Some radio supernovae have been monitored for several decades after the explosion. They become SNRs some 100-1000 years later. We do not know what happens during the transition from RSN to SNR, i.e., in the transition from interacting with the circumstellar material left over from the progenitor to interacting with the general interstellar medium.
The radio behavior during the SN phase diagnoses the history of
mass loss from the progenitor, since the emission comes from
the shock between the ejecta and the CSM. A sharp intensity
decrease indicates the outer edge of the CSM, giving a
timescale for the progenitor's slow wind, and yielding an
estimate for the total mass loss for that star. This can be
compared with stellar evolution and ISM enrichment theories, a
comparison that is particularly interesting if the SN was
observed optically. This has been done so far for only four
SNe.
The difficulty lies in tracking supernovae to very low
flux levels; SN 1961V in NGC 1058 (
= 518 km/s) was
recovered at about 0.2 mJy at 
20cm
To reach a similar level in the Virgo cluster (about a
factor of two further) we need 10 microJy rms noise,
achievable in about 1 hr with the enhanced sensitivity; even
with several pointings, the dozens of
optically-known SNe in that cluster could be detected easily.
SNRs indicate recent massive star formation (stars that blew
up 
years ago) and are the main source of the high
energy particles (cosmic rays) that form the diffuse
synchrotron emission in galactic disks and halos. They are
important both in terms of energy input, disturbing the ISM,
creating H I ``bubbles'' and possibly triggering further star
formation, and as sources of metals in galactic disks. The
current state of SNR detection is illustrated by recent
observations of M33, where some 50 SNR were detected at an rms
sensitivity of 50 microJy at 1.4 GHz.
The cumulative distribution of SNRs as a
function of flux density suggests either that
incompleteness sets in at 0.5
mJy, or that the source counts may turn over there.
Observations with a factor 10 higher sensitivity could be
achieved in less than 2 hrs with the proposed 2.4 GHz system,
and would settle this question beyond doubt. If the steeper
power law is correct, this would yield a factor 20 more
sources, or some 1000 SNRs. Confusion will be significant but
one should be able to determine the supernova rate
statistically with such large numbers. As well as obtaining a
crucial parameter for models of energy input into the ISM,
galactic abundances, etc. much more accurately, this would
also give the ``turnover time" for replenishing the cosmic
rays in the galactic disk, which with more SNRs would be
embarrassingly short. As well as such detailed studies, one
could extend such observations
to 
Mpc with
2-hr integrations, yielding more galaxies, and examining the
dependence of SN rates and the available pool of particle
accelerators on galaxy type and activity.
Resolving SNRs yields much more information than simple
detections-it lets us distinguish SNRs from background sources
by their morphology and also adds important physical information
about the SNRs. For instance, the size gives an indication of
the age of a remnant, and the relationship between size and
surface brightness (the 
-
relation) should also reflect
the evolution of the source. The results of such studies in the
Milky Way are muddled, at least in part, because of
inhomogeneous samples and uncertainties in SNR
distances. Studying SNRs in another galaxy provides better
statistics (more SNRs, some indication of dependence on galactic
environment, much better understanding of selection effects)
and a common distance for all SNRs seen. Resolved images
also give SNR types (filled, shell, or mixed), giving some
indication of SNR histories and current power sources (pulsar
vs. shock). Ideally one would like sub-parsec resolution
at 0.6-2.4 GHz. The current A configuration would be useful for the
larger sources, but the A+ configuration would be much
better, giving a 11cm resolution of 
pc at the
distance of M33 (800 kpc).
These observations require high sensitivity over the entire
primary beam at a low frequency. We need a high-sensitivity
system at some frequency between 0.6 and 2.4 GHz; which
frequency is not so important, nor does one need continuous
coverage. Low frequencies are best because these sources have
fairly steep spectra (
to 
). One should
also be able to image the entire primary beam at the highest
available resolutions, to cover an entire galaxy or more with
one pointing.