Saved in:
Bibliographic Details
Main Authors: Bai, Yuchen, Athanasiou, Georgios, Yu, Xin, Antonopoulos, Diogenis, Papoutsis, Ioannis, Hantson, Stijn, Carvalhais, Nuno
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.10298
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913111285432320
author Bai, Yuchen
Athanasiou, Georgios
Yu, Xin
Antonopoulos, Diogenis
Papoutsis, Ioannis
Hantson, Stijn
Carvalhais, Nuno
author_facet Bai, Yuchen
Athanasiou, Georgios
Yu, Xin
Antonopoulos, Diogenis
Papoutsis, Ioannis
Hantson, Stijn
Carvalhais, Nuno
contents Accurate next-day active fire forecasts can support early warning, disaster response, forest risk assessment, and downstream estimation of fire-related carbon emissions. Existing machine learning approaches to wildfire forecasting typically predict wildfire danger or fire probability on kilometre-scale daily grids, which is useful for regional warning but does not directly represent localized fire events. We propose Wildfire Ignition Set Predictor (WISP), a query-based model that reformulates next-day active fire forecasting as point-set prediction. From 48 hours of covariates including meteorology, satellite vegetation products, static land, and fire history, WISP predicts a fixed-size ranked set of future active fire cluster centres on a 375 m grid across globally distributed regions. The model is trained end-to-end with Hungarian matching; to address the conflicting roles of the classification score in assignment, ranking, and query activation, we use asymmetric classification-localization weighting in matching and loss. We further construct a globally distributed, hourly, multi-source benchmark for this task. On a held-out test set spanning fire regions worldwide, the best WISP variant achieves 38.2% average precision (AP) for ranked fire-centre detections, covers 53.4% of fire cluster mass weighted by fire radiative power (FRP), and localizes 54.1% of observed clusters within 5 km. These results establish sparse set prediction as a viable formulation for high-resolution wildfire forecasting and provide a benchmark for future work in this regime.
format Preprint
id arxiv_https___arxiv_org_abs_2605_10298
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Set Prediction for Next-Day Active Fire Forecasting
Bai, Yuchen
Athanasiou, Georgios
Yu, Xin
Antonopoulos, Diogenis
Papoutsis, Ioannis
Hantson, Stijn
Carvalhais, Nuno
Machine Learning
Accurate next-day active fire forecasts can support early warning, disaster response, forest risk assessment, and downstream estimation of fire-related carbon emissions. Existing machine learning approaches to wildfire forecasting typically predict wildfire danger or fire probability on kilometre-scale daily grids, which is useful for regional warning but does not directly represent localized fire events. We propose Wildfire Ignition Set Predictor (WISP), a query-based model that reformulates next-day active fire forecasting as point-set prediction. From 48 hours of covariates including meteorology, satellite vegetation products, static land, and fire history, WISP predicts a fixed-size ranked set of future active fire cluster centres on a 375 m grid across globally distributed regions. The model is trained end-to-end with Hungarian matching; to address the conflicting roles of the classification score in assignment, ranking, and query activation, we use asymmetric classification-localization weighting in matching and loss. We further construct a globally distributed, hourly, multi-source benchmark for this task. On a held-out test set spanning fire regions worldwide, the best WISP variant achieves 38.2% average precision (AP) for ranked fire-centre detections, covers 53.4% of fire cluster mass weighted by fire radiative power (FRP), and localizes 54.1% of observed clusters within 5 km. These results establish sparse set prediction as a viable formulation for high-resolution wildfire forecasting and provide a benchmark for future work in this regime.
title Set Prediction for Next-Day Active Fire Forecasting
topic Machine Learning
url https://arxiv.org/abs/2605.10298