For this study, an aerosol simulation was performed with the MOCAGE global CTM [Teyssèdre et al., 2007] (CNRM/GAME/Météo-France). MOCAGE has been designed for a range of applications, from regional air quality studies to global climate studies dealing with the evolution of tropospheric and stratospheric chemical species. The version used in this study was adapted to represent trace gases and aerosols at the global scale and is briefly presented in section 2.1; it is described in detail by Ménégoz et al. .
2.1. General Features of the CTM
 In our simulation, MOCAGE is used on a T42 Gaussian grid (about 2.8° × 2.8° horizontal resolution) and with 47 vertical sigma pressure layers from the surface to 5 hPa. Seven levels are within the planetary boundary layer, 20 in the free troposphere, and 20 in the stratosphere. The first layer is ∼40 m thick, while the resolution above 300 hPa is constant with altitude, around 800 m. A semi-Lagrangian scheme is used for the advection of tracers and chemical compounds. Based on the work of Williamson and Rasch , it is described in detail by Josse et al. . Time steps are 1 h for advection and 15 min for subgrid-scale processes. The turbulent diffusion follows Louis , whereas the convection scheme (mass-flux-type) is that of Bechtold et al. . The representation of dry deposition for gases, based on the work of Wesely , was presented by Michou and Peuch . In-cloud and below-cloud scavenging representation for gases was presented by Teyssèdre et al. .
2.2. Aerosol Representation in the Model
 MOCAGE can simulate the evolution of three types of aerosols: sulfate, BC, and dust. Organic carbon (OC) and sea-salt aerosols are not yet implemented in the model. BC and dust are emitted directly into the atmosphere, whereas sulfate is produced by chemical reactions involving precursor gases. Some of these gases are natural (dimethylsulfide [DMS] emitted by the ocean), and others are anthropogenic (SO2 and H2S). Concentrations of the oxidant (OH, H2O2, O3, and NO3) are prescribed and provided by a 1 year MOCAGE simulation with the full chemical scheme described by Teyssèdre et al. . Representation of the sulfur cycle, based on the study by Pham et al. , was described by Ménégoz et al. . Both aqueous and gaseous phase reactions produce sulfate. The parameterization of the dry deposition was based on the work of Seinfeld and Pandis ; its implementation in MOCAGE was presented by Nho-Kim et al. . The sedimentation velocity, negligible for BC and sulfate aerosols, is essential for the representation of large dust aerosols. Its parameterization was adapted from Stokes law [Seinfeld and Pandis, 2006]. Below-cloud scavenging depends on the collision efficiency between aerosols and cloud droplets, as computed by Seinfeld and Pandis . In-cloud scavenging is simulated according to the scheme of Langner and Rodhe . BC and sulfate transfer efficiencies from solid to aqueous phase are adjusted from the observations of Kasper-Giebl , as presented by Ménégoz et al. . This parameter is considered to be identical for mineral dust and BC. For each type of aerosol, the total distribution can be represented by several modes, all of which have a lognormal distribution. Diameter, standard deviation, and fraction number of the different modes for each aerosol are presented in Table 1. Distributions are discretized into bins of different sizes in MOCAGE, as described by Martet et al. . The last column of Table 1 shows the different bins used for each aerosol. In MOCAGE simulations, the chemical production of sulfate and the direct emissions of BC and mineral dust are injected into the atmosphere according to the distribution shown in this Table 1.
Table 1. Diameters, Standard Deviation, and Number Fraction of Lognormal Distribution for Mineral Dust, Black Carbon, and Sulfate
|Aerosol||Distributiona||Numbers and Sizes of Bins Used in the Modelb (m)|
|Diameter (μm)||Standard Deviation||Number Fraction|
|Mineral dust modes||0.22||1.59||0.38||5 bins (1.00E-8 to 6.31E-8; 6.31E-8 to 3.98E-7; 3.98E-7 to 2.51E-6; 2.51E-6 to 1.58E-5; 1.58E-5 to 1.00E-4)|
|Black carbon modes||0.015||1.8||0.92||4 bins (1E-9 to 1E-8; 1E-8 to 1E-7; 1E-7 to 1E-6; 1E-6 to 1E-5)|
|Sulfate modes||0.015||1.8||0.98331||4 bins (1E-9 to 1E-8/1E-8 to 1E-7/1E-7 to 1E-6/1E-6 to 1E-5)|
2.3. The 2000–2005 Simulation
 A simulation of the years 2000–2005 was computed for the aim of this study. In this simulation, air temperature, humidity, pressure, and wind components used to drive MOCAGE were provided from 6 h analyses obtained by the ECMWF IFS model. For the 6 years of the simulation, we used the AEROCOM global emissions inventory representative of the year 2000 [Dentener et al., 2006]. This inventory is one of the commonly used inventories of aerosols and precursor gases to perform aerosol simulations. Emissions of SO2, H2S, SO42−, and BC are constant over the year, except for biomass burning emissions, which have monthly variations. The AEROCOM inventory considers daily variations of DMS and dust emissions. However, we used monthly averages for these fields because we assume that daily variations of these emissions are very different from one year to another, and it would not make sense to take them into account in a 6 year simulation. We assumed that 2.5% of the anthropogenic elementary sulfur is directly emitted as SO42−, the rest being SO2.
2.4. Validation of the Simulation
 Before studying our aerosols simulation over the NAE region, it is necessary to evaluate the ability of the model to describe the main sources and sinks for each type of aerosol. Table 2 shows global aerosol burdens, sinks, and sources computed in 2000 by MOCAGE and the multimodel mean from the AEROCOM intercomparison exercise. Aerosol burdens simulated by MOCAGE are in the range of the burdens simulated by AEROCOM models. The MOCAGE BC burden is equal to the mean of that in the AEROCOM models. Concerning this aerosol, dry deposition is more efficient in MOCAGE than in all other models, but it is compensated by wet deposition, which is lower in MOCAGE than in other AEROCOM models. Nho-Kim et al.  validated MOCAGE simulations of BC, comparing simulations with surface observations. Dust burden is lower in MOCAGE than in AEROCOM models, presumably the result of the MOCAGE dry deposition and sedimentation fluxes, which are stronger than those simulated by AEROCOM models, on average. However, Martet et al.  validated the MOCAGE representation of mineral dust by comparing simulations with satellite data (Moderate Resolution Imaging Spectroradiometer; MODIS), light detecting and ranging data (AErosol RObotic NETwork), and station measurements (Interagency Monitoring of Protected Visual Environments network). Sulfate burden simulated by MOCAGE is similar to the higher estimate from the AEROCOM intercomparison. Dry deposition of this aerosol is of the same order of magnitude in MOCAGE as in AEROCOM models, whereas sulfate chemistry production simulated by MOCAGE is lower than in other AEROCOM models. High sulfate burden simulated by MOCAGE may be due to the wet deposition, which is less efficient in MOCAGE than in the other models, on average. Ménégoz et al.  compared a MOCAGE simulation of sulfate with observations of the European Monitoring and Evaluation Programme [Hjellbrekke, 2004) and confirmed that MOCAGE tends to underestimate scavenging fluxes of sulfate. Nevertheless, this study shows that despite a slight tendency to overestimate sulfate concentration, MOCAGE describes rather well the distribution of sulfate over Europe.
Table 2. Global Burden, Sinks, and Sources by MOCAGE and AEROCOM Modelsa
| ||Sulfate||Black Carbon||Mineral Dust|
|Burden||1.15||0.7 (0.3–1.2)||0.2||0.2 (0.11–0.37)||13.2||21.3 (6–30)|
|Emission + chem. prod. (sulfate only)||44.0||54 (30–80)||7.8||7.8 (-)||1670||1670 (-)|
|Dry deposition + sedimentation||6.6||6.8 (1–13)||3.2||1.6 (0.2–2.4)||1254||1120 (700–2100)|
|Wet deposition||37.2||47 (28–116)||4.5||6.2 (5.3–11)||427||498 (100–750)|
2.5. Aerosol Budget Over the North Atlantic Region
 Figure 1 shows the main sinks and sources over the North Atlantic region for sulfate, BC, and mineral dust. Sulfate originates essentially from SO2 oxidation, by an aqueous chemistry pathway, whereas gaseous chemistry pathways and direct emissions represent only 6% and 5% of total sulfate sources, respectively. Considering the north Atlantic region, the main sink for sulfate is wet deposition (0.28 mg[S] m−2 d−1), followed by dry deposition (0.11 mg[S] m−2 d−1) and transport outside of the domain (0.07 mg[S] m−2 d−1). Sedimentation is negligible for sulfate, because of the very small size of this aerosol. The mean sulfate burden is 4.59 mg[S] m−2 in this region. The mean BC burden is 0.38 mg m−2. The BC emissions, issued from biomass burning and human activities in Europe, Africa, and America, average 0.067 mg m−2 d−1 over the entire North Atlantic region. The main sink for this aerosol is dry deposition (0.022 mg m−2 d−1), followed by wet deposition (0.014 mg m−2 d−1) and export out of the domain (0.0017 mg m−2 d−1). Sedimentation is negligible for BC, because of the small size of this aerosol. The mineral dust burden is 68.4 mg m−2 over the NAE region, with emissions coming from African and Middle East deserts equal to 115.9 mg m−2 d−1, on average, over the whole domain. The main sinks for this aerosol are dry deposition (23.22 mg m−2 d−1), followed by transport outside of the domain (8.47 mg m−2 day−1), sedimentation (5.10 mg m−2 d−1), and wet deposition (2.40 mg m−2 d−1). The sum of all fluxes is equal to 0.026 mg[S] m−2 d−1 for sulfate, 0.029 mg m−2 d−1 for BC, and 76.7 mg m−2 d−1 for mineral dust. It is positive for all aerosols, indicating an accumulation of aerosols in the NAE region during the winter.
Figure 1. Sulfate, black carbon, and mineral dust budget over the North Atlantic region (20°N–80°N, 80°W–40°E) in winter (December–February) from 2000 to 2005. Burdens (top right corner of the square) are given in mg[S] m−2 for sulfate and in mg m−2 for the other aerosols. Fluxes are given in mg[S] m−2 d−1 for sulfate and in mg m−2 d−1 for the other aerosols. Tr., transport outside of the domain; Wet, wet deposition; Dry, dry deposition; Sed., sedimentation; Aq. chem., aqueous chemistry; Gas. chem., gaseous chemistry; Em., emissions.
Download figure to PowerPoint
 Sulfate is essentially concentrated over Europe and America in winter, because of high SO2 anthropogenic emissions (Figure 2a), the average burden varying between 5 and 9 mg[S] m−2. Over the North Atlantic Ocean, the sulfate burden varies from 0 mg[S] m−2 in the tropics to 5 mg[S] m−2 in the north. Over Greenland, sulfate column load (i.e., burden) is very low (i.e., about 2 mg[S] m−2). BC has a distribution similar to this of sulfate, with an averaged column load reaching 0.8 mg m−2 over Europe and America, 0.4 mg m−2 above the Atlantic, and 0.2 mg m−2 over Greenland (Figure 2b). There is, however, a large difference between BC and sulfate distributions over Africa: the sulfate burden is very low over this region, whereas the BC burden is comparable to that modeled in European polluted areas, because of strong biomass burning emissions in central Africa. Mineral dust is essentially concentrated over northern Africa, attributable to Saharan and Middle East desert emissions (Figure 2c). The winter average of dust column load varies between 200 and 500 mg m−2 above Africa, between 10 and 25 mg m−2 over Europe, and between 0 and 10 mg m−2 above the Atlantic Ocean and America. Only over the eastern part of the Atlantic is there a high concentration of dust: dust column load ranges from 200 mg m−2 in the West Coast of Africa to 0 mg m−2 in the middle of the North Atlantic region.
Figure 2. Sulfate (mg[S] m−2), black carbon (mg m−2), dust (mg m−2), and aerosol optical thickness (AOT) winter average from 2000 to 2005 simulated by MOCAGE.
Download figure to PowerPoint
 The resulting aerosol radiative forcing is estimated by using aerosol optical thickness (AOT). AOT is computed here from the aerosol column load of sulfate, BC, and dust with the coefficients estimated by Tegen et al. . Figure 2d shows that simulated AOT takes very high values over Africa, reaching 0.6, because of the extreme concentration of mineral dust in the atmosphere. AOT ranges from 0.2 to 0.3 over Eastern Europe and from 0.1 to 0.2 over Western Europe, because of the presence of mineral dust, sulfate, and BC. Over America, AOT varies between 0.2 and 0.25, mainly because of the presence of sulfate and BC. AOT is between 0.1 and 0.15 over the Atlantic Ocean, except near the African western coast, where it can reach values of 0.4 because of the transport of desert dust. AOT takes its lower values over Greenland and over the Caribbean Sea, ranging from 0.05 to 0.01.
 In the study by Tegen et al. , AOT is also computed with a CTM. It is generally lower than those simulated in this study over the continents: it does not exceed 0.3 over Africa and 0.016 over Europe and America. The same difference is noticed over the ocean, with AOT varying between 0.08 in the northern Atlantic and 0.2 in the tropical Atlantic in the study by Tegen et al. . In our study, we do not consider organic and sea-salt aerosols since they are not yet implemented in our model. It is then surprising that the AOT values of Tegen et al.  are lower than ours, because they consider all the aerosols in their study. However, from comparisons with sun photometer measurements and satellite retrievals, Tegen et al.  indicated that the AOT modeled in their study was often underestimated, in particular over Africa and Europe.
 Remer et al.  evaluated AOT by analyzing MODIS satellite images. They evaluate a set of 5 year monthly means of AOT at a global scale. In that study, AOT in winter ranges between 0.1 and 0.3 over Europe, between 0.1 and 0.2 over the north of the Atlantic Ocean, except near the African western coast where it can reach 0.4, and from 0 to 0.15 over the American eastern coast. Theses values are of the same order of magnitude as those simulated by MOCAGE, which suggests that AOT distribution modeled by MOCAGE is quite realistic.