Abstract
3-D cloud-resolving model simulations including explicit aerosol physics and chemistry are compared with observations of upper tropospheric (12 km) aerosol size distributions over the Amazon Basin. The model underestimates the aerosol number concentration for all modes, especially the nucleation mode (d < 18 nm). We show that a boundary layer SO2 mixing ratio of approximately 5 ppb would be needed in order to reproduce the high nucleation mode number concentrations observed. This high SO2 mixing ratio is very unlikely for the pristine Amazon Basin at this time of the year. Hence, it is suggested that vapours other than H2SO4 participate in the formation and growth of small aerosols. Using activation nucleation theory together with a small (0.4-10%) secondary organic aerosol mass yield, we show that isoprene has the potential of substantially increasing the number of small particles formed as well as reducing the underestimate for the larger aerosol modes.