Dirt is a snowscape architect. Whether the sun sculpts narrow peaks, rounded dimples or a web of ridges in a smooth snow field depends on whether the snow is clean, dirty, or really dirty, Meredith Betterton of Harvard University in Cambridge, Massachusetts, has found1.

'Ablation structures', as these mysterious melt formations are properly known, come in many shapes and sizes. 'Penitentes' are narrow peaks of snow up to six metres high. They look, as their name suggests, like a procession of white-robed monks, and occur high in the Himalayas. When Charles Darwin came across them in 1835 in the Andes, he recorded how they made it "difficult for the cargo mules to pass".

Elsewhere, snow melts into 'sun cups', smaller, smoother, rounded dimples, like the inside of an egg carton. It can also form webs of polygon-shaped dirt ridges, with cleaner depressions in the centre of the polygons. Or dirt can become marshalled into dirt cones: peaks of snow or ice coated in a thick layer of debris. Cones up to 85 metres high have been reported in the Himalayas.

A depression in snow, caused by the sun's heat, is self-amplifying. Snow reflects sunlight, so troughs receive more sunlight than peaks, with light bouncing to the bottom off the sloping sides. Once it begins to develop, a trough rapidly deepens.

Dirty snow dampens reflections, increasing the amount of light (and heat) that the snow absorbs. But thick dirt can insulate the snow and slow the melting. So dirt's effect on ablation is subtle and not easy to predict.

Betterton has used these ideas to construct a mathematical model that estimates the initial size and shape of irregularities that form in a field of roughly even snow exposed to sunlight. She shows that, in clean snow, self-amplifying depressions will appear with a characteristic size, like wind or wave ripples in sand, leading ultimately to a field of sharply peaked penitentes all of more or less the same size.

Fleetingly present ridges, the model predicts, occur when a thin layer of dirt suppresses reflections and increases light absorption, smoothing out the melting process. As the snow melts, the dirt gathers on ridges and thins out in valleys. The result is that ridges constantly form, gather dirt and melt faster until they disappear.

A thick layer of dirt behaves differently again, because it insulates the snow. In this case, dirt accumulating on ridges makes them ever more prominent, since they are protected from the sunlight eating away the valleys. This leads to the formation of polygonal webs of dirty ridges, or dirt cones.

Now Betterton hopes to test her model in the lab with artificial illumination -- far easier than watching the slow formation of these snow ablation structures in the inclement heights of the Andes.