Les régimes dynamiques dans les basses couches de l’atmosphère à l’île de la Réunion sont conditionnés par l’action du relief et du rayonnement sur l’écoulement synoptique. L’île est située en permanence dans le flux des alizés de sud-est et son relief élevé, culminant à 3000 m dans le centre de l’île et 2600 m au sud, constitue un obstacle important pour l’écoulement moyen. Le relief, le chauffage différentiel des pentes et le contraste thermique avec l’océan influent sur les échanges locaux entre la couche limite marine et la troposphère libre.L’analyse des phénomènes complexes de l’écoulement atmosphérique au niveau de l’île a pour but initial de caractériser les transferts de polluants émis localement. Cette étude a deux applications : • La première rentre dans le cadre préparatoire à la mise en place, à 2200 m d’altitude sur le Piton Maïdo, de l’observatoire atmosphérique de l’OPAR, à l’horizon 2011. L’objectif est de comprendre les circulations locales induites par le relief et les transports associés afin de discriminer d’éventuelles pollutions par les sources locales sur le signal qui sera mesuré in situ au sommet du Maïdo.• La seconde vise à étudier la diffusion des panaches volcaniques du Piton de la Fournaise. L’éruption majeure d’avril 2007 du volcan réunionnais a montré que des panaches pouvaient générer d’importantes pollutions dans diverses parties de l’île allant jusqu’à poser de réels problèmes environnementaux et de santé publique.Ces applications s’appuient sur une étude par modélisation numérique à haute résolution des écoulements atmosphériques dans les basses couches au niveau de l’île, au moyen du modèle météorologique de recherche MésoNH, permettant de conduire un ensemble de simulations sur cas idéalisés puis sur cas réels avec diffusion de traceurs passifs. / Reunion Island is an isolated mountainous island in Indian Ocean culminating in the center at 3000 m and in the southern part at 2600 m asl and is located in the influence of very regular south-easterly trade winds. The perturbation of the prevailing humid trade wind flow by the complex topography causes large spatial variations in local weather. The analysis of atmospheric circulations at local scale has been conducted with the aim of improving the understanding of small-scale transport and dispersion of pollutants emitted from local sources. This study has two applications:• It takes part in the perspective of the new atmospheric research station of Piton Maïdo, a summit at 2200 m above sea level on the west coast of the island, in the frame of the developing Atmospheric Physics Observatory of La Reunion (Observatoire de Physique de l’Atmosphère de la Réunion, OPAR as French acronym). The objective is to examine to what extent local orographic perturbations of the synoptic wind and local wind systems are responsible for vertical export of air pollutants originating from the island boundary layer, and could affect ground-based measurements at the future Piton Maïdo observatory devoted to the monitoring of background atmospheric composition (greenhouse gases and aerosols).• The second one aims to investigate the behavior of the volcanic plumes from the Piton de la Fournaise. During the April-May 2007 eruption, large amounts of volcanic gases, particles and ash were released into the atmosphere causing air-pollution at several inhabited locations. In this work, the three-dimensional, non-hydrostatic, limited-area numerical model, MESONH, was run in high resolution to simulate features of airflows over Reunion Island in idealized and real conditions. Various passive tracers have been released in the model in order to characterize dynamical processes in the lower atmospheric layers and the related transport of air-masses.

Doppler radar wind profilers are already widely used to measure atmospheric winds throughout the free troposphere and stratosphere. Several methods have been developed to quantify atmospheric turbulence with such radars, but to date they have remained largely un-tested; this thesis presents the first comprehensive validation of one such method. Conventional in-situ measurements of turbulence have been concentrated in the surface layer, with some aircraft and balloon platforms measuring at higher altitudes on a case study basis. Radars offer the opportunity to measure turbulence near continuously, and at a range of altitudes, to provide the first long term observations of atmospheric turbulence above the surface layer. Two radars were used in this study, a Mesosphere-Stratosphere-Troposphere (MST) radar, at Capel Dewi, West Wales, and the Facility for Ground Based Atmospheric Measurements (FGAM) mobile boundary layer profiler. In-situ measurements were made using aircraft and tethered-balloon borne turbulence probes. The spectral width method was chosen for detailed testing, which uses the width of a radar's Doppler spectrum as a measure of atmospheric velocity variance. Broader Doppler spectra indicate stronger turbulence. To obtain Gaussian Doppler spectra (a requirement of the spectral width method), combination of between five and seven consecutive spectra was required. Individual MST spectra were particularly non-Gaussian, because of the sparse nature of turbulence at its observation altitudes. The width of Gaussian fits to the Doppler spectrum were compared to those from the `raw' spectrum, to ensure that non-atmospheric signals were not measured. Corrections for non-turbulent broadening, such as beam broadening, and signal processing, were investigated. Shear broadening was found to be small, and the errors in its calculation large, so no corrections for wind shear were applied. Beam broadening was found to be the dominant broadening contribution, and also contributed the largest uncertainty to spectral widths. Corrected spectral widths were found to correlate with aircraft measurements for both radars. Observing spectral widths over time periods of 40 and 60 minutes for the boundary layer profiler and MST radar respectively, gave the best measure of turbulence intensity and variability. Median spectral widths gave the best average over that period, with two-sigma limits (where sigma is the standard deviation of spectral widths) giving the best representation of the variability in turbulence. Turbulent kinetic energies were derived from spectral widths; typical boundary layer values were 0.13 m 2.s (-2) with a two-sigma range of 0.04-0.25 m 2.s (-2), and peaked at 0.21 m 2.s (-2) with a two-sigma range of 0.08-0.61 m 2.s (-2). Turbulent kinetic energy dissipation rates were also calculated from spectral widths, requiring radiosonde measurements of atmospheric stability. Dissipation rates compared well width aircraft measurements, reaching peaks of 1x10 (-3) m 2.s (-3) within 200 m of the ground, and decreasing to 1-2x10 (-5) m 2.s (-3) near the boundary layer capping inversion. Typical boundary layer values were between 1-3x10 (-4) m 2.s (-3). Those values are in close agreement with dissipation rates from previous studies.