A brief summary of the methods commonly used for the analysis of rough surfaces and the errors associated with these techniques is given. A frequency domain digital differentiator is shown to reduce the bias errors for estimates of the properties of slopes and curvatures for a sample spacing determined by the profilometer probe size. A comparison of this “spectral” approach with finite difference methods for calculating slope and curvature characteristics for a ground surface show substantial underestimates for the latter methods.

This content is only available via PDF.
You do not currently have access to this content.