| Summary: | Parameterization of turbulent heat fluxes with bulk aerodynamic methods has been shown to perform relatively poorly in the presence of shallow katabatic winds. Identifying when and how the bulk methods under-perform is essential to reduce uncertainty in surface energy balance modelling of glacier melt. Here, we evaluate the most commonly used bulk method in simulating 30-min sensible heat fluxes over two summer months at the Kaskawulsh glacier in the Yukon, Canada. As our reference data, we use eddy-covariance (EC) measurements from one on-glacier site at three different heights (1 m, 2 m, and 3 m). To adequately process the EC-data, we propose two new methods: a statistical method that ensures the fluxes are derived from time windows with (near-)stationary turbulence, and a filtering method that ensures the fluxes are representative of surface conditions, in particular during shallow katabatic winds. We find that the agreement between EC-derived fluxes and those modelled with the bulk method substantially improves with the implementation of the two data-processing methods. The least data-processing is required for the measurements at 1 m above the surface, the height that also yields the best performance of the bulk method. Contrary to the previous findings, the bulk method does perform well during shallow katabatic flows if the measurements of temperature and wind speed are taken close to the surface (<= 1 m) and well below the wind speed maxima. We conclude that adequate data processing and filtering are critical in obtaining accurate sensible heat fluxes at glacier surfaces.
|
|---|