A method of retrieving PM10 particles concentrations at the ground level from AOT (Aerosol Optical Thickness) measurements is presented. It uses data obtained among five years during 2003 to 2007 summers in the Lille ...A method of retrieving PM10 particles concentrations at the ground level from AOT (Aerosol Optical Thickness) measurements is presented. It uses data obtained among five years during 2003 to 2007 summers in the Lille region (northern France). As PM10 concentration strongly depends on meteorological variables, we clustered the meteorological situations provided by the MM5 meteorological model forced at the lateral boundaries by the operational NCEP model in eight classes (local weather types) for which a robust statistical relationship between AOT and PM10 was found. The meteorological situations were defined by the hourly vertical profiles of temperature and (zonal and meridian) wind components. The clustering of the weather types were obtained by a self-organizing map (SOM) followed by a hierarchical ascending classification (HAC). We were then able to retrieve the PM10 at the surface from the AERONET AOT measurements for each weather type by doing non linear regressions with dedicated SOMs. The method is general and could be extended to other regions. We analyzed the strong pollution event that occurred during August 2003 heat wave. Comparison of the results from our method with the output of the CHIMERE chemical-transport model showed the interest to tentatively combine these two pieces of information to improve particle pollution alert.展开更多
文摘A method of retrieving PM10 particles concentrations at the ground level from AOT (Aerosol Optical Thickness) measurements is presented. It uses data obtained among five years during 2003 to 2007 summers in the Lille region (northern France). As PM10 concentration strongly depends on meteorological variables, we clustered the meteorological situations provided by the MM5 meteorological model forced at the lateral boundaries by the operational NCEP model in eight classes (local weather types) for which a robust statistical relationship between AOT and PM10 was found. The meteorological situations were defined by the hourly vertical profiles of temperature and (zonal and meridian) wind components. The clustering of the weather types were obtained by a self-organizing map (SOM) followed by a hierarchical ascending classification (HAC). We were then able to retrieve the PM10 at the surface from the AERONET AOT measurements for each weather type by doing non linear regressions with dedicated SOMs. The method is general and could be extended to other regions. We analyzed the strong pollution event that occurred during August 2003 heat wave. Comparison of the results from our method with the output of the CHIMERE chemical-transport model showed the interest to tentatively combine these two pieces of information to improve particle pollution alert.
