Significant scientific and functional feature updates

One of the standout enhancements in this release is the inclusion of an ensemble template that allows comparison between the Vesta, Vesta Mk2, and McArthur fire spread models. Another notable advancement is the overhaul of existing ensemble capability. The maximum number of ensemble members that can be run simultaneously has increased from 12 to 20, significantly boosting the platform’s power for probabilistic forecasting and scenario planning.

These changes mean users can now better capture the inherent uncertainty in weather and fuel variables, leading to richer, more informative outputs. Alongside this, ensemble templates have been improved to allow for the comparison of multiple spread models or fuel age multipliers within a single scenario, making Spark a more powerful tool for comparative risk analysis and operational planning.

The template system has been upgraded. The weather ensemble template now enables scenario-specific weather editing through comma-separated values (CSV) within the graphical user interface, empowering users to tailor simulations more simply than previous versions. The addition of an editable CSV interface for variable data, including fuel lookup tables, is a step-change for usability and scientific experimentation. Users can now adjust key simulation parameters within the Spark interface itself, facilitating rapid scenario testing without needing external software. These tools expand the scientific repertoire available to users, offering more validated and varied options for simulating bushfire behaviour under a range of conditions.

Spread diagnostics have been introduced with a raster based visual output to help users quickly identify why a fire may not have spread in a given simulation. This assists with live validation of input data and fire spread.

Additionally, the introduction of a small fire build-up sub-model allows for more accurate modelling of early-stage fire dynamics, supporting better risk assessment during critical periods.

This version also introduces a re-ignition model based on fireline intensity, with a new threshold column (‘SeFLI’) in the fuel lookup table. This innovation enables more nuanced simulation of extinguishment and reignition events, which is crucial for accurately modelling complex and evolving firegrounds.

Plume visualisation has also been introduced. Users can now visualise fire plumes for both individual runs and ensembles, providing deeper insights into fire spread and atmospheric interactions across a range of forecast conditions.

On the server side, simulation resolution is now a configurable parameter, giving advanced users more control over the spatial detail of their forecasts. Backend improvements mean the system now supports advanced plume and spotting models directly in ensembles, making it easier to account for ember transport and complex fire behaviour in real-time.

Looking ahead

While this release marks a substantial leap forward, the Spark team continues to refine models and enhance integration with agency tools. Ongoing development will focus on further scientific validation, better integration of real-time weather data, and additional tools for diagnosis and analysis of rapidly evolving fire situations. Planning for Phase 5 is in progress and the AFAC Spark team have run a series of webinars with agency users to refine requirements. Phase 5 will focus on ensuring Spark meets the needs of end users, before adding in new capabilities and new science.

Conclusion

The 1.3.0 release elevates Spark Operational bushfire simulation with new models, expanded ensemble analysis, and improved scientific transparency. These capabilities will help AFAC member Fire Behaviour Analysts better prepare for and respond to bushfire incidents, supporting safer communities and more effective emergency fire management.

Spark 1.3.0 will be released for download in September 2025.

For more information or if you require support with Spark Operational, please contact Chantelle O’Brien or Andrew Rosa at firepredictions@afac.com.au.