Patterned Ground: How Permafrost Sorts Stones into Polygons

The Freeze-Thaw Engine Behind Sorting

Patterned ground forms in the active layer, the shallow zone of soil above permafrost that thaws in summer and refreezes in winter. As freezing progresses downward from the surface, ice lenses and needle ice crystals grow within the pore spaces of the soil, expanding by roughly nine percent as water turns to ice. This expansion is not uniform: because larger stones conduct heat and transmit stress differently than fine-grained sediment, the freezing front exerts a net upward force on clasts, a process known as frost heave. Each winter, stones are nudged a fraction of a millimeter closer to the surface. Once at the surface, gravity and downslope creep during the thaw season carry them laterally, where they accumulate at the boundaries between sorting cells rather than sinking back into the matrix. Repeated year after year, this ratchet-like cycle behaves analogously to Rayleigh-Benard convection in a heated fluid: soil in the center of each cell effectively rises, degrades, and pushes outward, coarse debris migrates to the rim, and a self-organizing cellular pattern emerges from an initially uniform surface. The timescale for a clearly defined polygon or circle to form ranges from decades in active, ice-rich permafrost to many centuries in more stable, drier ground. Moisture content is critical to the whole process, because water must be present and mobile enough to form segregation ice; permafrost soils that are too dry or too coarse rarely develop strong patterning, while saturated silty soils with abundant fines are the most prolific pattern builders. The intensity of patterning is also linked to the number and severity of freeze-thaw cycles per year, which is why the sharpest, most geometrically regular sorted circles and polygons are found in continuous permafrost zones of the high Arctic and high alpine environments, where winters are long, cold, and punctuated by sharp seasonal transitions rather than constant deep freeze.

From Circles to Polygons to Stripes: The Role of Slope

The single most important control on the final shape of patterned ground is surface gradient. On perfectly flat terrain, sorting cells have no preferred direction to migrate, so convective sorting produces roughly circular or polygonal cells with stone borders enclosing fine-grained centers, often called sorted circles or sorted polygons. As slope angle increases, even slightly, gravity begins to bias the slow downhill creep of soil within each cell. The circular cells elongate in the direction of maximum slope, and neighboring cells merge along their downslope edges. Somewhere between about two and seven degrees of slope, circles typically transition into elongated ellipses; beyond roughly ten to fifteen degrees, the cells stretch out fully into parallel bands of alternating stone and fine sediment running straight downhill, known as sorted stripes. This transformation is a beautiful natural demonstration of how a single self-organizing process can produce an entire family of related morphologies depending on one continuous parameter. Field surveys across periglacial regions consistently confirm this relationship, and laboratory freeze-thaw simulations using trays of gravel and silt on tiltable platforms reproduce the same sequence, lending strong support to the convective sorting hypothesis over competing explanations. On very steep slopes, above roughly twenty-five degrees, other processes such as solifluction lobes and debris flow tend to dominate and patterned ground becomes discontinuous or is overprinted by mass-wasting features. Aspect also matters at a secondary level: slopes facing the sun experience deeper and more variable thaw, which can suppress pattern development, while shaded, persistently cold slopes tend to preserve crisper geometric forms. Researchers have also documented transitional forms called sorted nets, an irregular, elongated intermediate morphology between full circles and true stripes, which tend to appear on very gentle slopes of roughly one to two degrees, underscoring that the circle-to-stripe transition is a continuous spectrum rather than a series of abrupt categorical jumps. Cell diameter itself is not fixed either, it tends to scale with the depth of seasonal thaw and the size of the coarsest available clasts, so a site with deep summer thaw and large cobbles typically produces broader-spaced polygons than a site with shallow thaw and only small pebbles available for sorting.

Non-Sorted Patterns and Ice-Wedge Polygons

Not all patterned ground is defined by stone sorting; a second major family forms through thermal contraction cracking and produces what are called non-sorted or ice-wedge polygons. When winter temperatures plunge rapidly, the frozen ground contracts and cracks in a manner similar to drying mud, except the cracks are driven by cold rather than desiccation. Meltwater infiltrates these cracks each spring and refreezes, forming a thin vein of ice. Because ice is mechanically weaker along old crack lines than the surrounding frozen soil, the same cracks tend to reopen in subsequent winters, and the ice veins grow incrementally wider and deeper year after year, eventually becoming massive wedge-shaped bodies of ground ice meters across and extending several meters down. The surface expression is a network of polygons, typically four to thirty meters across, bounded by shallow troughs where the wedges lie beneath, sometimes with slightly raised rims where soil has been pushed up by the expanding ice. Ice-wedge polygons are especially important to permafrost scientists because they reveal information about paleoclimate: relict, fossil ice-wedge polygon networks found in regions no longer underlain by permafrost, including parts of the temperate United States and lowland Europe, are used as evidence of much colder climates during the last glacial period. Unlike sorted circles and stripes, ice-wedge polygons do not depend on stone content and can form in nearly stone-free silts and peats, making thermal contraction cracking a mechanistically distinct but often co-occurring partner process alongside frost-heave sorting across the periglacial landscape. In many high-Arctic settings the two processes overlap directly, with sorted stone borders tracing out cell boundaries that sit almost exactly above buried ice wedges, since both mechanisms independently favor drainage and stress concentration along similar low points in the microtopography. Distinguishing the two in the field usually comes down to trough geometry and excavation, ice-wedge troughs are typically narrower, deeper relative to their width, and directly underlain by a vertical ice body, while sorted-circle borders are broader stone accumulations without a discrete ice core beneath them.

Reading Patterned Ground as a Climate Indicator

Because the formation of patterned ground depends so sensitively on the presence, thickness, and thermal regime of permafrost, geomorphologists and climate scientists treat these features as natural instruments recording ground conditions. Actively forming, crisp sorted circles and polygons indicate permafrost that is currently experiencing vigorous seasonal freeze-thaw cycling with an intact, cold active layer. In contrast, degraded, blurred, or partially collapsed patterned ground is one of the clearest visual signals of permafrost thaw, since warming ground disrupts the frost-heave engine and allows previously sorted cells to slump and mix. Researchers monitoring Arctic warming routinely use aerial and satellite imagery to track changes in the sharpness and areal extent of patterned ground networks as a proxy for permafrost health across huge, otherwise inaccessible regions. Thermokarst features, including sagging troughs above melting ice wedges and irregular pits where ice-rich ground has collapsed, frequently develop directly out of formerly well-organized polygon networks, providing a visible timeline of degradation. Beyond Earth, patterned ground has also captured planetary scientists' attention because strikingly similar polygonal terrain has been photographed on Mars by orbiting spacecraft and surface rovers, interpreted as strong evidence for past or present ground ice and analogous freeze-thaw or thermal contraction processes on another planet. Comparative study of terrestrial patterned ground therefore extends well beyond Arctic field geology, informing both climate change monitoring on Earth and the search for water ice on other worlds. The scale invariance of the underlying physics, from centimeter-scale laboratory trays to kilometer-scale Martian polygons, is part of what makes patterned ground such a compelling subject for physical and computational modeling. Long-term monitoring plots established decades ago in places such as Svalbard, Alaska's North Slope, and the Antarctic Dry Valleys now provide multi-decade time series that document measurable acceleration in pattern degradation coinciding with regional warming trends, giving climate scientists ground-truthed evidence that directly complements the broader satellite record and strengthens confidence in patterned ground as a genuine, quantifiable indicator of permafrost health rather than a purely qualitative visual cue.

Modeling Patterned Ground: What Simulations Capture

Numerical and physical models of patterned ground formation generally combine three coupled elements: a heat-transfer model that tracks the depth and rate of the freezing front through the active layer, a mechanical model of frost heave that converts ice segregation into vertical and lateral forces on soil particles, and a sediment-transport rule that governs how heaved stones migrate once they reach zones of higher slope-driven stress. Early conceptual models, developed from the 1950s onward, treated the process largely as an analogy to convection cells in a heated fluid, which successfully explained the general circle-to-stripe transition with slope but could not capture fine details like polygon cell size or the timescale of pattern maturation. More recent cellular-automaton and finite-element approaches explicitly simulate individual clasts or discretized soil parcels responding to repeated heave-and-settle cycles, reproducing realistic cell diameters, sorting sharpness, and the sensitivity of pattern type to variables such as soil moisture, freeze-thaw frequency, and clast size distribution. These models consistently show that pattern wavelength, meaning the typical spacing between adjacent stone borders, scales with the depth of the active layer: deeper thaw layers produce wider, coarser-spaced polygons, while shallow active layers produce tightly packed, smaller-diameter cells. Validating these models against real Arctic field sites, where researchers have measured stone displacement rates directly using buried magnets, dyed tracer stones, and repeat high-resolution photography, remains an active area of periglacial geomorphology research, and discrepancies between predicted and observed sorting rates continue to refine scientists' understanding of exactly how efficiently frost heave converts thermal energy into mechanical sorting work.

Frequently asked questions

Try it live

Everything above runs in your browser — open Patterned Ground: How Permafrost Sorts Stones into Polygons and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.