Climate Change and Atmospheric Factors in Predicting Fire Season Severity

Abstract

Weather systems and climate drivers affect temperature and precipitation and are clearly associated with regional fire regimes; however, these links are poorly understood in many areas. In this thesis the relationships between fire season severity in the period 1950-2017 in seven locations in southern Australia, and synoptic-scale atmospheric variables and large-scale climate drivers are investigated. These relationships are found to be spatially coherent, with south-eastern and south-western Australia generally being affected by similar weather systems and climate drivers. Using the random forest technique – one not often used in climate science – the identified relationships are used to develop predictive tests for fire season severity which can be employed two to six months prior to the fire season. These predictive tests differ between the earliest years of the dataset and the most recent. This suggests changing influences of climate variables on fire season severity in each location. These changes are consistent with the expected effects of climate change, such as the intensification of climate drivers and of the subtropical ridge in the Australian region. This is the first study to produce a reliable tool for fire danger prediction in southern Australia with seasonal lead times. Such a tool could prove useful for fire agencies in planning and preparing for fire seasons in the future.