4 The infrared SN rate

Our monitoring program is still in progress. A final, detailed summary of our survey and its implications on the SN rate will be provided in a forthcoming paper (Mannucci et al. in prep.). Here we compare the number of SNe detected during our survey with the number of SNe expected by applying the current estimates of SN rates, expressed in SNu (Cappellaro et al. 1999), to our sample of galaxies. So far, our program has detected 4 SNe. One of these is a type Ia (SN 1999gd, Li 1999), which was also detected in the optical. Another SN was a type II (SN 2000bg, Sato & Li 2000), also detected in the optical. The other two SNe, SN 1999gw and SN 2001db, were discussed in this paper.

By assuming the conversion from blue luminosity to SN rate (all types) given in Cappellaro et al. (1999), i.e. ${\rm SNr} \simeq 10^{-12} (L_B/L_{\odot})~\rm yr^{-1}$, we would have expected to detect $\sim$0.5 SNe. Since we have detected 4 SNe, we roughly infer a SN production which is about an order of magnitude higher than estimated by conventional optical surveys.

This high SN rate reflects both the higher extinction affecting the B light of LIRGs (which is the normalizing factor of SNu) and the enhanced star formation in the galaxies of our sample. The enhanced SN rate obtained by our IR survey partly reconciles the contractictory results on starbursts obtained by optical surveys, that were discussed in the Introduction.

Note that a SN rate higher than expected by the blue luminosity is also obtained by the two SNe detected in the optical, implying that the high SN rate inferred by our study is not only due to obscuration. An important difference of our study, with respect to previous surveys, is that our sample is characterized by a much stronger starbursting activity; more specifically, if the star formation is estimated through the far-IR luminosity, the galaxies in our sample are characterized by a star formation rate which is about one order of magnitude higher than in the sample of Richmond et al. (1998). Probably, the failure of previous studies to detect an enhanced SN rate in starburst galaxies is partly to ascribe also to the selected sample, which consisted mostly of "mild'' starbursts.

Altough higher than found in optical surveys, the SN rate measured by us is still not as high as expected by the large star formation rate inferred from the far-IR luminosity of the galaxies in our sample. Indeed, if most of the far-IR luminosity is due to star formation, and we adopt the conversion from $L_{\rm FIR}$ to SN rate given in Mattila & Meikle (2000), i.e. ${\rm SNr} \simeq 2.7 \times 10^{-12} (L_{\rm FIR}/L_{\odot})~\rm yr^{-1}$, we find that our survey has detected only about 20% of the expected SNe.

There are various possible scenarios which could explain the shortage of SN detections. One possibility is that most SNe are so embedded in the dust that they are significantly obscured even in the near-IR. More specifically, in order to reduce the expected number of SNe by a factor of $\sim$5 (i.e. the fraction of missing SNe) they should be, on average, fainter by 3.0 mag in K, implying that they should be absorbed by A_K > 3.0, or A_V > 30. Another possibility is that most SNe occur in the nucleus (i.e. within the central 2''); in this case our limited angular resolution would have prevented disentangleing them from the peaked nuclear surface brightness of these galaxies, even when the PSF between multi-epoch images are optimally matched. In support of this scenario, recent mid-IR studies have found growing evidence that a large fraction of the starburst activity occurs in the nuclear region (Soifer et al. 2001). The possibility that most SNe occur in the nuclear region can be tested through near-IR monitoring from space (e.g. NICMOS on HST), which would provide higher angular resolution and more stable PSF. Alternatively, nuclear SNe could be disentangled spectroscopically through the detection of the broad hydrogen lines characteristics of type II SNe, by periodically monitoring the (near-IR) spectra of the nuclei of the galaxies in our sample. The discovery of a nuclear SN by Aretxaga et al. (1999) through the detection of broad hydrogen lines has shown the feasibility of the latter approach in the optical.

Finally, the assumption that the high far-IR luminosity is tracing an enhanced star formation rate might be incorrect. In particular, obscured AGNs may contribute substantially to the IR luminosity of these galaxies (Sanders & Mirabel 1996). Note that heavily obscured AGNs may be present even if not identified spectroscopically and may be totally obscured even in the X-rays (Marconi et al. 2000; Maiolino et al. 1998).