Fire Occurrence Predictions and Modelling Scripts for Victoria

This repository contains scripts and key data for modelling future fire occurrence in Victoria, Australia, under climate change. It supports the replication of results from the manuscript 'Forecasted increases in fire occurrence in Victoria under climate change', using a workflow based on Boosted Regression Trees (BRTs). The core components include functions for model fitting, cross-validation, and generating spatial probability predictions.

Use Cases

• Fitting Boosted Regression Tree models for fire probability based on climate and environmental covariates.
• Generating spatial predictions of future fire occurrence probability across Victoria.
• Conducting cross-validation to assess model performance for fire forecasting.
• Recreating publication figures from the associated climate change and wildfire study.

Strengths

• Provides a complete, documented R workflow for replicating a published scientific study.
• Released under a permissive CC-BY-4.0 license, facilitating reuse and adaptation.
• Cross-platform presence on figshare indicates established provenance and importance for the research community.

Limitations

• Critical external raw data (DEECA spatial layers, AGCD/AWO climate grids) are not included, limiting full replication without additional sourcing.
• Specific column names, row counts, and dataset structure are not detailed in the provided metadata.
• The repository is large (~22.4 GB), which may pose accessibility challenges.

Provenance

Source

Authors: Charlie Hart, Andrew Dowdy, Sarah C. McColl-Gausden, Hamish Clarke, Luke Collins, Amelia French, Trent D. Penman, Nevil Amos, Angie Haslem, Cindy E. Hauser, Jim Thomson, Josephine MacHunter, Matt White & William L. Geary.

Collection Method

Modelling workflow using Boosted Regression Trees (BRTs) on historical fire and climate data.

Time Range

null

Freshness

Last updated on 2026-03 24.

Geography

Victoria, Australia.