A diagnose of the occurrence of the SALLJ events for the 2002 was performed based on the modified Bonner’s first criterion for the strength and vertical shear of the wind field (Bonner, 1968; Saulo et al., 2000), using the 6-hourly analysis of the Global Data Assimilation System (GDAS) of the National Centers for Environmental Prediction (NCEP). This data set has one-degree horizontal resolution and is available every synoptic time (0000, 0600, 1200 and 1800 UTC), at 26 vertical pressure levels.

The information about fire spots over South America is obtained with remote sensors and after processing, it is freely available at http://www. cptec. inpe. br. The observations of aerosols in Buenos Aires that could give information of the intrusion of the regional smoke plumes consist on columnar aerosol content and derived quantities obtained from measurements at the CEILAP-BA (34.5° S, 58° W) (Buenos Aires) site of the AErosol RObotic NETwork (AERONET) from National Atmospheric and Science Administration (NASA) (http://aeronet. gfsc. nasa. gov).

The on-line atmospheric transport model CATT-BRAMS (Coupled Aerosol and Tracer Transport model to the Brazilian developments on the Regional Atmospheric Modeling System) was used to simulate the atmospheric transport of biomass burning smoke during the dry season of 2002. A detailed description of the CATT-BRAMS system can be seen at Freitas et al., 2009; Longo et al., 2010). The system considers the emission, transport and transformations of particulate matter (PM2.5) and gases (CO) and it is run operatively at Centro de Previsao de Tempo e Estudos Climaticos (CPTEC) with 40 km resolution over South America. It provides 72-hour predictions of the above mentioned aerosols and gases as well as the meteorological fields. Two SALLJ events were selected to perform a more in depth analysis of the transport patterns and the aerosol dispersion. The synoptic environment in which they took place was studied and the resulting spatial and temporal distributions of aerosols obtained with the CATT-BRAMS modelling system for each case were analysed.

3. Results