CanadaFireSat: Towards high-resolution wildfire forecasting with multiple modalities

Hugo Porta

doi:10.1016/j.isprsjprs.2026.05.050

CanadaFireSat: Towards high-resolution wildfire forecasting with multiple modalities

2026-06-19

0

OA

AI

DOI：10.1016/j.isprsjprs.2026.05.050

Original

Original request for help

Save

Abstract

En 中文

• CanadaFireSat is a high-resolution wildfire forecasting benchmark over Canada. • Sentinel-2 images and environmental variables yield more accurate fine-scale results. • We show CanadaFireSat’s potential by training two multi-modal models: ResNet and ViT.

Keywords:

Wildfire forecasting

Benchmark dataset

Multi-modal learning

Deep learning

Boreal ecosystem

CanadaFireSat: Towards high-resolution wildfire forecasting with multiple modalities

Abstract

Journal

Organization