A Météo France, le modèle opérationnel AROME a une résolution horizontale de 2,5 km. L'augmentation des moyens de calcul permettra au prochain modèle opérationnel de tourner à des résolutions de l'ordre ou inférieures au kilomètre. Il entrera donc dans une gamme de résolution appelée zone grise de la turbulence. A ces échelles, les plus grandes structures turbulentes, qui étaient jusqu'alors entièrement sous-maille, devraient être en partie résolues. Cette thèse a permis de définir ce que les modèles devaient obtenir aux échelles kilométriques et sub-kilométriques, c'est-à-dire les parts sous-maille et résolue de référence de la turbulence dans la zone grise. Ces références ont été établies dans le cas de couches limites convectives en convection libre ou forcée, nuageuse ou non. Elles permettent de prouver qu'à hauteur de couche limite égale, les thermiques sont plus larges dans les couches surmontées de nuages. Elles indiquent surtout que, quelle que soit la configuration, les paramétrisations actuelles ne sont pas capables de reproduire la zone grise. Ces échelles demandent donc de développer une nouvelle paramétrisation de la turbulence. La représentation de la turbulence non locale est la part qu'il faut faire évoluer. Nous avons donc pris le parti de modifier le schéma de thermique en flux de masse. Pour étudier les structures cohérentes sous-maille de couche limite, nous avons créé une analyse conditionnelle permettant de circonscrire la part de thermique qui influence le schéma sous-maille en fonction de la résolution. Cet outil nous a permis de définir les caractéristiques des thermiques sous-maille dans la zone grise, mais également de vérifier à micro-échelle les hypothèses de méso-échelle des schémas en flux de masse. Nous avons démontré que toutes les hypothèses ne sont pas valables. Finalement nous avons établi le système d'équations d'un schéma en flux de masse qui fonctionne aux échelles kilométriques. / The turbulence is well-represented on grid coarser than 2 km. Indeed, in meso-scale models, the turbulence is entirely sub-grid. The turbulence is also well-represented at very high resolution (10 to 100 m) by LES models for which turbulence is mainly resolved. However we do not know which part of the turbulence should be resolved and which part of it should be parameterized when a model runs at kilometric scales, the so-called “Terra Incognita“ from Wyngaard (2004). Thanks to increasing computational resources, in a near future, limited area NWP models will reach grid spacings on the order of 1 km or even 500 m. The aim of this study is to develop a parameterization which will provide adequate turbulence to these new-generation, high-resolution models. At first, this study describes a new diagnostic based on LES, which clarifies which part of turbulence should be parameterized at kilometric scales. This reference called “partial similarity function“ is a precious tool to quantify the error made by atmospheric models when running at kilometric scales. These errors are quantified for a state-of-the-art meso-scale model (Méso-NH) with several turbulence mixing options : different mixing lengths, different dimensionalities, a K-gradient scheme or an EDMF approach (K-gradient with a mass-flux scheme). K-gradient turbulence schemes are unable to reproduce the counter-gradient zone. In the grey-zone, this characteristic has a disastrous effect. As the instability is too large, the boundary layer is mixed by the dynamic of the model and the resolved mixing is too strong. The counter-gradient zone can be reproduced by adding a mass-flux scheme to the K-gradient turbulence scheme (Pergaud et al. (2009)). However the mass-flux scheme in its original form only produces wholly subgrid thermals at a grid size for which boundary-layer thermals should be partly resolved. In this case, the subgrid mixing is too strong. So the question arises as what is a subgrid thermal in the “grey zone“, when the mesh contains one thermal at the most and a part of the thermal has to be resolved by the advection scheme of the model. A conditional sampling is defined in order to detect the subgrid part of the thermals. It allows to determine the characteristics of the subgrid thermals in the “grey zone“ and to find out which assumptions of the mass-flux schemes are not verified. In the light of this study, the mass-flux scheme equations are established by taking the thermal fraction and the resolved vertical velocity into account. Finally, the system of equations is closed. The new parameterization is valid in the grey zone.

The first part of the thesis assesses the aerosol direct radiative effect (ADRE) with a focus on ground-based AERONET and satellite MODIS measurements. The AERONET aerosol climatology is used, in conjunction with surface albedo and cloud products from MODIS, to calculate the ADRE and its normalized form (NADRE) for distinct aerosol regimes. The NADRE is defined as the ADRE normalized by optical depth at 550 nm and is mainly determined by internal aerosol optical properties and geographical parameters. These terms are evaluated for cloud-free and cloudy conditions and for all-mode and fine-mode aerosols. We find that the NADRE of fine-mode aerosol is larger at the TOA but smaller at the surface in comparison to that of all-mode aerosol. Cloudy-sky TOA ADRE with clouds is sensitive to the relative location of aerosols and cloud layer. The high-resolution MODIS land surface albedo is also applied to study the clear-sky ADRE over North Africa and the Arabian Peninsula for summer 2001. TOA ADRE shows the high spatial variability with close similarity to that of surface albedo. The second part of the thesis is to develop a 2-D conceptual model for a climatological convective boundary layer over land as a persistent and distinct component in climate models, where the convective-scale motion is explicitly described by fluid dynamics and thermodynamics while the smaller scale effect is parameterized for a neutral stratification. Our conceptual model reasonably reproduces essential statistics of a convective boundary layer in comparison to large eddy simulations. The major difference is that our model produces a better organized and more constrained spatial distribution with coherent convective cells. The simulations for a climatological convective boundary layer are conducted for a prescribed constant and homogenous surface heat flux and a specified cooling term representing the background large scale thermal balance. The results show the 2-D coherent structures of convective cells with characteristic scales comparable with PBL height; downward maximum velocities being 70-80% of the accompanying upward maxima; vertical profiles of a constant potential temperature and linear decreasing heat fluxes; a square-root increase in the velocity magnitude with increasing surface heat flux.

Submitted by Morgana Andrade (morgana.andrade@ufes.br) on 2016-03-22T17:05:04Z No. of bitstreams: 2 license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Pedro Henrique Bonfim Pantoja.pdf: 1965227 bytes, checksum: c6e11a3e91300bad3c95be08b469415f (MD5) / Approved for entry into archive by Patricia Barros (patricia.barros@ufes.br) on 2016-03-23T14:21:49Z (GMT) No. of bitstreams: 2 license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Pedro Henrique Bonfim Pantoja.pdf: 1965227 bytes, checksum: c6e11a3e91300bad3c95be08b469415f (MD5) / Made available in DSpace on 2016-03-23T14:21:49Z (GMT). No. of bitstreams: 2 license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Pedro Henrique Bonfim Pantoja.pdf: 1965227 bytes, checksum: c6e11a3e91300bad3c95be08b469415f (MD5) / Neste trabalho, apresenta­se uma solução para equação de difusão­advecção considerando o  termo  contragradiente  que  é um termo  adicional. Esse termo  adicional  contém informações  sobre a assimetria, escala de tempo Lagrangeana e velocidade turbulenta vertical. A solução  da equação foi obtida pela utilização da técnica de Transformada de Laplace, considerando a  Camada  Limite  Planetária  (CLP)  como  um  sistema  de  multicamadas.  Os  parâmetros  turbulentos foram derivados da teoria de difusão estatística de Taylor, combinada com a teoria  da similaridade. Assim, são apresentadas simulações para diferentes valores de assimetria, o  que  propiciou  a  obtenção  de  uma  concentração  de  contaminantes  em  diferentes  alturas,  em  uma  camada  limite  convectiva.  A  avaliação  do  desempenho  do  modelo,  considerando  a  assimetria  no  processo  de  dispersão  de  poluentes  atmosféricos, foi realizada  através  de  um  experimento  de  tanque  convectivo  tradicional.  Nesse  experimento,  o  termo  contragradiente  influenciou a concentração de poluentes para uma camada limite convectiva. Entretanto, com  as  parametrizações  utilizadas,  o  modelo  não  conseguiu  captar  de  forma  eficiente  o  comportamento da concentração em pontos mais distantes da fonte. / In this paper presents a solution to the advection­diffusion equation considering the term is an  additional  term  countergradient.  This  additional  term  contains  information  asymmetry,  Lagrangian  time  scale  and  vertical  turbulent  velocity.  The  solution  of  the  equation  was  obtained  by  using  the  technique  of  Laplace  transform,  considering  the  planetary  boundary  layer (PBL)  as  a multilayer system. The  turbulent  parameters were  derived from statistical  distribution  theory  Taylor,  combined  with  the  theory  of similarity.  Hence,  Simulations  for  different  values  of  asymmetry, which  allowed to  obtain  a  concentration  of  contaminants  at  different  heights  in  a  convective  boundary  layer  is  displayed.  The  evaluation  of  model  performance,  considering  the  asymmetry  in  the  dispersion  of  air  pollutants  process  was  conducted  through  an  experiment  of  traditional  convective  tank.  In  this  experiment,  the  countergradient  influenced  the  concentration  of  pollutants  in  a  convective  boundary  layer.  However, with the parameterizations used, the model failed to capture efficiently the behavior  of concentration at points further away from the source.