Let's try a 'fer instance' using a relatively modern, more or less common example of what we're looking at here -
- let's take the NWS's WSR-88D which employs a 28' dish, an S-band freq of 2.4 GHz (thereabouts) which results in 45 dB of gain and a 3 dB beamwidth of just under a (1) degree ...
... do you know what kind of distance an arc of 1 degree covers at 35,786 kilometers (19,323 nautical miles or 22,241 statute miles) above the earth's surface?
The problem that arises is that - MULTIPLE 'targets' would be indicated with a beamwidth at that distance - and I've not consdered the path-loss that would be encountered *both* directions - with a poor reflection/reflector (the bogey aircraft) as the source for the return (signal) ...
A better solution might be increased use of land-based OTH (Over The Horizon) HF (High Frequency) RADAR rather than a) balloons or b) an impossible satellite solution ...