SUMMARYFor very stable boundary layers there is no well-accepted theory today. In this study, an improved Prandtl model with varying diffusivity is applied to less than ideal conditions for pure katabatic flow pertaining to very stable boundary layers. We find that the improved Prandtl model adequately describes the usual and persistent katabatic glacier wind on Breidamerkurjökull. This is true even for flows with very different heights and strengths of the jet. A theoretical estimate of the katabatic jet height, based on temperature deficit and lapse rate, is verified. The calculated surface fluxes compare well with the measured turbulence parameters. A possible reason for the robustness of the katabatic jet (and other low-level jets) is given in terms of the Scorer parameter.

A realistic simulation of katabatic flows is not a straightforward task for numerical models. One complicating factor is that katabatic flows develop within a stably stratified boundary layer, which is poorly resolved and described in many numerical models. To capture the jet-shaped shallow flow a model set-up with high vertical resolution is also required. In this study, 'a state of the art' mesoscale numerical model is applied in a simulation of katabatic flow over a melting glacier. A basic agreement between observations and model results is found. From scale analysis, it is concluded that the simulated flow can be classified as katabatic. Although the background flow varies in strength and direction, the simulated katabatic flow over Breidamerkurjökull is persistent. Two factors vital for this persistence are identified. First, the melting snow maintains the surface temperature close to 0 • C while the air temperature warms adiabatically as it descends the slope. This provides a 'self enhanced' negative buoyancy that drives the flow to a balance with local friction. Second, the jet-like shape of the resulting flow gives rise to a large 'curvature term' in the Scorer parameter, which becomes negative in the upper jet. This prevents vertical wave propagation and isolates the katabatic layer of the influence from the free troposphere aloft. Our results suggest that the formation of local microclimates dominated by katabatic flow is a general feature over melting glaciers. The modelled turbulence structure illustrates the importance of nonlocal processes. Neglecting the vertical transport of turbulence in katabatic flows is not a valid assumption. It is also found that the local friction velocity remains larger than zero through the katabatic jet, due to directional shear where the scalar wind speed approaches its maximum.

The Ekman boundary-layer model is extended analytically for a gradually varying eddy diffusivity K(z) ! 0, z ! 0. A solution for the Ekman layer is provided having similar structure to the constant-K case; that is, exponentially decaying sine functions for the two horizontal wind components. The analytical asymptotic solution compares well with its numerical counterpart for various K(z). The result can be useful in theoretical studies such as Ekman pumping, for efficient estimation of the Ekman layer profiles in various analyses with near-neutral stratifications, or for a rapid initialization of mesoscale models.

No abstract