Operational fire models like Spark and PHOENIX rely on fuel layers that are static, manually updated, and spatially coarse. They assume a uniform fuel response to time since fire — ignoring how growing conditions, management, and fire severity fundamentally alter fuel trajectories across Australia's diverse fire landscapes.

No single sensor captures every fuel attribute. Optical imagery tracks phenology but can't penetrate canopy. Spaceborne LiDAR measures structure but only in sparse footprints. Airborne surveys deliver wall-to-wall coverage but are costly and infrequent. Field plots offer detail but lack scale. The challenge isn't data scarcity — it's fusing disparate sensors, resolutions, and cadences into operationally consistent maps.

We fuse optical time series with spaceborne LiDAR through AlphaEarth — a geospatial model extracting structural and spectral features at 10 m resolution, enabling wall-to-wall fuel mapping ready for fire-spread and landscape risk models.

Foundation models offer a different path: one unified representation across sensors, seasons, and landscapes. We use AlphaEarth to fuse optical satellite time series with field plots and GEDI LiDAR into annual fuel maps of Victoria at 10m resolution.

FuelMaps delivers these datasets through a browser-based map explorer — no GIS software required. Browse, query, and compare fuel metrics at any location. Cloud-Optimised GeoTIFFs with per-pixel hover queries. Custom reporting tools in the next batch.