While I agree on your additional criteria, I feel the roughness metric itself (at least as explained here) is not as informative as it could be: a generally smooth road surface with sudden discontinuities in level (e.g.potholes) seems qualitatively worse (and damaging) than would be a smoothly-varying one with the same roughness. Perhaps an alternative metric might be based on the maximum speed at which a typical car or truck could travel without experiencing vertical accelerations above a certain threshold? ('typical', here would be with regard to things like its mass, suspension travel and stiffness, and wheelbase.)