Roughness is present for the majority of surfaces because atomically flat interfaces are not common. Even though the surface "roughness layer" may be very thin, on the order of countable atomic layers, it frequently has a very significant effect on light reflection, chemical reactivity, adhesion, and the like. Consequently the analysis of surface roughness is of interest. Surfaces of metal oxide films deposited by chemical vapor deposition were scanned using an atomic force microscope (AFM). Initial processing of the very large AFM data sets, typically over one megabyte in size in ASCII format, included tilt correction, "scattering noise" removal, surface smoothing, and surface compression. Techniques for enhanced visualization of the surface using hard limited and band images are presented. Histograms (approximate probability density functions) are computed for surface possessing long range order and it is observed that both the long and short range orders are essentially Gaussian. Consequently the void fraction dependency is very close to an error function for these cases. A case (much less common in our data) in which the surface pdf was Gamma rather than Gaussian is also shown for comparison.