Drumlin Formation Under Ice Sheets

What a Drumlin Actually Is

A drumlin is a streamlined hill built from till, the chaotic, unsorted debris that glacial ice picks up, transports, and deposits as it moves across bedrock and older sediment. Unlike sand or gravel sorted by flowing water, till contains everything from fine clay to boulders jumbled together, a direct product of being ground beneath and dragged along by ice rather than carried in a stream. The defining feature of a drumlin is its shape: an asymmetric, teardrop or half-egg profile. One end is blunt, broad, and relatively steep, while the opposite end tapers gradually into a long, low, gentle slope. Typical drumlins range from a few hundred meters to over a kilometer in length, tens of meters in height, and are often several times longer than they are wide. Their surfaces are smooth and rounded, lacking the sharp, angular relief typical of erosional bedrock features. Critically, the blunt end always points in the direction the ice came from, called the up-ice or stoss end, while the tapered tail points in the direction the ice was heading, the down-ice or lee end. This asymmetry mirrors, at a much larger scale, the shape water or wind carves into loose sediment when flow encounters an obstacle: blunt resistance on the upstream face, a longer sheltered wake trailing downstream. Drumlins are composed primarily of till, though some contain cores of older, more resistant material, bedrock knobs, or pre-existing sediment mounds that the ice draped and molded around. This variability in internal composition is itself a clue in the formation debate, since a purely depositional process would be expected to build hills from till alone, while cored drumlins suggest erosion and shaping of pre-existing material also plays a role in at least some cases. Drumlins are found exclusively in regions that were once covered by continental ice sheets or large valley glaciers, including much of Canada, the northern United States, Scandinavia, Ireland, Scotland, and northern Europe. Their presence is unambiguous physical evidence that grounded, moving ice once occupied a landscape, distinguishing glaciated terrain from surrounding areas that escaped glacial cover.

Drumlin Swarms as Ice-Flow Indicators

The scientific power of drumlins comes less from any single hill and more from how they behave collectively. Drumlins virtually never occur in isolation. Instead they cluster into swarms, sometimes called drumlin fields, that can contain many thousands of individual hills spread across tens of thousands of square kilometers. Within a given swarm, the long axes of the drumlins are remarkably consistent, often varying by only a few degrees from one hill to the next across enormous distances. This consistency is the basis of drumlins' most well-established and least controversial scientific use: as a direct proxy for the direction that an ice sheet was flowing at the time the drumlins formed. Because the blunt, steep end faces up-ice and the tapered end trails down-ice, a geologist standing among a drumlin swarm can simply look at the alignment of the hills and read off the ancient flow direction as confidently as reading an arrow. This technique has been applied for well over a century and remains a cornerstone of glacial geomorphology today. By mapping drumlin orientations across entire regions and combining them with other flow indicators such as glacial striations scoured into bedrock, elongated ridges called flutes, and crag-and-tail landforms, researchers have reconstructed detailed flow maps of former ice sheets including the Laurentide Ice Sheet that covered most of Canada and the northern United States during the last glaciation, and the British-Irish Ice Sheet. These reconstructions reveal that ice flow was not a simple uniform spreading from a single dome. Flow direction changed over time as ice domes grew, shifted, and thinned, and swarms of differing orientation sometimes overlap or cross-cut one another, preserving a layered record of changing ice dynamics through a glaciation. This directional application of drumlins is considered robust and widely trusted, largely independent of which specific formation mechanism ultimately proves correct, since virtually every proposed mechanism agrees that elongation aligns with ice flow. The open scientific question lies not in what drumlins tell us about direction, but in the physical process that carves or builds that elongated shape in the first place.

The Instability Model: Till, Water, and Ice in Feedback

The leading modern framework for explaining drumlin formation treats the process as an emergent instability arising from coupled interactions between deforming, water-saturated subglacial till and the ice flowing above it, rather than as a simple, one-step carving or dumping event. Beneath an active ice sheet, the bed is often not rigid rock but a layer of till that has been softened by pressurized meltwater trapped between the ice and the ground below. This water-saturated till can deform and flow slowly under the weight and shear stress of the overlying ice, behaving somewhat like a thick slurry rather than a solid. Ice sliding and till deforming together set the stage for the instability process. The process is thought to begin with a tiny, essentially random irregularity in the bed, perhaps a patch of stiffer till, a small bedrock bump, or a slight thickening of sediment. This minor obstacle locally disturbs the smooth flow of ice passing over it, subtly increasing ice pressure on its up-ice face and altering the pattern of subglacial water pressure around it. Where water pressure drops locally, till tends to stiffen and resist erosion; where it rises, till weakens and is more easily entrained by the moving ice above. This is the essence of a self-reinforcing feedback loop. The initial bump perturbs the ice-water-till system just enough to encourage a small amount of extra deposition or reduced erosion immediately down-ice of the obstacle, while the flow disturbance itself gets reinforced and sharpened. Over time, repeated cycles of this feedback progressively amplify the initial irregularity, stretching and streamlining it into an elongated, teardrop-shaped mound aligned precisely with the ice flow direction, since flow-aligned shapes minimize drag and are mechanically favored to grow and persist. Researchers explicitly compare this dynamic to self-organizing bedforms produced by any flowing fluid moving over a deformable, granular surface, such as sand dunes shaped by wind or ripples formed on a riverbed by flowing water. In both cases, an initially minor, almost random irregularity is not simply eroded away or ignored by the fluid but instead grows through feedback into a stable, repeating, larger-scale pattern. The proposal is that subglacial till beneath kilometers of moving ice behaves analogously, though the physical variables involved, ice rheology, water pressure, till strength, and effective pressure at the ice-bed interface, are far more complex than for wind over sand.

Why the Exact Mechanism Remains Debated

Although the instability model has become the leading explanatory framework, it has not fully resolved the century-old debate over how drumlins actually form, and several competing or complementary ideas remain scientifically active. One historical camp emphasized erosional formation, proposing that drumlins are carved by ice and meltwater eroding into a pre-existing, thicker layer of till or sediment, essentially sculpting hills out of an already-present mass the way wind erosion can leave streamlined remnant ridges (yardangs) in a desert. This view is supported by drumlins that contain cores of older, harder material, which are difficult to explain through pure deposition since deposition alone would not naturally embed a pre-existing bedrock knob inside a hill. A second camp emphasized depositional formation, proposing that drumlins are built up progressively as ice deposits successive layers of till in favorable locations, somewhat like sediment accreting around an obstacle in a current. This view fits well with drumlins composed of internally layered or structured till that suggests incremental accumulation over time rather than carving from a single pre-existing mass. The modern instability approach attempts to unify these views by proposing that both erosion and deposition can occur simultaneously and locally as part of the same coupled feedback process, with erosion dominant in some positions relative to the forming hill and deposition dominant in others. However, translating this elegant conceptual model into a precise, testable, quantitative theory that correctly predicts drumlin size, spacing, shape, and internal structure across many different real-world settings has proven genuinely difficult. Several factors keep the debate alive. Direct observation is essentially impossible, since the process occurs beneath kilometers of active ice in remote polar and formerly glaciated regions, forcing researchers to rely on indirect evidence: the internal structure of drumlins exposed in quarries and road cuts, geophysical surveys, numerical and laboratory models, and comparison with landforms beneath modern, still-active ice sheets such as parts of Antarctica. Drumlins in different regions also show real physical variability in composition, internal layering, and the presence or absence of cores, suggesting that no single mechanism may apply universally, and that drumlin formation could represent a spectrum of related processes rather than one uniform recipe. This is a healthy, ongoing area of glaciological research rather than a sign that the basic science is unsettled.

Reading the Landscape: Drumlins in Practice

Field geologists identifying and mapping drumlins rely on a combination of direct observation, topographic maps, and increasingly, high-resolution digital elevation data derived from satellite and airborne laser scanning (lidar), which can reveal subtle streamlined landforms even in areas obscured by vegetation or modern land use. A well-preserved drumlin swarm is often visually striking from the air or on a shaded relief map, appearing as a field of parallel, elongated ovals all oriented the same direction, sometimes likened to a basket of eggs or a school of fish. This pattern-recognition approach allows researchers to rapidly establish regional ice flow directions across huge areas without needing to physically visit every hill. Beyond simple direction, the size, spacing, and density of drumlins within a swarm are thought to carry additional information about the conditions under which they formed, including ice velocity, till availability and water saturation, and the duration of stable flow needed to fully develop the streamlined shape. Larger, more elongated drumlins are generally associated with faster, more sustained, and more consistently directed ice flow, while shorter, less streamlined forms may indicate flow that was slower, shorter-lived, or more variable in direction before the ice retreated or shifted. Drumlins are studied alongside other landforms left by moving ice, including eskers (sinuous ridges deposited by meltwater rivers flowing within or beneath the ice), moraines (ridges of debris marking where an ice margin paused), and flutes (smaller, more elongated ridges formed by similar processes to drumlins but on a finer scale). Together these features let researchers reconstruct not just flow direction but the broader life history of an ice sheet, including where it thickened, thinned, sped up, slowed down, and eventually disappeared. Understanding drumlin formation also has relevance beyond historical reconstruction. The physical processes governing till deformation and subglacial water pressure that are thought to drive drumlin formation are the same processes that control how fast modern ice sheets in Greenland and Antarctica can slide toward the ocean, making this seemingly historical puzzle directly relevant to predicting the future behavior of Earth's remaining ice sheets under climate change.

Frequently asked questions

Try it live

Everything above runs in your browser — open Drumlin Formation Under Ice Sheets and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.