6533b7d6fe1ef96bd1265920

RESEARCH PRODUCT

Prévision et spatialisation des concentrations en ozone troposphérique en Bourgogne

subject

description

This PhD done in the Centre de Recherche de Climatologie, UMR 5210 CNRS of University of Burgundy has been supported by the Conseil Régional de Bourgogne, and the AASQA of Burgundy (Atmosf'Air). The ozone, a poison gas, is the core of this research because it is considered as one of the major worrying atmospheric pollutants in this region. We have conducted a study focused on hourly concentrations of measured ozone over several years for 12 stations. Their spatial and temporal variability are related to some features outside of the region (ground-level ozone, synoptic-scale conditions) and inside of the region (physical properties, land cover, meteorological features and precursor potential emissivity). The second part presents the working out of a statistico-dynamic method for a 1-day forecast of the ozone maximum concentration based on five stations from Atmosf'Air in Dijon urban area, two in Chalon-sur-Saône one, two in Mâcon, one in Montceau-les-Mines, and one rural station located in the Morvan (Saint-Brisson). The forecast models built in collaboration with Météo-France integrated meteorological forecast coming from the Aladin model. The chemistry is also taken into account in multiple regression models with the integration of Nitrogen Oxides, and the amount of observed ozone measured the day when the forecast is done. The forecast mean error reaches around 20 μg/m3. The ozone peaks are correctly forecasted while using a correction factor that increases the forecast “dynamic”. The residual errors are analysed and have pointed out that models slightly reduce the observed concentration variability when abrupt changes of weather type occur. Isolated errors are linked to meteorological forecast errors or unexpected concentrations of precursors. These forecasts are thus outstanding, and have a more accurate spatial resolution than those from determinist models. The benefit of this kind of model is important at a local scale to give information and to warn the concerned population. Finally, a ground measurements campaign (104 sample points using radial diffusive samplers) was done over the entire Burgundy during August 2000. Using the results of this campaign, we tried to better understand the factors of the ozone spatial distribution, and then the concerned mechanisms. A downscaling approach based on specific analysis and spatial statistics methods has been developed First, some indexes of the surface properties have been created from several variables (Digital Elevation Model, land cover, modelled atmospheric variables) using a GIS. Then, the ozone spatial pattern has been investigated based on a variogram and crossvariogram. We have explained these behaviours from the atmosphere states and some surface indexes. The results highlighted a strong anisotropy of ozone spatial pattern, and two types of pattern in accordance with two specific meteorological patterns. The first type is associated to pattern of bad conditions to form the ozone where the ozone spatial pattern is strongly related to surface rugosity at a regional scale. The second type that gives good conditions to form ozone is dominated by a surface “biological” forging via the combination of surface rugosity and surface land cover. Thus, a strong correlation between ozone concentration at a point and mixed forest ratio above that point has been highlighted. Globally, our results pointed out the complexity of the associated mechanisms and the ozone spatial pattern forced by the surface in Burgundy.