Bushfire_specific_PM25_Aus_2001_2020_v1_3

This dataset contains bushfire specific daily PM2.5 predictions for Australia from 2001 to 2020 based on a random forest model. For more information and access see http://cardat.github.io/data_inventory/bushfire_specific_pm25_aus_2001_2020_v1_3.html. This is version 1.3 of the CAR Bushfire Smoke Exposure project, superseding v1.2. Additionally, a Seasonal-Trend decomposition using LOESS (STL) decomposition was calculated for years 2001-2019, using a seasonal window of 45. Due to the abnormal levels of PM2.5 in January 2020 from severe bushfires, the year 2020 was not included in the STL calculation. However, the decomposition was extrapolated to 2020 using 2019 seasonal and trend values. Flags for classifying days of high PM2.5 (e.g. dust event, bushfire event) were taken from v1.2. As the v1.2 prediction grid was slightly smaller than that of v1.3, an extrapolation was performed on the flags to cover coastal pixels. The creator acknowledges the Sydney Informatics Hub and the University of Sydney’s high performance computing cluster Artemis for providing the high performance computing resources that have contributed to the production of this dataset.