Seabed roughness was measured over an area of the south shore of O'ahu, Hawaii, with a boat-mounted acoustic altimetry system. The study site is characterized by a highly inhomogeneous rough seabed. Roughness was resolved in the 40–200-cm range over an area extending from the wave breaking zone to a depth of 20 m. Theory suggests that roughness length scales and distribution both play an important role in wave energy dissipation mechanisms. In this paper, we investigate methods for quantifying bed roughness within this range of scales. Various methods for quantifying roughness are examined, including spectral and statistically derived variables. These methods are compared qualitatively with sidescan imagery and quantitatively via in situ diver-based roughness surveys. Results show that wavenumber spectra can effectively differentiate smooth from rough areas. Results also suggest that the bed roughness in the study site exhibit a high spatial variability within the examined range, with no dominant length scale but with a characteristic spectral slope. In addition, roughness also exhibits variation between the shallow and deep ends of the study site.