What is different btw fixedValue & supersonicFreestream?
 

I am running several compressible flow simulation, and I tried two different velocity options for input surface, fixedValue and supersonicFreestream

Both two options converged well, but problem is that their data show so much difference. Drag force for fixedValue was about 30% higher than that of supersonicFreestream.

Due to their name, I guess supersonicFreestream may be more accurate, but as these two option show so drastic difference, it will cause huge singularity at transonic region because supersonicFreestream is unavailable at subsonic domain, and I would have to use different bc for M0.99 ad M1.01...

What are difference of these two boundary options? Which data should I trust more?