Glacier flow

How can solid ice flow downhill like a very slow river?

Ice is brittle. Drop a cube on a tiled floor and it shatters; it does not spread. Yet a valley glacier delivers ice from its high basin to its snout year after year, bending around bedrock spurs and thickening into hollows, without melting and refreezing its way down. So what is a solid doing when we say it flows — and is it one thing happening, or two?

Ask first what a glacier has that an ice cube does not. Depth. A hundred metres of ice presses on whatever lies beneath it, and on a slope that pressure has a component pointing downhill. The deeper you go, the more weight is stacked above, so the shear stress on a layer grows with depth and with the sine of the slope. That already tells us where to look for movement: near the bed, not near the surface.

But stress is not motion. What can a crystal actually do under a sustained push? Ice crystals deform preferentially along one set of internal planes, rather as a deck of cards slips card over card. Defects in the lattice migrate; grains rotate, recrystallise, and reorganise so their easy-slip planes line up with the shear. None of this is melting. It is creep, and it is slow — but a glacier has centuries.

Here is the consequence worth pausing on. Measure how fast ice creeps against how hard you push it and the relationship is not a straight line: strain rate rises roughly as the cube of the stress, the exponent commonly taken as about three, though its true value is argued and probably varies with stress regime and impurity content. Double the driving stress and you get something like eight times the deformation rate. So a glacier that thickens a little speeds up a lot, and almost all deformation concentrates in the deepest, most stressed ice.

Now a second question. If internal creep were the whole story, ice welded to bedrock would not move at all and the velocity would fall smoothly to zero at the bed. Is that what we see? Sometimes — cold-based glaciers, frozen to their beds in polar interiors, behave close to that. But many glaciers, when you drill through them, are moving at the bed itself, and sometimes most of the surface speed is bed motion.

What could let ice slide on rock? Ice at the bed of a temperate glacier sits at its pressure-melting point, so a thin film of water is present; around small bumps it melts on the high-pressure upstream side and refreezes downstream, and around larger ones it simply creeps past. Meltwater from the surface, draining down crevasses and moulins, raises water pressure at the bed and effectively floats the ice off it, reducing friction. That is why many glaciers speed up measurably within hours of a warm afternoon — an unambiguous sign that the bed, not the ice, is in charge that day. A third mechanism belongs here: where the bed is soft sediment rather than rock, the sediment itself deforms and carries the ice along.

The strongest evidence that basal conditions govern flow comes from surge-type glaciers, which spend decades quiescent, thickening upstream, then abruptly accelerate tenfold or more for months to a year or two. The classic instrumented case, Variegated Glacier in Alaska, surged in 1982-83 and ended with an outburst of turbid water — consistent with the subglacial drainage reorganising from a distributed, high-pressure system into efficient channels that drained the bed and restored the friction. The ice itself did not change. The plumbing did.

The tempting misconception is that a glacier slides because the pressure of its own weight melts the ice underneath it. Pressure does lower the melting point, but only slightly — far too little to melt a bed that is genuinely cold. Warm-based glaciers have wet beds mainly because geothermal heat, frictional heating from the sliding itself, and surface meltwater draining down keep them there. Whether a glacier slides is a question about its thermal state and its drainage, not about how heavy it is.

Note also what "flow" is not. Crevasses are real and they are brittle fracture: near the surface, where confining pressure is low, ice subjected to fast extension cracks rather than creeps. The same material is ductile below and brittle above, depending on stress rate and depth. And this is bulk flow, a different matter from why an ice surface is slippery underfoot — that is a question about a thin surface layer, not about a kilometre of deforming crystal.

• Why ice streams in Antarctica move a thousand times faster than the ice on either side of them.
• How a glacier's response time to a change in climate depends on its thickness and slope.

Every term the collection defines is gathered in the glossary.