Data Clouds to Prevent Pyro Clouds

During the devastating Australian bushfires in the summer of 2019/2020, fires burnt more than 18 million hectares of land. More than 30 lives were lost, billions of animals were killed and damage bill estimates are greater than $200 billion. The scale and complexity of the bushfires and weather systems was unprecedented and it pushed people, processes and existing technology to their limits. Kablamo took on the challenge of rethinking how data and machine learning could be used to better manage, predict and prevent bushfires in the future.

From interviewing volunteer firefighters, brigade captains, fire-behavioural analysts and fire scientists and technology teams, we uncovered some opportunities to apply our expertise and experience:

• Introducing more power, scalability and AI capabilities to the data platforms
• Enabling fast integration with new fire prediction models as they became available
• Reducing manual steps and introducing new digital workflows
• Designing an accessible interface to relay information from the analysts to incident controllers and front line teams

Within a matter of weeks, Kablamo had created a working prototype, which provided a scalable cloud platform and intuitive user interface for displaying real bushfire data and managing predictions. Importantly, the data platform was capable of ingesting and processing almost limitless amounts of data, which is essential for managing large and complex bushfire events.

The interoperability of the new data platform with existing core mapping systems, mathematical prediction models and decision workflows was critical. To address this, we created APIs within the platform to interact with common prediction models and geospatial mapping systems.

Importantly, the new data platform will assist State and Territory fire fighting services to be more prepared for a future of longer, more intense fire seasons. The scalable data platform and extensible architecture enables a future of more sophisticated data analysis, machine-learning based fire prediction models and even automated fire response mechanisms, such as robotics and drones.