Dust Influences in the Eastern Mediterranean: Multi-scale Assimilation of MODIS AOD

Air pollution is one of the major environmental problems in the Mediterranean basin since the limit values of the pollutants are often exceeded. Considering its location, Turkey is downwind of Europe and on the crossroad of long-range dust transport and local emissions, meaning high amount of population living in Turkey (Population of Turkey is 75 million) are exposed to high particulate matter (PM) concentration. This study aims to quantify Saharan dust contribution to high PM concentration over Eastern Mediterranean, especially in Turkey via ground observations, atmospheric modeling and aerosol data assimilation. An online Weather Research and Forecasting model with chemistry (WRF-Chem) is configured with 0.1°×0.1° spatial resolution HTAP emission inventory to cover Europe in the west, Caspian Sea in the east, Scandinavia in the north and Sahara in the south with 30km horizontal resolution. NOAA Gridpoint Statistical Interpolation (GSI) data analysis system is used to assimilate Moderate Resolution Imaging Spectoradiometer (MODIS (collection 6)) aerosol optical depth (AOD) retrievals over the region for April 2008. Real-time Air Quality Modeling System (RAQMS) 2°×2° global analyses is used to provide the lateral boundary conditions (LBC) for 30km run (30km_Assim). The 30km run is then used to provide LBC for a second higher resolution (10km horizontal resolution covering Anatolian Peninsula) nested run to avoid the errors caused by the complex topography. 10km run is utilized with and without data assimilation (10km_Assim and 10km_NoAssim, respectively). This is the first WRF-Chem data assimilation study investigating natural dust influences on air quality in Anatolian peninsula. WRF-Chem prediction is consistent with the ground observations as the model captures the increasing and decreasing pattern. Daily comparisons of the WRF-Chem 30km_Assim, 10km_Assim and 10km_NoAssim runs to ground observations show that the 10km_Assim (-11 μg/m3) run has an overall low bias than the 30km_Assim (-5 μg/m3) and 10km_NoAssim (-10.8 μg/m3) runs, although the correlations between the runs and the ground observations are almost the same.