Modeling surface exchange and heat transfer for the shallow snow cover at SHEBA

The snow cover on sea ice acts as an insulator against frigid Arctic air and inhibits wintertime ice growth at the ocean interface. It has an especially disproportionate effect in the fall, when snow is less dense, and even a shallow cover can dramatically slow cooling of the sea ice. We use a detailed, one-dimensional model of snow, SNTHERM, to simulate heat exchange within and at the surface of the shallow snow cover at SHEBA (Surface Heat Budget of the Arctic Ocean). Although SNTHERM was developed originally for snow-covered ground, we recently adapted it to also handle snow-covered sea ice. We have tested this polar version with data from Russian drifting station North Pole 4 (NP-4) and from Ice Station Weddell (ISW). Extensive yearlong measurements from SHEBA provide data for further testing our model and for investigating the thermal regime within this shallow snowpack.

Our investigation simulates snow cover at the "Pittsburgh" thermistor location near the main SHEBA camp. The SHEBA site had 7 cm of snow when measurements began in mid-October 1997. The snow cover built to a maximum depth of around 45 cm before totally melting in mid-June of 1998. Snowfall resumed in mid-September. We drive SNTHERM with hourly-averaged observations of air temperature, relative humidity, wind speed, incoming and outgoing solar radiation, and incoming longwave radiation from the Atmospheric Surface Flux Group site about 100 m from Pittsburgh. Because SNTHERM does not predict ice accretion or ablation at the ice underside, we derive a lower boundary condition from thermistor readings at 180-cm depth in the ice.

We estimate turbulent fluxes at the snow surface with a routine recently developed for ISW, which includes new parameterizations for the roughness length z0 and a variable von Kármán "constant". Simulated momentum and heat fluxes correlate well with approximately 3,500 hours of eddy-correlation measurements. Simulated time series of surface temperatures match radiometric observations to within 1°C, thus corroborating the correctness of our surface flux simulation. We validate in-snow thermal processes by comparing simulated temperature traces and profiles with thermistor readings at 5 levels within the snow cover. Agreement is close, except during a cold period in February when SNTHERM overpredicts temperatures in the lower snowpack by about 5°C. Possible reasons for this discrepancy could be that SNTHERM underpredicts snow density or omits wind effects in computing an effective thermal conductivity.

Because detailed snow models are not practical for GCM applications, it is instructive to examine the accuracy of simpler snow treatments. We therefore present two alternate SNTHERM simulations of turbulent exchange and in-snow temperatures. In the first, we assume a homogeneous snowpack; and, in the second, we additionally limit the model to two snow layers.