OzCBI: the composite burn index adapted to assess fire severity and key fauna habitat features in Australian ecosystems

ABSTRACT Although fire management is essential to conserve ecosystems and protect communities in Australia, the best way to achieve these goals is controversial. Recent advances that use satellite imagery to map fire severity focus on canopy effects and provide limited ecological information. Fires and prescribed burns can also occur over weeks to months, limiting the relevance of a single post-fire image. We adapted the composite burn index, a field-truthing method commonly used in North America, to include parsimonious metrics that assess resprouting dynamics common among Australian eucalypts and the retention and recovery of key fauna habitat attributes. We used multitemporal satellite imagery to compile the greatest impacts within fire boundaries over several months and derive the maximum differenced normalised burn ratio (dNBR max). We used the dNBR max and a new field method, the OzCBI, to create distinct fire-severity models for three forested biogeographical regions in southwest Western Australia. Ecological outcomes were used to set thresholds between severity classes that were common to the three models. Balanced accuracies averaged above 0.7 across the models. This study demonstrates a practical method to incorporate ecological assessment into fire-severity maps. The approach informs efforts to achieve best-practice fire management that balances risk reduction and the conservation of natural environments. Key policy highlights The OzCBI is a parsimonious method for relating key habitat features to distinct fire-severity classes. Different biogeographic regions require distinct models relating satellite imagery and field-truthing. Multitemporal imagery increases the accuracy of severity maps when fires burn over extended periods (weeks to months).