Dataset Card for CanadaFireSat 🔥🛰️

In this benchmark, we investigate the potential of deep learning with multiple modalities for high-resolution wildfire forecasting. Leveraging different data settings across two types of model architectures: CNN-based and ViT-based.

• 📝 Published paper from ISPRS (ArXiv Version)
  
• 💿 Dataset repository on GitHub
  
• 🤖 Model repository on GitHub & Weights on Hugging Face
• 🟰 Another "Raw" version of the data with NPY files organized in different folders per-modality types can be found at CanadaFireSat-Raw

Disclaimer: [23/06/2026] CanadaFireSat does not contain yet the ignition proxy predictors used in Appendix E.

📝 Summary Representation

The main use of this dataset is to push for the development of algorithms towards high-resolution wildfire forecasting via multi-modal learning. Indeed, we show the potential through our experiments of models trained on satellite image time series (Sentinel-2) and with environmental predictors (ERA5, MODIS, FWI). We hope to emulate the community to benchmark their EO and climate foundation models on CanadaFireSat to investigate their downstream fine-tuning capabilities on this complex extreme event forecasting task.

Sources

We describe below the different sources necessary to build the CanadaFireSat benchmark.

🔥📍 Fire Polygons Source

• 💻 National Burned Area Composite (NBAC 🇨🇦): Polygons Shapefile downloaded from CWFIS Datamart
  
• 📅 Filter fires since 2015 aligning with Sentinel-2 imagery availability
  
• 🛑 No restrictions are applied on ignition source or other metadata
  
• ➕ Spatial aggregation: Fires are mapped to a 2.8 km × 2.8 km grid | Temporal aggregation into 8-day windows

🛰️🗺️ Satellite Image Time Series Source

• 🛰️ Sentinel-2 (S2) Level-1C Satellite Imagery (2015–2023) from Google Earth Engine
  
• 🗺️ For each grid cell (2.8 km × 2.8 km): Collect cloud-free S2 images (≤ 40% cloud cover) over a 64-day period before prediction
  
• ⚠️ We discard samples with: Fewer than 3 valid images | Less than 40 days of coverage
  

🌦️🌲 Environmental Predictors

• 🌡️ Hydrometeorological Drivers: Key variables like temperature, precipitation, soil moisture, and humidity from ERA5-Land (11 km, available on Google Earth Engine) and MODIS11 (1 km, available on Google Earth Engine), aggregated over 8-day windows using mean, max, and min values.
• 🌿 Vegetation Indices (MODIS13 and MODIS15): NDVI, EVI, LAI, and FPAR (500 m) captured in 8 or 16-day composites, informing on vegetation state.
• 🔥 Fire Danger Metrics (CEMS previously on CDS): Fire Weather Index and Drought Code from the Canadian FWI system (0.25° resolution).
• 🕒 For each sample, we gather predictor data from 64 days prior to reflect pre-fire conditions.

🏞️ Land Cover

• ⛔️ Exclusively used for adversarial sampling and post-training analysis.
• 💾 Data extracted is the 2020 North American Land Cover 30-meter dataset, produced as part of the North American Land Change Monitoring System (NALCMS) (available on Google Earth Engine)

📊 CanadaFireSat Dataset Statistics (Without Test Hard):

Statistic	Value
Total Samples	177,801
Target Spatial Resolution	100 m
Region Coverage	Canada
Temporal Coverage	2016 - 2023
Sample Area Size	2.64 km × 2.64 km
Fire Occurrence Rate	39% of samples
Total Fire Patches	16% of patches
Training Set (2016–2021)	78,030 samples
Validation Set (2022)	14,329 samples
Test Set (2023)	85,442 samples
Sentinel-2 Temporal Median Coverage	55 days (8 images)
Number of Environmental Predictors	58
Data Sources	ERA5, MODIS, CEMS

📍 Samples Localisation:

Figure 1: Spatial distribution of positive (left) and negative (right) wildfire samples.

🛰️ Example of S2 time series:

Figure 2: Row 1-3: Samples of Sentinel-2 input time series for 4 locations in Canada, with only the RGB bands with rescaled intensity. Row 4: Sentinel-2 images after the fire occurred. Row 5: Fire polygons used as labels with the Sentinel-2 images post-fire.

Dataset Structure

Name	Type	Shape	Description
date	timestamp[s]	-	Fire Date
doy	sequence<int64>	-	Sentinel-2 Tiles Day of the Year
10x	sequence<array3_d>	(4, 264, 264)	Sentinel-2 10m bands, Order: ["B4", "B3", "B2", "B8"]
20x	sequence<array3_d>	(6, 132, 132)	Sentinel-2 20m bands, Order: ["B5", "B6", "B7", "B8A", "B11", "B12"]
60x	sequence<array3_d>	(3, 44, 44)	Sentinel-2 60m bands, Order: ["B1", "B9", "B10"]
loc	array3_d<float32>	(2, 264, 264)	Latitude and Longitude grid
labels	array2_d<uint8>	(264, 264)	Fire binary label mask
tab_cds	array2_d<float32>	(8, 6)	Tabular CDS variables
tab_era5	array2_d<float32>	(8, 45)	Tabular ERA5 variables
tab_modis	array2_d<float32>	(8, 7)	Tabular MODIS products
env_cds	array4_d<float32>	(8, 6, 13, 13)	Spatial CDS variables
env_cds_loc	array3_d<float32>	(13, 13, 2)	Grid coordinates for CDS
env_era5	array4_d<float32>	(8, 45, 32, 32)	Spatial ERA5 variables
env_era5_loc	array3_d<float32>	(32, 32, 2)	Grid coordinates for ERA5
env_modis11	array4_d<float32>	(8, 3, 16, 16)	Spatial MODIS11 variables
env_modis11_loc	array3_d<float32>	(16, 16, 2)	Grid coordinates for MODIS11
env_modis13_15	array4_d<float32>	(8, 4, 32, 32)	Spatial MODIS13/15 variables
env_modis13_15_loc	array3_d<float32>	(32, 32, 2)	Grid coordinates for MODIS13/15)
env_doy	sequence<int64>	-	Environment Variables Day of the Year
region	string	-	Canadian Province or Territory
tile_id	int32	-	Tile identifier
file_id	string	-	Unique file identifier
fwi	float32	-	Tile Fire Weather Index

📊 Performance Analysis: In this table, we describe the models' performances across data settings and architectures.

Encoder	Modality	Params (M)	Val PRAUC	Val F1	Test PRAUC	Test F1	Test Hard PRAUC	Test Hard F1	Avg PRAUC	Avg F1
🔷 ResNet-50
	SITS Only	52.2	45.9	49.4	54.0	59.9	26.2	36.7	42.0	48.7
	ENV Only	97.5	41.6	46.7	50.8	55.2	24.5	33.1	39.0	45.0
	Multi-Modal	52.2	46.1	51.2	57.0	60.3	27.1	37.4	43.4	49.6
🔶 ViT_S
	SITS Only	36.5	45.2	50.6	51.2	51.9	25.7	33.8	40.7	45.2
	ENV Only	54.8	34.8	45.7	49.2	59.9	21.2	35.1	35.1	46.9
	Multi-Modal	37.7	43.9	50.0	56.2	59.2	24.7	35.6	41.6	48.3
⚫ Baselines
Baseline (FWI)	ENV Only	-	20.0	32.7	43.1	50.3	21.1	32.7	28.1	38.6
Baseline (UNet) ¹	ENV Only	9.1	33.6	43.2	51.4	58.4	25.1	34.2	36.7	45.3
Baseline (UTAE) ²	ENV Only	1.1	32.9	43.8	47.2	52.5	22.0	31.7	34.0	42.7
Baseline (ConvLSTM) ³	SITS Only	1.2	41.4	46.0	50.2	58.9	23.1	35.0	38.2	46.6

¹ Prapas et al., 2023 ² Michail et al., 2025 ³ Yang et al., 2021

Citation

The paper has been published in the ISPRS Journal of Photogrammetry and Remote Sensing.

@article{porta2026canadafiresat,
  title={CanadaFireSat: Towards high-resolution wildfire forecasting with multiple modalities},
  author={Porta, Hugo and Dalsasso, Emanuele and McCarty, Jessica L and Tuia, Devis},
  journal={ISPRS Journal of Photogrammetry and Remote Sensing},
  volume={239},
  pages={555--572},
  year={2026},
  publisher={Elsevier}
}

Contacts & Information

• Curated by: Hugo Porta
• Contact Email: hugo.porta@epfl.ch
• Shared by: ECEO Lab
• License: MiT License