Machine Learning & AI Meetup

Talk description: Initial attack refers to the first set of firefighting actions undertaken at a newly reported fire, with the aim to control the fire’s spread. When multiple fires are reported and resources constrained, decisions need to be made about triaging fires for an escalated response. We developed empirical initial attack models for grass and forest fires for the specific purpose of triaging fires. Using a recently developed dataset of spreading fires, we tested three statistical modelling techniques to estimate the probability of unsuccessful initial attack. We assessed the models' performance and how practical it would be to use them in an operational setting using statistical techniques and case studies. In this talk, we will discuss how the models have limited skill overall and are unable to distinguish fires where bottlenecks affect the initial attack suppression. We will highlight the importance of taking an application specific approach to developing and validating initial attack models. A preprint of this work can be found on bioRxiv.

Speaker bio: Dr Elena Tartaglia is a Principal Data Scientist at the Victorian government Department of Energy, Environment and Climate Action, where she has worked for the past three years in Bushfire and Forest Services (BFS). BFS works to keep Victorians and the things they care about safe from bushfires, through land management actions and by responding to bushfires on public land. Elena and her team use statistics and machine learning to support data-driven decision-making in BFS. Elena is also the current President of the Victorian and Tasmanian Branch of the Statistical Society of Australia.

Ping Siriamnat: "Impact of fire suppressant drops on fire severity"

Talk description: One of the common suppression methods of bushfires is aerial drops, but there is limited research on its effectiveness. Aerial drops refer to aircraft flying over a fire and droping water or fire suppressant to slow or stop the spread of the fire. Through the National Aerial Firefighting Centre, aircraft attending to bushfires in Australia are tracked, so we have data on when and where drops were done, along with what type of suppressant was used and how much. In this project, we investigated the impact of drops on fire severity by measuring the association between drops and fire severity level, as calculated by the Arthur Rylah Institute, accounting for the influence of terrain steepness and land cover characteristics.

Speaker bio: Ping Siriamnat is a Master of Business Analytics graduate (Monash University) and previously interned at DEECA in the DAIS Data Science team. Across his projects, he applies statistical methods and machine learning models, alongside his business acumen, to transform data into actionable insights. He has a strong interest in market analytics, business strategy, and applying data science to uncover opportunities and improve the performance of strategic decisions.