Problems in the determination of sensible heat flux over snow

Measurements, as well as the majority of presentday GCM simulations, indicate an increase in air temperature, especially for polar regions. In order to improve the quality of GCM predictions for these regions, it is necessary to achieve good parameterization of the energy exchange, especially of the sensible heat flux between atmosphere and snow, because the high latitudes are snow-covered for as many as nine months. During the snowmelt periods in 1995 and 1998, eddy-covariance and profile measurements of wind speed and air temperature were carried out in the Kdrkevagge, a valley in the Abisko Mountains in Swedish Lapland. Sensible heat flux was calculated by the profile and bulk method and compared to the directly measured sensible heat flux. The profile and the bulk method gave reliable results as long as all data were measured within the surface layer and radiation errors for air-temperature data were small. The bulk method yielded better results than the profile method in periods with high values of upwelling shortwave radiation. All measurements conducted outside the surface layer, which was frequently shallow in 1998, were unusable for the profile method. Similarly, the eddy-covariance method produced unreliable results if the instrument was operating outside the surface layer. In such a case, however, the bulk method produced at least a minimal estimate for the sensible heat flux. None of the methods yielded reasonable results for all meteorological conditions. This aspect should be noted when seeking new methods to improve the calculation of sensible heat flux in GCMs.