Friday, 12 March 2010

The mid-IR luminosities of normal galaxies

In the last few years there has been a lot of discussion about the effect of TP-AGB stars on stellar mass estimates for high redshift galaxies. Such stars produce a large fraction of the rest-frame NIR light from youngish (~1 Gyr old) stellar populations, so if observations at these wavelengths are used when fitting stellar population models, it is important to take these stars into account. However, given all of the modeling and observational uncertainties associated with our understanding of TP-AGB stars, we are a long way away from being able to robustly incorporate them into stellar population synthesis (SPS) models.

This letter by Dan Keslon and Brad Holden looks at another important effect of TP-AGB stars: their contribution to the MIR luminosity, which is frequently used as an indicator of the star formation rate. The authors estimate contribution to the MIR luminosity by adding the observed K-MIR colors for Galactic TP-AGB M and C stars to the expected contribution to the K-band luminosity that comes from Maraston's SPS model. This figure shows some of the results. The upper blue lines show the maximum contribution to the MIR, which occurs when the TP-AGB C stars dominate (although this comparison must depend somewhat on the star formation history assumed in the models). It thus appears that the MIR luminosity of galaxies is consistent with being entirely due to the TP-AGB stars.

There's a lot of stuff in this (very long) letter, but what is most striking is their figure 2d, which I haven't shown here. That figure shows that the authors are able to reproduce the observed correlation between 24um luminosity and star formation rate from Chary & Elbaz (2001) very well, but only when the SFR is averaged over the last 1.5 Gyr (which is when the TP-AGB stars are important). But I do find this result a little weird, since presumably that correlation was made against SFR indicators that are sensitive only to more recent SF. Also, it would suggest that very little of the MIR luminosity comes from dust that is not immediately surrounding TP-AGB stars.


Daniel said...

Ryan, perhaps I can help clarify the issue with the Chary & Elbaz (2001) correlation. It is that for a constant star-forming system, one need not look back 1.5 Gyr to derive an average rate of SF that agrees with CE2001. But if a system has a non-constant SFH, then, essentially, the on-going instantaneous rate of SF is simply out of sync with the number of intermediate mass TPAGB stars present because those are produced over such a long timescale.

Now, you raise a subtle question: how CE2001 was derived in the first place. If you recall, CE2001 is all about the correlation between L_12um (and L_15um, etc) with L_TIR, and then L_TIR vs SFR is simply from Kennicutt (1998). Setting aside any implications for where L_TIR arises (!), this calibration is derived for galaxies selected to be star forming. Odds are that something forming stars today probably was forming stars 1 Gyr ago. What the correlation is between the ongoing rate of today and the rates in the previous 1 Gyr is not obvious but one can evoke the central limit theorem to suppose that things probably average out fairly well. But if one were to select a bunch of galaxies, say, by their stellar continuum and make plots of L_12 vs H_alpha star-formation rate, it might look a little different than CE2001 (see, e.g., Calzetti et al, Kennicutt et al, etc) but the lore is that the difference is all "obscured star formation".

Or is it?

Anonymous said...

不只BLOG內容很棒留言也很精采 XDDDD.........................