Nanoprobe Network

This question conerns interpretation of surface roughness data on very smooth surfaces. It is my understanding that in measuring surface roughness only height data (either contact or tapping mode) should be used. Is there a point however, when a surface is smooth enough that deflection data (contact mode) should give you the same value or is maybe even more accurate? The source of this question is that the data we collect on highly polished (CMP) sapphire wafers may give a surface roughness of 1nm in height mode and 2A in deflection mode and published data does not normally say what mode it was collected in. Is it a well known standard operating procedure to only use height data? If the roughness of the surface is physically lower than the average cantilever deflection shouldn't deflection data be just as relevant? Any opinions on this would be highly regarded. Thanks. (Mike Kamrath, Nalco Co., Naperville, IL)