Long term modelling of the dynamical atmospheric flows over SIRTA site. Issue 172 (January 2018)

• Record Type: Journal Article
• Authors: Chahine, A.; Dupont, E.; Musson-Genon, L.; Legorgeu, C.; Carissimo, B.
• Is Part Of: Journal of wind engineering and industrial aerodynamics. Issue 172(2017)
• Page Start: 351
• Page End: 366
• Publication Date: 2018-01
• Subjects: Turbulence -- Clustering -- Downscaling -- Code_Saturne -- SIRTA -- Meteorological clustering -- Microscale; Wind-pressure -- Periodicals; Buildings -- Aerodynamics -- Periodicals; Pression du vent -- Périodiques; Constructions -- Aérodynamique -- Périodiques; Buildings -- Aerodynamics; Wind-pressure; Periodicals
• DOI: 10.1016/j.jweia.2017.09.004
• Languages: English
• ISSNs: 0167-6105

Abstract

Abstract: The atmospheric flow knowledge is important for its role in pollutant dispersion and wind energy. In this work, the hourly atmospheric flow output (8760 states) from Weather Research and Forcasting (WRF) model for the year 2011 over SIRTA (Site Instrumental de Recherche par Télédétection Atmosphérique) are analyzed and clustered into a finite number of representative atmospheric states using two clustering methods: non-controlled clustering and controlled clustering. The resulting representative situations of those clusters are used to specify boundary conditions for flow downscaling over the heterogeneous SIRTA. For flow downscaling, the CFD code Code_Saturne is used to simulate each representative atmospheric state. To assess the efficiency of WRF clustering and Code_Saturne downscaling, the measurements in SIRTA over the same year are used as reference. The Mean Absolute Error (MAE) and the Kullback-Leibler divergence (KL) metrics were computed for the distributions of the atmospheric flow features in order to: (i) compare the difference between the performance of the two clustering procedures, and (ii) compare the distribution of flow properties between WRF mesoscale model and Code_Saturne. It is clearly demonstrated that the two clustering methods are comparable in benefit, and that Code_Saturne improves considerably the flow features modeling in comparison to measurements. Highlights: Clustering is a good method to analyze big data relative to the atmospheric states. CFD tool combined with clustering methods allows reduction of the CPU cost of computation. CFD code Code_Saturne allows the simulation of all encountered atmospheric stability states. The nesting of CFD code Code_Saturne with the mesoscale model WRF allows improvement of the flow features over a complex site. The atmospheric flows obtained with the controlled and non-controlled clustering are similar.