Temperature measurements and heat transfer in near-surface snow at the South Pole

To study near-surface heat flow on the Antarctic ice sheet, snow temperatures were measured at South Pole Station to a depth of 3 m at 15 min intervals during most of 1992. Solar heating and water-vapor transport were negligible during the 6 month winter, as was inter-grain net thermal radiation, leaving conduction as the dominant heat-transport mechanism. The rate of temperature change at depth over 15 min intervals was smaller than that at the surface, by one order of magnitude at 20 cm depth and two orders of magnitude at I m depth. A finite-difference model, with conduction as the only heat-transfer mechanism and measured temperatures as the upper and lower boundary conditions, was applied to four sets of three thermistors each. The thermal conductivity was estimated as that which minimized the difference between modeled and measured 15 min changes in temperatures at the center thermistor. The thermal conductivity obtained at shallow depths (above 40 cm) was lower than that given by existing parameterizations based on density, probably because the snow grains were freshly deposited, cold and poorly bonded. A model using only vertical conduction explains on average 87% of the observed 15 min temperature changes at less than 60 cm depth and 92% below 60 cm. The difference between modeled and measured temperature changes decreased with depth. The discrepancies between model and observation correlated more strongly with the air-snow temperature difference than with the product of that difference with the square of the wind speed, suggesting that the residual errors are due more to non-vertical conduction and to sub-grid-scale variability of the conductivity than to windpumping. The residual heating rate not explained by the model of vertical conduction exceeds 0.2 W m(-3) only in the top 60 cm of the near-surface snow.