Observation and modelling of stable atmospheric boundary layer over complex terrain : the katabatic wind process
jury members
ROTACH Mathias, Université d’Innsbruck, Rapporteur
MASSON Valéry, Meteo-France de Toulouse, Rapporteur
STAQUET Chantal, Université de Grenoble Alpes, Examinateur
GRISOGONO Branko, Université de Zagreb, Croatie, Examinateur
COHARD Jean-Martial, Université de Grenoble Alpes, Examinateur
BRUN Christophe, Université de Grenoble Alpes, Directeur de thèse
Abstract
The stable atmospheric boundary layer, particularly in complex terrain, is not yet fully understood and it is thus still inadequately modelled. A surface cooling of a sloping terrain generates a gravity flow (known as a katabatic wind) due to local density increase. This flow behaves as a wall-bounded turbulent jet, often simply modelled by a local balance between the buoyancy force and the turbulent friction (the Prantdl model). In mountainous regions, the wind maximum is typically observed at a height (zj) of 1 to 10 m above the ground. The wall-bounded jet is responsible for a momentum-flux sign change and a heat-flux variability close to the ground. Those turbulent-flux variabilities are fully conflicting with the aplicability of the Monin-Obukhov similarity theory (MOST), which is nevertheless universally used in the atmospheric models to provide the surface boundary condition. If the MOST is already questionable for the very stable cases, it is obviously not valide over sloping surfaces because it neglectes the coupling of the wind and temperature equations, which constitutes the katabatic source. Hence, it is not possible to adequatly model a katabatic flow (zj O(1m)) using the MOST, especially with a vertical resolution of the order of magnitude of the wind-maximum height. The aim of the this PhD work is thus to improve the current understanding and modelling capacity of the katabatic winds.
Since data sets of turbulent-katabatic-flow measurements are still scarce, a new field campain was carried out on a steep slope (20◦ to 40◦) overlooking the Gresivaudan valley, near Grenoble : the west face of the Grand-Colon mountain (Belledonne ridge, French Alps). The experimental setup was mainly composed of a 6m mast with four sonic-anemometer levels (1m, 2m, 4m and 6m) to measure the turbulence on both sides of the katabatic jet. The spectral analysis shows the hight sensitivity of the local flow to external perturbations, even when these are weak. The hight-frequency subrange shows a classical behaviour (energy-injection frequency, inertial subrange), but the spectra of the intermediate and low-frequency subranges are less typical : turbulent perturbations with an energy of the order of magnitude of the local injection are present. A specific cospectra Cuw behaviour of the katabatic flows is shown : negative and positive cross-correlations overlap gradually, increasing z. The MOST fails in representing the observed flow and a surface-flux alternative estimation is succesfully used to describe the friction vellocity.
The multi-layer 1D surface model of ISBA (Surfex, Météo-France), which is initially only diffusive (or aimed at modelling exchanges through the canopy) is modified to model katabatic flows. The model is firstly validated with a standard calibrated Prandtl model (with variable eddy difusivity). Secondly, the field data are modelled both with a prescribed effective diffusivity (from data) and using the 1.5-order turbulence scheme. The mean velocity and temperature fields are well reproduced, but it appears that the model is over-diffusive (which generates excessive fluxes), even when an adapted mixing-length is used (z-less parametrisation, depending on local shear and stability).
Realistic 3D LES simulations (Meso-NH, Météo-France) are computed with high resolution (dx × dy × dz = 10m×10m×2m) to model the field data. Spatial flow variabilities over sloping terrain are finely represented, but are biased, mainly due to the using of MOST for the surface boundary counditions. The using of MOST shifts the start of the katabatic source detection by the atmospheric model to a height of 2 m, while the katabatic source reaches its maximum at the surface. Analytical katabatic models (of the Prandtl type, which could be easily used to feed surface boundary counditions) need an a priori definition of the eddy and heat diffusivities. Currently, the general definition of these diffusivities is only possible by the use of turbulent models that include closures. The coupling of the previously-presented multi-layer 1D surface model (validated off-line) is suggested to overcome the lack of physics description in the classic surface boundary counditions. Preliminary work on this coupling is developed and perspective solutions are proposed.