The surface stays the same. The operational definition changes.
A 2023 USGS-listed journal study examined how the measured roughness of natural fault surfaces changes with scale. Using a standard height-to-profile-length ratio, the surfaces appear much rougher at millimeter scales than at scales of tens of meters.
When the author instead averaged height over wavelength, the strong trend nearly disappeared—and the small remaining trend pointed the other way, toward slightly greater roughness at long wavelengths.
This is not proof that measurement is arbitrary. It is evidence that the word “roughness” must name its metric before a comparison becomes meaningful.
How both results can be true
A surface contains features of many sizes. One method asks how height varies across a profile of a given length. The other organizes the same geometry by wavelength. Those operations emphasize and normalize scale differently, so they do not have to preserve the same apparent trend.
What the paper concludes
The USGS abstract says the scale dependence is weak under the wavelength-based treatment and much smaller than the scale dependence of several other fault and shear-zone properties. It also says some natural surfaces may be consistent with Brownian behavior within measurement uncertainty.
Why this belongs on LossyInfo
A result can lose its measurement definition while the number survives. “Faults are rougher at small scales” and “faults are slightly rougher at long wavelengths” sound contradictory until the metric is restored. The missing context is mathematical, not rhetorical.
Primary research record
U.S. Geological Survey: On the scale-dependence of fault surface roughness
Journal article published in JGR Solid Earth in 2023. LossyInfo review completed 20 August 2026. This card summarizes the USGS abstract rather than independently reproducing its analysis.