Degree Name

Doctor of Philosophy


School of Chemistry


The work presented in this thesis aims to improve our understanding of Australian vegetation fire emissions through measurements of smoke and the evaluation of biomass burning emission inventories. In particular, this thesis presents three studies.

The first study was aimed at investigating the potential impact that the amount of coarse woody debris would have on emission factors of trace gases in vegetation fires typical of south eastern Australian forests. As part of The Smoke Emissions and Transport Project commissioned by the Department of Environment, Land, Water and Planning, Victoria, a series of experiments were conducted at the CSIRO Pyrotron, a wind tunnel designed to study vegetation fires. Four loads involving a constant fine fuel load (10 t ha-1) and increasing coarse fuel loads (0 t ha-1, 2 t ha-1, 6 t ha-1 and 12 t ha-1), were set ablaze while an open-path Fourier Transform InfraRed (FTIR) spectrometer and two off-axis Integrated-Cavity-Output Spectroscopy (ICOS) instruments sampled trace gases in the smoke plume. The combustion was separated into three stages during the analysis: propagation, aming and smouldering. Sampling geometry was found to lead to significant divergence in the retrieval of species mixing ratios. No substantially significant trend was found between emission factors and the amount of coarse fuel. The three stages of the burns were found to have significantly different combustion efficiencies, with modified combustion effciency (MCE) values decreasing from 0.98 ± 0.01, during propagation, to 0.82 ± 0.05 during smouldering. This trend was also found to have a significant effect on emission factors, with clear linear trends between emission factors and MCE values. Stage dependent emission factors from this work have been incorporated into the CSIRO forecast modelling system to improve the prediction of air quality impacts from vegetation fires.

The second study was related to delivering new emission factors for trace gases and particles from Australian savanna fires. The 2014 Savanna Fires in the Early Dry Season campaign, the largest of its kind in Australia, was a 4 week study at the Australian Tropical Atmospheric Research Station (12.25° S, 131.05° E), with the objective of studying emissions from savanna fires. It brought together seven organisations and a large suite of instruments to provide updated emission factors for CO2, CO, CH4 and N2O, the first Australian savanna emission factors for Hg and sub-micron speciated aerosols, as well as the world's first emission factors for Aitken and accumulation aerosol modes. Those values were extracted from eight smoke events which impacted the station through the four week period. Desservettaz et al. [2017] summarised the results from this study.

The third and final study endeavoured to evaluate biomass burning inventories within global chemical transport models for the Australian region. GEOS-Chem was run for a period of three years, from 2008 to 2010, with three separate biomass burning emission inventories: FINN1.5, QFED2.4 and GFED4s; and the model ACCESS-UKCA was run by CSIRO with GFED4s for an extended period including those three years. These four simulations were compared to a range of observations including surface mixing ratios, ground-based FTS total columns and satellite-based MOPITT total columns at five sites within three regions. CO, C2H6 and CH2O were the species studied because of their different atmospheric lifetimes. A previously documented high bias in CO levels was found in both models, although this was slightly higher in GEOS-Chem. The three inventories were found to have up to an order of magnitude difference in their estimates of Australian biomass burning emissions, with FINN lowest and QFED highest. Bias aside, GEOS-Chem with GFED and QFED led to better correlation against measurements than either GEOS-Chem with FINN or ACCESS-UKCA with GFED. This study led to a recommendation to improve FINN emission estimates over North Australia. It also demonstrated that GFED performs the best over the Australasian region and should therefore be used for this region. Finally, this study highlighted that the current observational network is ill-suited to constrain Australian biomass burning emissions.

Together these three studies have improved the available emission factors for Australian temperate forest and savanna fires, and have identified areas where improved characterisation of Australian vegetation fires emissions is needed in global inventories.



Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.