The Thawing Ground of the Far North

AGround that has stayed below freezing for two years or more is, by definition, permafrost, and for most of the twentieth century the explanation of why it thaws was almost as simple as the definition. Warm air in summer deepens the layer that melts each year, and cool summers allow that layer to recover. On such an account the fate of the frozen north could be read off a single curve, the rise in mean summer temperature. Measurements made since have complicated the picture in an awkward way. At a number of sites where summer air has warmed by less than half a degree since 1990, boreholes show the ground itself warming twice as fast, and at some of them the deepest thermometers are climbing while the surface record stays flat. Something other than summer heat is at work, and it appears to be at work in the season when nothing is supposed to happen at all.

BThe explanation now attracting most attention concerns snow. Yelena Marchuk, a permafrost physicist at the Beloyar Institute for Cold Regions Research, contends that the winter blanket matters more than the summer sun, because snow conducts heat poorly and therefore holds the ground beneath it far warmer than the air above. Where the cover is thin, the cold of a Siberian January passes straight into the soil and recharges it for the year ahead; where the cover is deep, that recharge never takes place. Between 2009 and 2019 her group instrumented 64 boreholes across the tundra, pairing each with a snow gauge. Sites carrying more than 40 centimetres of snow in midwinter finished the season with soil at minus 4 degrees, against minus 11 degrees at otherwise similar sites where wind had swept the surface bare. The difference in mean annual air temperature between the two groups of sites was 0.3 degrees.

CWhat makes the mechanism more than a curiosity is that the vegetation is changing as well. Shrubs which once grew waist-high in sheltered hollows now spread across open tundra, and a shrub is, among other things, a fence, since it holds drifting snow where wind would otherwise carry it away. Marchuk therefore describes a loop in which warming encourages shrubs, shrubs trap snow, and trapped snow warms the ground further still. She concedes, however, that the loop turns only under conditions which are far from universal. It needs a landscape windy enough for snow to be redistributed rather than laid down evenly, and it needs soil wet enough to carry heat downwards; on well-drained gravel ridges her instruments record almost no effect at all. She is careful to add that her boreholes measure temperature rather than carbon, and that no direct measurement of gas release was made at any of them.

DNils Aabye, a plant ecologist at the Tarnaby Centre for Northern Ecosystems, does not question the physics but insists that the accounting is one-sided. The shrubs which trap snow in winter, he points out, also shade the surface in summer and draw carbon from the air throughout the growing season, and neither effect shows up in a borehole record. In a valley he has monitored since 2014, shrub cover rose by 19 per cent while soil temperature at one metre rose by only a tenth of a degree, and the plant matter above ground took up an additional 21 grams of carbon per square metre each year. Warming, on this reading, buys roughly as much as it costs. A tundra that grows more plants is, in his phrase, an account with entries on both sides, and the entry which happens to be easiest to measure has been mistaken for the balance itself. Aabye has argued in print that models built on ground temperature alone will overstate the release of carbon from the north by something close to a quarter.

EThe weaknesses of that case are not hidden. Aabye's figures come from a single valley, cover three growing seasons and count only what grows above the surface, whereas the carbon at issue lies metres beneath it, in a layer his method never reaches. Marchuk's reply is that a plant which fixes carbon for a century and a soil which releases it for a millennium cannot be entered in the same column of the same ledger, and that the comparison flatters the vegetation. Aabye accepts the point about duration. He does not accept that it disposes of his objection, since an overestimate of a quarter, repeated across every model in use, would change what any of them is worth. Both men agree that the argument would be settled by measuring gas above the same ground for which soil temperature is recorded, and that hardly anybody does this.

FFor the present the field has arranged itself around a compromise nobody finds satisfying. Snow depth is accepted as a genuine control on ground temperature, shrub growth is accepted as a genuine sink, and the balance between them is admitted to be unknown at any scale larger than a valley. That uncertainty has consequences beyond the science, because national pledges to cut emissions are written in tonnes, and a term which might be either a source or a partial sink cannot be entered in a budget with any confidence. Monitoring, meanwhile, remains sparse: fewer than 300 boreholes report winter data across a region of some 15 million square kilometres, and most of them stand within a day's travel of a settlement, in the sheltered places where people choose to build. The very landscapes most likely to hold deep snow are the least likely to be measured.