This is a short summary of the 2003 paper by Belokurov and Evans.

Also see the supernovae information sheet (pdf) from the Gaia web-site.

Numbers detected

This figure shows the fraction of SNe within a distance $D$ which enter the field of view of Gaia's telescopes ASTRO-1 and ASTRO-2. Gaia records data on $30 \%$ of all the SNe Ia within 630 Mpc, which marks the limit of the most distant SNe Ia accessible. The distance cut-offs for the intrinsically less bright SNe Ib/c, II-L and II-P are 355 Mpc, 335 Mpc and 195 Mpc respectively. Gaia records data on $\sim 20 \%$ of all SNe Ib/c, $\sim 31 \%$ of all SNe II-L and $\sim 48 \%$ SNe II-P within these distances. This means that Gaia will provide some (perhaps rather limited) information on $14\,300$ SNe Ia and $1400$ SNe Ib/c during its five-year mission. For SNe II, the number depends on the relative frequency $\f$ and is $\sim 8700 \f + 2700 (1-\f)$. If SNe II-L and SNe II-P occur equally frequently ($\f = 0.5$), then the total number of SNe II is $\sim 5700$. In other words, Gaia will provide some information on $\sim 21\,400$ SNe in total. For comparison, H{\o}g et al. (1999) used simple scaling arguments to estimate that the total number of SNe in the Gaia observations would be $\sim 100\,000$.

These are huge numbers, both compared to the sizes of existing catalogues and to the likely datasets gathered by other planned space missions. Almost all the SNe that Gaia misses explode in the 20 days just after Gaia samples that location in the sky. Before the next transit of ASTRO-1 or ASTRO-2, the SN reaches maximum and then fades to below Gaia's limiting magnitude ($G \sim 20$). It may be wondered whether some SNe are missed because light from the background galaxy can overwhelm the SN. This is clearly a problem for distant galaxies, which are wholly contained within Gaia's PSF ($\sim .35''$ at FWHM). However, rough calculations show that this is not a problem for SNe Ib/c and II, as they occur relatively close by; we estimate that it may affect $\lta 10 \%$ of SNe Ia.

This figure shows the fraction of the detected SNe as a function of phase of the lightcurve. Some $44 \%$ of the detected SNe Ia are caught before maximum, $37 \%$ of the detected SNe Ib/c, $37 \%$ of the detected SNe II-L and $9 \%$ of the detected SNe II-P. The low fraction for SNe II-P is largely a consequence of the fact that they are intrinsically the faintest. The total number of all SNe found before maximum during the 5 year mission lifetime is $\sim 8500$. This number can be broken down into $\sim 6300$ SNe Ia, $\sim 500$ SNe Ib/c and 1700 SNe II (assuming $\f = 0.5$). If data on a SN is taken before maximum, then Gaia has an excellent chance of identifying the rapidly brightening object as a SN.

Identification strategy

Every new object in a field of view is potentially a SN. Before Gaia can identify a SN, it must have visited that location on the sky at least once before. To provide SN alerts, we must ensure that the brightening object is not just a common variable star. We use the General Catalogue of Variable Stars (Kholopov et al. 1999) to build a subsample of variables with periods in excess of 10 days. Some $34 \%$ of this subsample have periods less than 6 months and $86 \%$ have periods less than 1 year. In practice, $\sim 12$ months baseline photometry may be needed to discriminate against common forms of stellar variability. One way round this problems is to restrict SN alerts to high galactic latitudes ($|b| > 30^\circ$), where the problem of variable star contamination is mitigated.

The objects that can cause most confusion are fast-moving solar system asteroids and novae. Main Belt asteroids move at $\sim 10$ mas s$^{-1}$ and Near-Earth objects at $\sim 40$ mas s$^{-1}$ (e.g., Mignard 2002). H{\o}g (2002) has shown that fast moving objects can be detected by a single field of view crossing. This offers quick discrimination between solar system asteroids and SNe. More problematical are novae. Shafter (1997) gives the Galactic nova rate as $35 \pm 11$ yr${}^{-1}$. We assume that this is typical of large galaxies. We take the absolute magnitude of novae to be in the range $-6 < M < -9$ (Sterken \& Jaschek 1996). Given Gaia's limiting magnitude, there are between 20 and 150 galaxies in the CfA catalogue for which novae are detectable. This means that there are $\sim 1000$ novae per year in the Gaia datastream. These can possibly be distinguished from SNe on the basis of colour information and spectroscopy. However, contamination by novae from external galaxies -- which is the bulk of the numbers -- is restricted to a number of small and pre-determined areas of the sky. These can, if necessary, be excised from the SNe survey.

Therefore, a reasonable expectation is that Gaia will alert on all SNe caught before maximum, that is $\sim 1700$ SNe a year. Riess et al. (1998) show how the distance to a SNe Ia can be estimated to within $10 \%$ from a single spectrum and photometric epoch. For a 20th magnitude SN, the most suitable combination is a 2m telescope for imaging and a 4m for spectroscopy. On a 2m telescope, it is feasible to carry out high signal-to-noise UBVRI photometry on a 20th magnitude SN in $\lta 1$ hr. The typical signal-to-noise needed for identifying the type from spectroscopy is $\sim 30$. This needs $\sim 2$ hrs on a $V \sim 20$ point source in dark sky at low spectral resolution. However, roughly half the SNe will not be suitable for follow-up from the ground as they will be daytime objects. Assuming we wish to follow up the sample of (nighttime) SNe alerted before maximum, then roughly two dedicated telescopes (say one 2m and one 4m) are required to get distance and phase estimates. This will confirm detection and type. Based on the information from the one night snapshot, selected SNe can be chosen for more detailed monitoring. Candidates for intensive monitoring might include all the SNe Ib/c and II-L (as there is little information on their lightcurves), subluminous and over-luminous events, all SNe Ia caught well before maximum and any SN for which the snapshot gives an unusual luminosity or spectral composition. Tammann \& Reindl (2002) have also recently emphasised the value for the extragalactic distance scale of such a follow-up program of Gaia SNe Ia alerts.

Bibliography