Ice surface roughness is a parameter of importance to many glaciological studies. Surface roughness, a derivative of microtopography, provides the high-resolution information necessary to characterize morphological types of ice surfaces and complements satellite data in investigations of surface features across scale. There has been a gap in observational scale between high-resolution satellite data, aircraft data and analyses at microscopic scale, the latter investigating material properties rather than micromorphology. To fill this gap, we designed and built the Glacier Roughness Sensor (GRS), a towed instrument that collects surface roughness data mechanically in swath-survey style with 0.1 m along-track spacing, 0.2 m across-track spacing, and sub-centimeter accuracy for areas typically 100 m by 175 m. As a result of this instrument development, the variable ice surface roughness has been added as a geophysical observable in the study of glaciers and ice sheets. The method utilized for analysis of GRS surface roughness data is geostatistical classification employing a range of parameters extracted from vario functions, which are generalized spatial structure functions. In a seasonal comparison of data from spring (May 1997) and summer (late July 1999), characteristic parameters of the spring and summer ice surface were calculated, and, as a result, an answer to the morphogenetic question of ice surface processes could be derived: in this part of the Jakobshavn Isbræ drainage basin, ice-surface structures develop in an interplay of ablation, refreezing, snowfall, wind and (distant) crevassing, each process yields characteristically different components of the surface structure, and ablat ion is the dominant morphogenetic process.