In line with the rest of this chapter, I will summarise the lessons learned on SFR indicators by separating the global, whole-galaxies case from the one describing local, sub-galactic regions.
1.5.1. Global SFR indicators
As the integrated light from galaxies is a weighted average of the most luminous contributors (i.e., star-forming regions), it is perhaps not surprising that global SFR indicators show a high level of consistency, as summarised in Kennicutt & Evans (2012). As long as the assumptions for the stellar IMF are factored in the calibration constant, stochastic sampling of the IMF is not an issue, and both the dust-obscured and dust-unobscured star formation are accounted for, different SFR indicators should yield similar answers.
All other conditions being equal, SFR indicators that are sensitive only to short timescales, i.e., only probe the presence of short-lived, massive stars, should be preferred to long-timescale ones. Examples of short-timescale SFR indicators are those using ionised gas tracers (e.g., H). Conversely, the IR probes emission from stellar populations covering a large range of ages, and its use will depend on the dominant stellar population contributing to the IR emission and on the required accuracy for the SFR measure: as shown in Table 1.1, the calibration constant changes by a factor of 1.75 if the timescale of the star formation changes from 100 Myr (e.g., a LIRG or more luminous galaxy) to 10 Gyr (a normal star-forming galaxy).
At high attenuation values, which generally correspond to high SFR values, about a few times 10 M yr-1 or above, the dust starts competing with the gas for Lyman continuum photon absorption. Combining an ionised gas tracer (e.g., H) with a dust emission tracer (e.g., 24 µm) is likely to not only mitigate this problem, but also provide a general answer to the question of how to correct UV/optical tracers for the effects of dust attenuation. Mixed SFR indicators (H + 24 µm, UV+24 µm, etc.) have, indeed, broad applicability in all cases where stochastic IMF sampling is not a concern.
At low SFR values, below ~ 10-3 M yr-1, stochastic sampling of the IMF affects the use of ionised gas tracers as SFR indicators. The longer-lived UV emission may thus become a preferable choice, as long as it is corrected for the effects of dust attenuation. Even the UV, however, is not a `panacea', since stochastic IMF sampling starts affecting the UV emission barely a factor three to four below the SFR level of the ionised gas.
Exclusive use of the UV (even after dust attenuation corrections) may complicate the discrimination between star-forming galaxies and post-star-forming galaxies (e.g., Weisz et al. 2012), i.e., galaxies whose active star formation terminated many tens of Myr ago and for which the use of any of the calibration constants in Table 1.1 will only yield a lower limit.
1.5.2. Local SFR indicators
Unlike whole galaxies, regions within galaxies are not isolated systems, and a variety of issues needs to be considered when attempting to convert any luminosity into a SFR. Evolution and mixing of stellar populations and the ability of stellar continuum light and ionising photons to leak out of star-forming regions and travel to distances of 1 kpc, and possibly more, are important effects that need to be taken into account for deriving local SFRs. Most of the problems do not reside with the regions of star formation proper, but with the faint regions that may have little or no in-situ star formation. These cumulatively termed `non-in-situ star formation' regions can still emit at many wavelengths, including H, because of leakage from surrounding areas.
The different estimates on the importance of these effects that can be found in the literature (e.g., more than a factor of two difference between Liu et al. 2011 and Leroy et al. 2012) attest to the complexity of the issue. An example of the problems induced by inaccurate estimates of local SFRs is given in Calzetti et al. (2012). Here, it is shown that the relation between the SFR and cold gas surface densities, also known as, the Schmidt-Kennicutt law, strongly depends on the treatment of the `non-in-situ' star formation.
A safe approach, at least for now, is to measure SFRs only in regions where there is ample independent evidence that star formation is actually occurring, such as in spiral arms and in the central regions of many galaxies. The study of M 33 has shown that dust heating is mainly powered by recent star formation in these locales, down to SFR/area ~ 0.002 M yr-1 kpc-2 (Boquien et al. 2011).
Even when star-forming regions or areas have been identified, care should be taken with how the different SFR indicators are applied: the one to choose for a specific case may depend on the star formation timescale of interest. In general, the shorter the timescale, the lower the dependence of the SFR indicator on the evolution of the stellar population. Ionising photon tracers may `fit the bill', although leakage of ionising photons out of star-forming regions will need to be accounted for.
I am extremely grateful to the organisers of this Winter School, Johan Knapen and Jesus Falcón-Barroso, for inviting me and for the opportunity to deliver these lectures. I am also grateful to the Instituto de Astrofísica de Canarias and its Director, Prof. Francisco Sánchez, for the hospitality. Many of the results presented in this manuscript are the products of the SINGS (Spitzer Infrared Nearby Galaxies Survey), KINGFISH (Key Insights in Nearby Galaxies: a Far Infrared Survey with Herschel), and LVL (Local Volume Legacy) collaborations, to whose members I am profoundly indebted.
I also want to thank my long-time, long-distance collaborator Robert C. Kennicutt; scientific discussions with him are always enlightening. We often end up with friendly disagreements, and both of our healths have benefited from never residing closer than about 2000 km from each other. Finally, parts of this manuscript have been improved thanks to discussions with another long-time collaborator, John S. Gallagher.
The preparation of this manuscript was supported in part by the NASA-ADAP grant NNX10AD08G and in part by the NASA Herschel grant JPL-1369560 to the University of Massachusetts.