This, of course, was one of the topics at the recent GeoMars conference. None of the immediately planned radar missions is a SAR, they are merely depth sounders, operating at different frequencies to give different attempts at penetration vs resolution.

Personally, I am also in favour of a radar-specific mission (no other distractions allowed onboard), with SAR and multifrequency sounding capability. Out of this will come a detailed surface elevation map to fill all the gaps that MOLA left, as well as a map of surface/near-surface radar reflectivity at several wavebands.

This will tell us a lot about the near-surface of Mars and in an extreme case might just possibly reveal the water that is supposed to be pouring madly out of the frozen surface of modern Mars in gullies, dark streaks etc.

It will give us hints about mineralogical distributions and may be a useful angle at working out what on Mars those dark dusts really are.

In sounding mode, or low-frequency SAR it will give reasonable penetration into the polecaps and help us understand what is going on with the only water we have, in fact, discovered on Mars

(Not liquid water, but the odd million cubic km of H2O-rich ice).

Of course, to really take advantage of such a mission, we also need a Mars relay system. One one level the radar satellite can itself do this, since its radar antenna would also serva as a handy 40m sub-DSN transmitter, allowing us to pick up broadband signals on Earth from 40m dishes or smaller. OTOH, any time spent transmitting with the radar aerial is neccesarily time wasted from mapping Mars (probably with everything, since the whole spacecraft will probably have to point at Earth - consuming more propellant over time and shortening the total mission endurance.

I would rather see an independant Mars communication network in place, with capacity to receive broadband but omnidirectional signals from other orbiters, landers, rovers etc. Apart from anything else, it's the first step in a proper long-term Mars-orbiting positioning system that does away with the stupidity of measuring how close one is to Mars from the vantage point of Earth, and also permits proper rover operations on the surface of Mars which are currently hampered by navigation control, as much as anything else.

But here I am spilling the beans about GeoMars without having sent the details to Alex first, so that's enough for now, while I go off and complete my GeoMars report...