Radio studies of supernovae and supernova remnants provide information on stellar evolution, stellar mass loss, particle acceleration mechanisms, energy input to the interstellar medium, and the structure of the ISM. The high sensitivity proposed for high frequencies in the enhanced VLA will allow studies of the very earliest phases of the radio emission after the supernova explosion.
The radio emission from supernovae (SNe) is thought to arise at
the shock between the outgoing ejecta and the progenitor star's
circumstellar wind; the shock compresses magnetic fields and
accelerates particles to relativistic speeds, giving synchrotron
emission. This process is the same as in
other synchrotron sources, but here we
know where, when, and on what physical scale particle
acceleration begins and proceeds, and can watch it occur over a
large fraction of the life of the source. Further, as we can
now model optical/UV spectra and photometry to derive the
densities of both the ejecta and the circumstellar material
(CSM), as well as the shock velocity, we can learn more about SN
shocks than about any others outside the solar system.
Unfortunately radio observations are somewhat complicated by the
very CSM that provides the radio emission: this material is
ionized by the initial optical/UV blast, and free-free
absorption hides the intrinsic emission until days, weeks,
months, or years after the explosion. This loses many of the
interesting physical constraints on how fast the particles are
accelerated, particularly at the highest energies. This changes
dramatically if we can make sensitive observations at the high
frequencies where the source first becomes optically thin
(free-free absorption declines as 
). The enhanced
VLA will make a dramatic impact on this field by probing
directly the most interesting energy regime within days of the
explosion. (SN 1994I was detected at 99 GHz within four days of
optical discovery, and only logistical difficulties prevented it
from being seen even earlier).
Aside from probing the effects of the young shock, early high-frequency detection would allow reliable determinations of the progenitor's mass loss and mass loss rate based on how much material is needed to mask the source at long wavelengths. This approach relies on knowing the inherent unabsorbed emission strength-currently a model-dependent extrapolation, not a direct observation. The derived CSM densities and density profiles are important first because they tie down models of SN optical/UV emission, and second because they tell us about the mass-loss history of the progenitor, a fundamental issue in stellar evolution. Finally, radio data taken during the first days after the explosion are the most useful, because they may be tied to simultaneous optical and UV data to develop and constrain detailed models of these sources.
Critical instrumental requirements are: frequencies of 40-50 GHz to access the highest-energy electrons and to minimize the effects of free-free absorption, rapid scheduling response (within days of optical detection) and high sensitivity (sub-mJy, phase coherent). The current 15-22 GHz system is ill-adapted to this work because of its low sensitivity.
One gains enormously by measuring the frequency dependence of the emission over as wide a band as possible. The basic model parameters-CSM density, intrinsic spectral index and time dependence-are determined much more accurately by multi-frequency observations than by detailed observations at one wavelength. Also, the fundamental assumptions of most simple models-e.g., that emission/absorption processes are scale-free-can be tested more directly. The main instrumental need is appropriate sensitivity in as many different bands as possible. This argues for the 40-50 GHz systems, as well as new 0.6 and 2.4 GHz receivers. Frequency coverage need not be continuous, but we need sensitive (few 10's of microJy in 5 min) systems for at least one high (40-50 GHz) and one low (0.6-2.4 GHz) band.
The biggest gain for detailed studies of individual objects is at high frequencies: a 2 GHz wide 40 GHz system could follow a 10mJy-peak supernova like SN 1994I for 40 days past maximum light with 5 minute scans, and obtain some data near maximum on sources as faint as 0.1 mJy. This would define the high-energy particle spectrum for normal Ib/c supernovae to a distance of 85 Mpc, and for the spectacular type II's a factor of 3 further out. The latter are particularly interesting, as none has yet been seen at mm wavelengths, although at least one (SN 1986J) emits copiously in the near-infrared.
Detection experiments become much more interesting with the enhanced VLA because sensitivity is increased comparably over a wide range of frequencies. For example, starburst galaxies are predicted to have about one supernova per year, yet directed searches have found no new radio supernovae. The current limits cannot rule out a population of less radio-luminous SNe (either intrinsic or heavily obscured by intervening free-free absorption). Both possibilities could be checked by deeper surveys at higher frequencies, where such absorption is minimized. H II galaxies, with star-formation rates of order a solar mass per year, should also make RSNe, and searches for those would also be aided by the increased sensitivity. It is important to realize that the supernovae we can study now at radio wavelengths are exceptional, with extremely high mass loss rates and strong shocks-lowering the detection threshold will let us detect more ordinary objects, and so learn about more typical stellar evolution.
Much improved SN searches will be possible if there is at least
one low frequency (1-2 GHz) where one can reach a noise level
of a few microJy in 12 hrs over the entire primary beam. (Low
frequencies are important because radio SNe decay more slowly at
those frequencies.) For example, one might spend 12 hrs at
20cm monitoring a distant starburst galaxy every six
months or so in a wide configuration to minimize confusion. One should
see radio SNe down to 20 microJy (
), corresponding to
the detection of a type Ib/c to 220 Mpc (z = 0.08) or a strong
type II (1986J) to 660 Mpc (z = 0.21, 
). For a
starburst galaxy at 100 Mpc, one should see even the dullest
supernova, or a normal type Ib SN even fairly far down its light
curve (within 200 days of the explosion), giving excellent
limits on supernova rates in starbursts. At the same time one
gets a good pencil-beam survey, and if the starburst is close on
the sky to a relatively nearby galaxy, good limits on radio
emission from type Ia SNe. The new correlator could be used for
an H I survey at the same time. Another option would be to
spend 12 hrs looking at a cluster: at 100 Mpc, the FWHM of the
20cm primary beam is about 1 Mpc, so one could measure
the radio SN rate in all the galaxies in the cluster at once.