Spatial Uncertainty in Wildfire Forecasting Using Multi-Modal Earth Observation

TMLR Paper4994 Authors

Abstract:

Accurate wildfire forecasting from remote sensing data is essential for climate resilience and
emergency planning. Beyond predictive performance, understanding where and why uncer-
tainty arises is critical for operational trust. We analyze the spatial structure of predictive
uncertainty in wildfire spread forecasts using multimodal Earth observation (EO) inputs,
including Sentinel-2 vegetation indices and VIIRS thermal reflectance. Using Monte Carlo
dropout, Deep Ensembles, and Bayesian Neural Networks for uncertainty quantification, we
find that uncertainty estimates are spatially structured and concentrated near predicted fire
perimeters, consistent with the expected uncertainty in fire spread forecasts. We introduce
a novel and interpretable centroid-oriented distance metric that reveals high-uncertainty
regions consistently form 20–60 meter buffer zones around predicted firelines. Feature attri-
bution using integrated gradients highlights vegetation condition and recent fire activity as
primary drivers of model confidence. Together, these results suggest that spatial uncertainty
in EO-based wildfire forecasting is structured, interpretable, and operationally actionable.
The code for all experiments is available on GitHub.\footnote{\url{https://github.com/roloccark/wildf-UQ}}

Submission Length: Regular submission (no more than 12 pages of main content)

Changes Since Last Submission: Changes made throughout the manuscript based on reviewer feedback.

Assigned Action Editor: ~Evan_G_Shelhamer1

Submission Number: 4994