Evaluation of mesoscale dispersion modelling during CAPITOUL
Titre de la revue : Meteorology and Atmospheric Physics
Volume : 102
N° : 3-4
Pagination : 263-287
Date de publication : 10/12/2008
Atmospheric transport and dispersion were investigated during CAPITOUL campaign using the measurements of sulfur hexafluoride (SF6) tracer. Six releases of SF6 tracer were performed (2004 March 9-10-11 and July 1-2-3) in a sub-urban area of Toulouse agglomeration, in relation with IOP of CAPITOUL. Concentration data were collected both at the ground along axes perpendicular to the wind direction at distances between 750m and 5000m from release point, and at 100m and 200m heights thanks to aircraft flights. Meteorological situations were all associated to daytime anticyclonic conditions with weak winds and convective boundary layer. However, concentration measurements show a large broad of plumes and complex dispersion associated to inhomogeneous turbulence. A meso-scale dispersion modelling system, PERLE, developed at Meteo-France for environmental emergency in case of atmospheric accidental release, has been evaluated for the different release experiments, in its operational configuration. PERLE is based on the combination of the non hydrostatic meso-scale MESO-NH model, running at 2km horizontal resolution, and the lagrangian particle model SPRAY. The system has captured the essential features of the plume, with different degrees of agreement. Meso-NH correctly reproduced the meteorological and turbulent fields, with only a few errors on the wind directions and an absence of shallow cumulus formation one day that induced an overestimation of the sensible heat flux. In the cases where Meso-NH estimates accurately the meteorological and turbulent fields, PERLE gives a realistic prediction of the plume according to SF6 concentration measurements. But it also shows a tendency to consistently underestimate the horizontal dispersion at low levels. This discrepancy was mainly attributed to the inappropriate assumption of isotropic turbulence to calculate the turbulent velocity variances, whereas the formulation of the lagrangian time scales has also a significant impact on the lateral spread. Based on research tools, the system PERLE has been implemented into a reliable software package, prompting for user-input in the context of an emergency response situation.