[hansvm]

That's only a problem if your favorite method of representing uncertainty is in-band signalling via a technique like significant figures.

[justonceokay]

You could translate the units and then also translate separately the error bands (assuming +/- 100 km). But “2237 mi/hr +- 62 mph” sounds pretty silly.

[floatrock]

Use the error bands to find a close-enough round number: 2200 or 2300 mph or "almost 3 times the speed of sound on earth"

Anyone pedantic enough to notice the conversion mismatch should be smart enough to understand the concept of "error bars".