2.1. Aerosol Satellite Products: POLDER and MODIS
 The Polarization and Directionality of the Earth Reflectance (POLDER) instrument is a spaceborne radiometer developed by the French space agency (Centre National d'Etudes Spatiales) [Deschamps et al., 1994], launched aboard the Advanced Earth Observation Satellite (ADEOS 1 and 2) in 1996 and 2003. The POLDER orbits have a local overpass time around 10:30 am and provide a quasi-global coverage every day, although clouds often prevent aerosol retrieval [Deschamps et al., 1994]. POLDER measures the polarization and directionality of the solar radiation reflected by the Earth, which allows the monitoring of aerosol characteristics, derived separately over land and ocean using independent algorithms. Both the total aerosol optical thickness (AOT) and the fine mode aerosol optical thickness (referred to as AOTf) are retrieved over the oceans [Deuzé et al., 2000]. POLDER can retrieve optical characteristics of particles whose radius is smaller than 0.5 μm (fine mode) over land [Deuzé et al., 2001] using its polarization capabilities, while larger particles generate low polarization and cannot be observed with this technique. The data set used in this study is from the POLDER-2 mission, which covered the period from April to October 2003, and consists of daily AOTf at 865 nm (AOTf_865) regridded in this study at the 2° × 2.5° model resolution. Initial validation of the POLDER products were presented by Goloub et al. , Deuzé et al.  and Deuzé et al. . Several improvements were applied to the algorithms, as described for instance by Herman et al. .
 The Moderate-resolution Imaging Spectroradiometer (MODIS) on board the Terra satellite and developed by the National Aeronautics Space Administration (NASA) has the same local overpass time than POLDER. MODIS provides multispectral measurements that are used to retrieve aerosol properties over land and ocean using separate algorithms. Retrievals over ocean include AOT (at seven wavelengths from 0.43 to 2.13 μm) and AOTf (radius smaller than 0.5 μm) [Tanré et al., 1997; Remer et al., 2002]. Over land, aerosol retrieval is possible over dark surfaces but not available over bright land (i.e., desert or snow/ice-covered surfaces) [Kaufman et al., 1997; Chu et al., 2002]. The MODIS data used in this study come from Terra Collection 005 and consist in daily AOTf at 550 nm (AOTf_550) globally gridded at 1° × 1° horizontal resolution, and regridded here at the 2° × 2.5° model resolution. Comparisons to Aerosol Robotic Network (AERONET) ground-based measurements show that, the MODIS AOT (τ) are accurate to within ±0.05 ± 0.15 τ over land and ±0.03 ± 0.05 τ over ocean [Remer et al., 2005].
 As remote sensing aerosol products can present high bias especially over land, we used two of the currently available products. A general good agreement is found between the POLDER and MODIS products used in this study [Gérard et al., 2005]. Discrepancies are found mostly in the presence of large nonspherical particles (e.g., dust), but this is not expected to be the case here since biomass burning aerosols are mostly in the fine mode. Note that a recent study by Smirnov et al.  indicates that satellite retrieved aerosol optical thickness (in this case MODIS products) are generally higher than ground-based measurements.
2.2. CO Satellite Products: MOPITT
 The Measurement Of air Pollution In The Troposphere (MOPITT) instrument is a thermal and near-infrared nadir-viewing gas correlation radiometer launched aboard the NASA Terra platform (December 1999). The instrument allows the retrieval of CO vertical profiles and provides a near-global coverage of the Earth within three days. MOPITT retrievals are derived from the maximum likelihood method [Rodgers, 2000] and is therefore a statistical combination of the measurements and a priori CO information [Deeter et al., 2003]. In this study we used the Level 2 Version 3 products, which consist in retrieved CO mixing ratio profiles and total column for all cloud-free scenes. The CO mixing ratio are reported at the seven pressure levels: surface, 850, 700, 500, 350, 250, 150 hPa. However, for a given vertical profile no more than 2 levels provide independent pieces of information, in general [Heald et al., 2003b; Deeter et al., 2004]. Emmons et al.  report error of 0.9 ± 10.4% at 500 hPa, 1.6 ± 10.1% at 350 hPa and −0.5 ± 12.1% for the data acquired after August 2001, although local biases could be somewhat larger. Further information about MOPITT retrieval method, vertical resolution and validation are given by Deeter et al. , Deeter et al.  and Emmons et al. . We restrict our analysis to the MOPITT data obtained with an a priori contribution of less than 40% at 500 hPa.
 We used the GEOS-Chem chemical and transport model (CTM) (http://www-as.harvard.edu/chemistry/trop/geos/), version v7-02-03, to conduct global three-dimensional simulations of coupled oxidant-aerosol chemistry for 2003 with a 2° × 2.5° horizontal resolution and 30 vertical levels. The model is driven by assimilated meteorology from the Goddard Earth Observing System (GEOS-4) of the NASA Global Modeling and Assimilation Office (GMAO), which includes winds, temperature, surface pressure, water content, clouds, precipitation, convective mass fluxes, mixed layer depth and surface properties with a 6-hour temporal resolution (3-hour for surface variables and mixing depths). The model simulates the tropospheric ozone-nitrogen oxides (NOx)-hydrocarbon chemistry [Bey et al., 2001a] as well as the tropospheric aerosol types including sea salts [Alexander et al., 2005], mineral dust [Zender et al., 2003], sulfate-nitrate-ammonium aerosols [Park et al., 2004], carbonaceous aerosols (including black and primary organic carbon further referred to as BC and OC, respectively) [Park et al., 2003], and Secondary Organic Aerosol (SOA) [Chung and Seinfeld, 2002]. Aerosol and oxidant chemistry simulations interact through sulfate, nitrate and SOA formation, heterogeneous reactions, and aerosol effects on photolysis rate [Martin et al., 2003].
 Biomass burning emissions for carbonaceous aerosols are derived from the Bond et al.  annual inventory. Climatological inventories of biomass burning trace gas emissions are described by Lobert et al.  and Duncan et al.  for CO, NOx, alkanes, acetone, and from Park et al.  and Park et al.  for sulphur dioxide and ammonia. These climatological inventories are redistributed in space and time according to the occurrence of open fires detected by the spaceborne Advanced Along Track Scanning radiometer (AATSR) [Arino and Melinotte, 1995] to account for seasonal and interannual variability. The method (including the specific biomass burning regions used) is described in detail by Generoso et al. , except that the fire count data set used in the present study was extended to account for the latest available satellite data (up to 2005). For 2003, our global estimates are 3.34 and 26.3 Tg C for BC and OC, respectively, and 418 Tg for CO.
 All aerosols experience dry deposition following the size-dependent scheme of Zhang et al.  for dust and sea salts and a resistance-in-time scheme for the other species as described by Balkanski et al. . Hydrophilic aerosols are in addition subject to wet deposition, which includes both scavenging in convective updrafts and rainout/washout from large-scale precipitation [Liu et al., 2001]. A fraction of eighty percent of SOA are assumed to experience wet deposition [Chung and Seinfeld, 2002]. A fraction of twenty and fifty percent of BC and OC, respectively, are emitted as soluble, the hydrophobic part being converted into hydrophilic fraction during particles ageing with an e-folding time of 1.15 days [Cooke et al., 1999; Chin et al., 2002].
 The aerosol optical thickness is calculated assuming externally mixed aerosol and lognormal size distributions and as a function of the local relative humidity. The aerosol optical properties and hygroscopic growth factors used in the model are described by Martin et al. . AOT865 and AOT550 are calculated online in the model using optical properties at 550 and 865 nm taken from the Global Aerosol Data Set (GADS) [Köpke et al., 1997; Patterson et al., 1977]. To compare model results with the observed fine aerosol optical thickness, we compute a simulated fine aerosol optical thickness which includes all carbonaceous aerosols (primary and secondary particles), sulfates and fine sea salts. The effective radius at seventy percent of relative humidity are 0.24, 0.04, 0.10, 1.3 μm for sulfate, black carbon, organic carbon, and fine sea salt, respectively [Martin et al., 2003]. The dust fine fraction is not accounted for in our simulated fine aerosol optical thickness, however we discuss its possible contribution in section 3. The simulated CO columns are compared to MOPITT after transformation by the MOPITT averaging kernel to describe the vertical sensitivity of the instrument to the true CO profiles. MOPITT averaging kernels typically show highest sensitivity in the middle and upper troposphere and low sensitivity in the boundary layer [Deeter et al., 2003].
 We conducted a standard simulation as described above starting in July 2002 and analyzed the outputs from May to August 2003. Anthropogenic and biomass burning emissions south of 45°N are mostly from biofuel burning and anthropogenic activities, while wildfire emissions dominate north of 45°N (Figure 1). The sensitivity simulations described in section 4 are performed for the period from April to August 2003, with initial conditions taken from the standard simulation.
Figure 1. BC emissions between May and August 2003 (top) from biomass burning and (bottom) from biofuel and anthropogenic activities.
Download figure to PowerPoint