Fire in arid and semi-arid Australia: 1998-2004.

Abstract

Fire is a crucial element in shaping our world, whether caused naturally by lightning or by humans, either accidentally or on purpose. These fires can have both positive and negative consequences and impacts on our natural environment, human health, society and its economics, and global climate through carbon emissions. In arid and semi-arid Australia (70% of the continent), individual fires frequently exceed 1 million hectares, and have collectively burnt up to 9% of this total area in a single year. Stakeholders all have different outlooks and priorities about these phenomena. Little objective information about the fire regime and its drivers has been available for this vast area with its very low population density, with previous analyses limited in spatial and/or temporal extent. This lack of knowledge has hampered attempts at effective management. Satellite imagery enables active fires to be detected as fire hotspots, and burnt areas to be mapped as patches from the change of surface reflectance properties in successive images. The National Oceanographic and Atmospheric Administration’s (NOAA) Advanced Very High Resolution Radiometer (AVHRR) imagery has recently been used to map both fire hotspots (FHS) and fire affected areas (FAA) for the entire Australian continent dating back to 1998. In this dissertation, validation of these two datasets is performed in arid and semi-arid Australia for the first time. Results show that mapping accuracy can be highly variable between areas, and from year to year within the same area. There are many factors which may contribute to the errors of omission and commission. Some are common to all remote sensing, while other issues are more specifically related to conditions in arid and semi-arid environments. The distribution, seasonality, frequency, number and extent of fire hotspots (FHS) and fire affected areas (FAA) across arid and semi-arid country of Australia were analysed from 1998 to 2004, giving us a picture of fire patterns across the entire area for the first time. This includes a number of high fire years in certain areas following above-average rainfall. This analysis highlights similarities and differences between regions, giving policy makers and managers a basis from which to make more informed decisions in the present, and with which to compare future regimes. Exploratory regression analysis helps to gain a predictive understanding of the spatial and temporal pattern of risk of large uncontrollable fires. Results show that the strongest influence is exerted by biomass or fuel load. As this is highly dependant on antecedent rainfall, we can anticipate a strong effect of climate change on the fire regime. The strongest combinations of relationships may be used as spatial indicators in the development of long-lead fire risk models for these areas. This can help improve the timing of pro-active strategies to manage fire, and in the allocation of sparse funds and resources. This analysis has highlighted regional patterns of fire across different land tenures. Heightened awareness of these patterns may encourage a more cooperative and coordinated approach to fire management amongst stakeholders.