Evolution of southeast Australian fire activity seasons

Abstract

It is widely reported that climate change is impacting on wildfire risks in Australia. Fire seasons are seen to be starting earlier and finishing later. Climate modelling produces equivalent changes in fire weather. However national-scale datasets on fire activity are not widely available. A number of studies have used algorithm-generated "hotspots" from the pair of MODIS sensors carried on the TERRA and AQUA satellites for over 17 years. This dataset includes all types of fire-whether planned or not, and does not pickup low intensity fire activity. It is also sensitive to cloud cover. Nevertheless, the MODIS hotspot dataset, covering 4.5 million records since July 2002, provides a coverage that is broadly complete and consistent within those constraints. Hotspots can be classified according to geographic location and time of year. McRae & Featherston (2015) used a technique of producing "wind roses" to show fire activity seasonality, showing relative occurrences by month. To do this hotspots were aggregated to one degree grid-cells. In this paper hotspot "wind roses" are generated though aggregation to half-degree grid-cells over southeast Australia. To detect changes in fire activity patterns, the dataset is split into two epochs: Epoch 1 spans from July 2002 to June 2012, and Epoch 2 from July 2012 to June 2019. Impacts of climate change may be assessed through changes in key parameters of the pattern between the epochs. A number of grid-cells lack sufficient hotspots to form clear patterns. Others have fire activity through much of the year, and a seasonality pattern cannot be described. Change detection requires a defined pattern for both epochs.