More data and more support in the heat of the moment

The Problem

While the 2019/2020 Australian bushfire disaster unfolded, the team at Kablamo sought to find ways we could help by using the skills we had: developing technology to assist firefighters and other emergency personnel.

DELWP Victoria utilised modelling framework to calculate the likelihood and severity of a bushfire in any area called the Residual Risk metric. After four years, it needed an injection of cloud technology and machine learning capabilities to continue reducing bushfire risk across the state. Kablamo was the team to build this infrastructure.

The new platform had to overcome issues in the previous prediction metric, such as bottlenecks preventing scalability, while adding new opportunities such as increased spatial resolution. It needed to be accessible to a range of operators, such as scientists and fire chiefs, who applied the resulting metric in different and important ways. It also needed to be underpinned by cloud-based data storage and processing at many orders of magnitude greater than had previously existed at DELWP.

An upgrade like this would not be likely for several years, so the project needed to be as advanced and fit-for-purpose as possible.

Our Approach

Kabalmo worked with the DELWP team to design and deploy a cloud-native, fully scalable AWS machine learning platform, capable of handling large increases in workloads and processing with existing and future bushfire prediction models.

The bushfire modelling currently in use is PHOENIX RapidFire - a fire behaviour simulator that puts Australia at the forefront of bushfire tools and analysis innovation. To be even more accurate, PHOENIX RapidFire could be supported with more clean data from geospatial, temporal and historical sources. This vital risk metric was now going to be produced with the most accurate and powerful analytical and machine learning driven technologies underpinning it.

As such, Kablamo architected a solution for DELWP that focused on scalability, automation, cost-effectiveness, performance visibility, and removing bottlenecks. It was to be an advanced cloud platform, with automated workflows (including post-processing) that would future-proof this critical metric for bushfire prevention.

The Results

Kablamo designed and built a fully scalable, future-proof and cost effective AWS cloud platform for performing bushfire risk assessments that was completely cloud-native. The solution took DELWP’s initial modelling platform and launched it to a new level. Serverless processes and automated data pipelines with enhanced AWS machine learning capabilities overcame previous bottlenecks and manual workflows, enabling almost limitless scaling. Simulations can now be run with more than six times the number of ignition points and double the simulation resolution.

A rich user interface was introduced that allowed users to quickly create scenarios for simulation using predefined or newly uploaded data. A simple and powerful user experience, the team could quickly and easily repeat previous runs with slight modifications.

This redevelopment of an important resource - the ability to predict risk and subsequently manage bushfires - is an absolute milestone in Australia’s role as a leader in bushfire management, but also as we continue to live in a country with such a high level of environmental risk. Now, with a suit of AWS cloud infrastructure, infinitely scalable data processing, and machine learning advantages, we have a new and powerful tool with which to save lives, property and bushland.