@conference {bnh-6527, title = {Fuels3D: barking up the wrong tree and beyond}, booktitle = {AFAC19 powered by INTERSCHUTZ - Bushfire and Natural Hazards CRC Research Forum}, year = {2019}, month = {12/2019}, publisher = {Australian Institute for Disaster Resilience}, organization = {Australian Institute for Disaster Resilience}, address = {Melbourne}, abstract = {

Improvement of the understanding of how fuel characteristics correlate with fire behaviour and severity is critical to the ongoing handling of risk and recovery in fire-prone environments. Current standards and protocols for describing fuel hazard (for example, {\textquoteleft}Overall Fuel Hazard Assessment Guide{\textquoteright}, Victorian Department of Sustainability and Environment) and post-burn severity (for example, {\textquoteleft}Fire Severity Assessment Guide{\textquoteright}, Victorian Department of Sustainability and Environment) were written for collection of information in the field. The data collected are largely subjective descriptions of the landscape. The ability of information from these assessment techniques to be adapted to modern risk assessment tools such as fire behavior models, or for the calibration and validation of datasets, is limited. Quantitative data-rich methods of measuring and assessing fuel load and structure are the missing link between the knowledge of land management personnel in the field, and the model drivers and decision makers at organizational level.

Handheld devices with high quality sensors, in the form of offthe-shelf cameras, are increasingly ubiquitous, as is the availability of 3D point cloud data collected from active sensing instruments on terrestrial and aerial platforms. Rapid and comprehensive capture of information by these devices, coupled with the use of computer vision techniques, allows for the 3D description of the surrounding environment to be exploited to provide robust measurement of metrics that can be built into existing fuel hazard assessment frameworks. Providing key metrics as data products rather than a single product enables flexibility across jurisdictions and ecosystem types, and capacity to adapt as end-user requirements change.

The Fuels3D project has created a suite of tools and methods for image capture in the field during fuel hazard assessments. 3D point clouds are generated using computer vision and photogrammetry techniques. From these 3D point clouds, scale is added, and decision rules are programmed to calculate quantifiable surface / near-surface metrics that replicate those
used in current fuel hazard visual assessment guides. Case studies are highlighted here.

Download the full non-peer reviewed research proceedings\ from the Bushfire and Natural Hazards CRC Research Forumhere.

}, keywords = {data collection, Fire behaviour, fuel hazard, risk management, technology}, url = {https://knowledge.aidr.org.au/resources/australian-journal-of-emergency-management-monograph-series/}, author = {Karin Reinke and Luke Wallace and Samuel Hillman and Bryan Hally and Simon Jones} } @conference {bnh-6524, title = {Intelligent warnings: a twenty-first century approach to encouraging protective action in emergencies}, booktitle = {AFAC19 powered by INTERSCHUTZ - Bushfire and Natural Hazards CRC Research Forum}, year = {2019}, month = {12/2019}, publisher = {Australian Institute for Disaster Resilience}, organization = {Australian Institute for Disaster Resilience}, address = {Melbourne}, abstract = {

The Victorian approach to the delivery of public information and warnings is robust and leads the country in many respects. Despite this, when at-risk individuals do receive warnings, existing research clearly highlights they are unlikely to immediately act, and instead, will seek out further information and take time to process the information to determine whether any action is required. This verification process may include talking with family, friends, neighbours or colleagues, resulting in a delay before protective action is taken. To counteract this problem and provide communities with as much time as possible to take action, we need to minimise the likelihood of delays infiltrating our decision-making and warnings dissemination process. Artificial Intelligence (AI) and automation can help us to achieve this.

To investigate this further, a desktop review of existing literature was undertaken, supported by informal discussions and semi-structured interviews with a diverse range of academics, emergency managers and other experts across Victoria, New South Wales, Queensland and the United States. The discussions and interviews highlighted exciting opportunities to enhance our current approach through the use of AI and automation, which could be particularly helpful in enhancing the effectiveness of warnings for rapid impact emergencies, such as flash flooding and severe thunderstorms. During these types of emergencies, where community consequences are often experienced very rapidly after initial onset of the event, AI and automation can support the tailoring of language and content in warning products according to affected communities and likely consequences, minimise warning issuance delay and maximise the effectiveness of decision-making.

The investigation found these technologies are not a replacement for human decision making, however, if leveraged effectively they will enable us to better understand risk in realtime and reduce the considerable time taken to manually process intelligence and apply our pre-determined triggers and business rules. For the Victorian emergency management sector, these results suggest there is a need to prioritise investment in innovative new approaches to support the dissemination of potentially life-saving public information and warnings, whilst also supporting researchers to further understand human behaviour and decision-making upon receipt of a warning. In summary, this investigation has identified opportunities to better support communities to take protective action in a timely manner, to ensure we achieve our shared vision of safer and more resilient communities.

Download the full non-peer reviewed research proceedings\ from the Bushfire and Natural Hazards CRC Research Forumhere.

}, keywords = {Emergency management, protective action, risk management, technology, warning}, url = {https://knowledge.aidr.org.au/resources/australian-journal-of-emergency-management-monograph-series/}, author = {Riley, Jacob} }