The 2013 Forcett-Dunalley fire : a geospatial analysis of fire severity, pyrocumulonimbus dynamics, and smoke emissions

Extreme fires have become an issue of great concern due to their effects on society and terrestrial ecosystems. Such events have been attributed to anthropogenic climate change that is linked to an increase in the occurrence of dangerous fire weather conditions globally. In Australia, elevated fire weather has resulted in changes in fire regimes manifested by an increase in the severity, frequency and the extent of wildfires, particularly in temperate Eucalyptus landscapes. This has resulted in an increased risk of pyrocumulonimbus (pyroCb; fire thunderstorm) events; a change in the structure and composition of vegetation communities to more flammable vegetation classes; possible atmospheric effects from radiative forcing associated with smoke and greenhouse gas emissions; changes in habitat quality and biodiversity; and an increased risk of geomorphological processes such as landslides. This thesis uses the extreme 2013 Forcett–Dunalley fire that occurred in south-eastern Tasmania as a case study to understand various aspects of an extreme fire, including the coupling of fire behaviour with the atmosphere. In Chapter 2, I focus on the assessment of fire severity, an attribute of a fire regime that has received less research attention than burnt area mapping and fire frequency. I use Remote Sensing techniques to map the geographic patterns of fire severity following the fire. Specifically, the reliability of differential Normalised Burn Ratio (dNBR), a fire severity mapping technique routinely used in North America, is tested in Eucalyptus landscapes by validating different thresholds of dNBR with mapping from the field-validated fine-scale aerial photography. This study also investigates the effect of environmental factors such as vegetation, topography and fire weather on the variability of fire severity. The findings show that the site-specific dNBR had a higher congruence with aerial photography (45%). This shows the potential of using dNBR in Eucalyptus forests, although there is need for local calibration of dNBR through ground and aerial photography assessments for the index to be more reliable. The two highest severity classes contributed 47–55% of the total fire area (25,950- ha, including unburnt patches), and the short period of extreme fire weather (2% of the total fire duration) led to burning of 46% of the total fire area, yielding modelled extreme fireline intensities that reached 68,000 kW m\(^{−1}\) . Generalized linear modelling revealed that fire severity was strongly influenced by slope angle, aspect, and interactions between vegetation type and fire weather. Under elevated fire weather conditions, the long-unburnt dry Eucalyptus forest burnt at high severities due to high fuel loads and high vertical and horizontal fuel continuities, resulting in prolific crowning. To understand extreme fire behaviour associated with the fire, I describe in Chapter 3, the evolution of a fire thunderstorm (pyroCb) that developed on the afternoon of 4 January 2013. Weather radar data are used to relate storm development to the temporal progression of near-surface xiv and lower atmospheric fire weather indices (C-Haines index and McArthur forest fire danger index (FFDI), respectively) and to the progression of fire severity derived from Chapter 2. I contextualise this event with spatial patterns of elevated fire weather for Tasmania and compare its fire weather environment with fire weather of previous large Tasmanian fires during the period 2007–2016. The geospatial analysis shows that the pyroCb rapidly developed over a 24 min period from around 15:24 local time to reach a cloud top height of 15 km in the lower stratosphere. That period was characterised by elevated fire weather in the lower atmosphere (C-Haines value of 10–11) and at the near-surface (FFDI 60–75), and total crown defoliation of the forest beneath the pyroCb. Findings also show that fire weather conditions in Forcett–Dunalley were extreme relative to previous Tasmanian fires; and that eastern and south-eastern Tasmania are prone to the conjunction of elevated values of both indices, thus an increased risk of pyroCb development. In Chapter 4, I investigate the spatial patterns of smoke emissions (CO\(_2\) and PM\(_{2.5}\)) from the fire to represent emissions from Australian temperate Eucalyptus forests, which have received less attention than Australian tropical savannas and North American temperate forests. The adopted methodology is based on a basic model that incorporates local fuel and fire severity attributes in emissions estimation. I compare the results with a global model (Global Fire Emissions Database; GFED) to determine the reliability of the global model in emissions analysis in the event of the absence of site-specific data. Results from the basic model show that 1.125 ± 0.232 Tg of CO\(_2\) and 0.022 ± 0.006 Tg of PM\(_{2.5}\) were emitted into the atmosphere. Both the fine scale and GFED inventories produced comparable estimates for CO\(_2\), although PM\(_{2.5}\) estimates were lower by a factor of three compared to the fine scale inventory. GFED was able to produce reliable emissions estimates within the limits of emissions uncertainties, although the model did not accurately capture the spatial distribution of the two emissions. Based on this analysis, I discuss the deficiencies of current emission approaches, and identify data required to improve characterisation of smoke emissions from Australian temperate Eucalyptus forests. I show that fuel attributes, especially the amount of coarse woody fuels within a forest stand and the fraction of fuel consumed, contributed the most to uncertainties in emissions estimates. To improve emission from these ecosystems, more field inventories on these attributes and an upward revision of emissions factors for PM\(_{2.5}\) (in GFED model) are suggested. In conclusion, this thesis has elucidated the various aspects of an extreme fire in a temperate Eucalyptus ecosystem and it is the first ever study to publish mapped fire severity patterns for the island state of Tasmania. I show that dNBR has a potential application in Eucalyptus forests, although there is a need for validating the index for it to be more reliable in these ecosystems. This study has also highlighted the vulnerability of south-eastern Tasmania to extreme fire events, which has implications for fire weather forecasting and fire management in Australia and globally. Although the GFED model produced reliable CO\(_2\) emissions estimates, I have provided recommendations on further improving emissions estimations in Eucalyptus landscapes, including more detailed field assessments of coarse woody fuels in the forests and measurement of fuel consumption following fires. This will provide an important framework for modelling of air quality and climate dynamic