Despite approximately 150 years of research on drumlins, no consensus on how they form has emerged: their origin is probably the longest-standing problem in glacial geology. Solving this problem would answer important questions about the basal conditions of past and modern sediment-based ice masses and their dynamics. Although many models of drumlin formation have been proposed, in the last several decades those that involve deformation of a till bed have received the most attention. In these models subglacial growth of hills is in response either to effective stress heterogeneity in the bed or to flow instability at the ice-till interface.

The forefield of Múlajökull—a warm-based, surge-type glacier in central Iceland—comprises the only known field of drumlins shaped by a modern glacier and provides an opportunity to study the stratigraphy, patterns of sediment deformation, and past effective stress distributions in and around these drumlins. Their geometric characteristics fall within ranges for Pleistocene drumlins (Johnson et al., 2010; Benediktsson et al., submitted), but glaciological conditions during drumlin formation are better known than for the Pleistocene drumlin fields that have been the focus of most previous work.

This study combines field measurements, laboratory analyses of fabrics based on till anisotropy of magnetic susceptibility (AMS), consolidation testing, and analysis of aerial photos and LiDAR data from the Múlajökull forefield. Data indicate that the drumlins formed by both erosion and deposition. (1) Attitudes of basal till units that constitute drumlins and patterns of till deformation within them indicate that deposition occurred on drumlin slopes. (2) Clast and AMS fabrics in drumlins and interdrumlin areas are dominated by shear strain, with no evidence of the longitudinal compression or extension necessary if sediment flux divergence in a deforming bed caused drumlin relief. (3) The package of basal tills that dominates the stratigraphy of the drumlins thickens with proximity to the glacier, where the forefield has been subject to more frequent surging. (4) Drumlin elongation is greater in parts of the forefield where more surging has occurred. (5) Observations of an unconformable uppermost till unit that often truncates underlying tills at the drumlin flanks indicate that erosion occurred there. (6) Data reflecting quiescent periods, when effective stresses are inferred to have been higher than during surges, indicate that past effective stresses were highest in interdrumlin areas at those times. These data, along with the stratigraphic evidence for erosion on the drumlin flanks, suggest that erosion occurred during quiescence rather than during surging and imply that sediment transport mechanisms with rates that increase under increasing effective stress likely were responsible for this erosion.

Thus, the stratigraphic, morphological, topographic, geotechnical, and fabric data from the Múlajökull forefield support a model of drumlin formation in which till is eroded from interdrumlin areas and adjacent drumlin flanks during quiescent periods. On the other hand, stratigraphic evidence connecting specific till layers to some surge moraines (Johnson et al., 2010) indicates that deposition occurs during surges. Such deposition could result from negative flux divergence in a shearing bed, but till fabrics, as noted, do not reflect this style of strain. Rapid but relatively uniform bed shear during surging—consistent with the low and relatively uniform effective stresses expected during such periods—that generates basal melt rates sufficient to release debris from ice and lodge it onto the bed is a more likely origin for the basal till. Thus, till stratigraphy and fabrics reflect lodgment and associated bed deformation during surging. Although this conceptual model is founded on observations of the Múlajökull drumlin field, it implies that drumlins elsewhere formed by slow flowing ice will be dominantly erosional, whereas drumlins that show evidence of deposition may have experienced episodes of fast ice flow.