1Max Planck Institute for Chemistry, Department of Biogeochemistry, Mainz, Germany
2Institute of Physics, University of São Paulo, São Paulo, Brazil
3Max Planck Institute for Biogeochemistry, Jena, Germany
4Atmospheric Sciences Department, IAG, University of São Paulo, São Paulo, Brazil
5Divisão de Química Ambiental, Laboratório de Química Atmosférica, IPEN, São Paulo, Brazil
6Center for Advanced Studies in Arid Zones (CEAZA), La Serena, Chile
Abstract. As part of the LBA-SMOCC (Large-Scale Biosphere-Atmosphere Experiment in Amazonia – Smoke, Aerosols, Clouds, Rainfall, and Climate) 2002 campaign, we studied the emission of carbon monoxide (CO), carbon dioxide (CO2), and aerosol particles from Amazonian deforestation fires using an instrumented aircraft. Emission ratios for aerosol number (CN) relative to CO (ERCN/CO) fell in the range 14–32 cm-3 ppb-1 for most of the time, in agreement with values usually found from tropical savanna fires. The number of particles emitted per amount biomass burned was found to be dependant on the fire condition (combustion efficiency). Variability in the ERCN/CO between fires was similar to the variability caused by variations in combustion behavior within each individual fire. This was confirmed by observations of CO-to-CO2 emission ratios (ERCO/CO2), which stretched across the same wide range of values for individual fires as for all the fires observed during the sampling campaign, indicating that flaming and smoldering phases are present simultaneously in deforestation fires. Emission factors (EF) for CO and aerosol particles were computed and a correction was applied for the residual smoldering combustion (RSC) fraction of emissions that are not sampled by the aircraft. The correction, previously unpublished for tropical deforestation fires, suggested an EF about one and a half to twice as large for these species. Vertical transport of biomass-burning plumes from the boundary layer (BL) to the cloud detrainment layer (CDL) and the free troposphere (FT) was found to be a very common phenomenon. We observed a 20% loss in particle number as a result of this vertical transport and subsequent cloud processing, attributable to in-cloud coagulation. This small loss fraction suggests that this mode of transport is very efficient in terms of particle numbers and occurs mostly via non-precipitating clouds. The detrained aerosol particles released in the CDL and FT were larger due to coagulation and secondary growth, and therefore more efficient at scattering radiation and nucleating cloud droplets than the fresh particles. This process may have significant atmospheric implications on a regional and larger scale.