in J and K, contains few late-type galaxies but many cluster galaxies, Seyferts and other active nuclei. We also lack of intrinsically faint galaxies. The color-magnitude relations we have determined should therefore not be extrapolated at fainter than . Large samples as the Sloan Digital Sky Survey in the optical or DENIS and 2MASS in the NIR should solve this problem.

The last important problem is the bias between the two estimators we use, especially in the slope of the color-magnitude relation. The discrepancy tends to disappear however when we consider effective colors, which have smaller uncertainties. The MCES estimator seems to be closer to the true slope than the ML, but the latter provides a better estimate of the intrinsic scatter and has a lower variance.

Combining the NIR and the optical enlarges considerably our vision of the stellar populations and the dust content of galaxies. In our next paper, we will extend our analysis to the near ultraviolet-optical wavelength range and will publish our catalog. We will also provide "color" energy distributions of galaxies from the near-UV to the NIR that could be used as templates to constrain the star formation history of galaxies of different types and masses. A further step will be the extension of these techniques to the far-UV and the far-IR.

© European Southern Observatory (ESO) 1999

Online publication: November 16, 1999
helpdesk.link@springer.de