FireCastNet: Earth-as-a-Graph for Seasonal Fire Prediction

TL;DR

FireCastNet addresses global seasonal wildfire forecasting by modeling the Earth as a graph and integrating 3D convolutional encoding with a GraphCast-based multi-mesh GNN. The approach uses the SeasFire datacube to predict burned-area patterns up to six months ahead, achieving superior global performance over baseline and state-of-the-art models and enabling region-specific refinements via Local Area Modelling. Key findings show that longer input time-series improve robustness, spatial context enhances forecasts across extended horizons, and region-focused models can yield higher resolution predictions in fire-prone zones. The work demonstrates the importance of Earth-system interactions in long-term wildfire prediction and provides a practical framework for scalable, high-resolution forecasts with public code availability.

Abstract

With climate change intensifying fire weather conditions globally, accurate seasonal wildfire forecasting has become critical for disaster preparedness and ecosystem management. We introduce FireCastNet, a novel deep learning architecture that combines 3D convolutional encoding with GraphCast-based Graph Neural Networks (GNNs) to model complex spatio-temporal dependencies for global wildfire prediction. Our approach leverages the SeasFire dataset, a comprehensive multivariate Earth system datacube containing climate, vegetation, and human-related variables, to forecast burned area patterns up to six months in advance. FireCastNet treats the Earth as an interconnected graph, enabling it to capture both local fire dynamics and long-range teleconnections that influence wildfire behavior across different spatial and temporal scales. Through comprehensive benchmarking against state-of-the-art models including GRU, Conv-GRU, Conv-LSTM, U-TAE, and TeleViT, we demonstrate that FireCastNet achieves superior performance in global burned area forecasting, with particularly strong results in fire-prone regions such as Africa, South America, and Southeast Asia. Our analysis reveals that longer input time-series significantly improve prediction robustness, while spatial context integration enhances model performance across extended forecasting horizons. Additionally, we implement local area modeling techniques that provide enhanced spatial resolution and accuracy for region-specific predictions. These findings highlight the importance of modeling Earth system interactions for long-term wildfire prediction.

Figures (7)

• Figure 1: High-level representation of our architecture. The FireCastNet architecture consists of three main components: (a) a cube embedding block for spatio-temporal feature extraction using a 3D convolution layer that reduces the spatial and temporal dimensions of the input timeseries data (b) a GraphCast block, which leverages a multi-mesh Graph Neural Network (GNN) to model long-range spatial interactions across a refined icosahedral mesh structure, and (c) a sub-pixel convolution block for upscaling the output to match the resolution of the original input.
• Figure 2: Target variable and prediction (per horizon) of the best global model, using a time-series of length 24, on 30 September 2019. The target variable is always the same. The input variables depend on the prediction horizon, i.e. for $h=1$ we use a time-series of length 24 ending one $8$-day period before, for $h=2$ we use a time-series of length 24 ending two $8$-day periods before, and so on. The sigmoid output represents the prediction confidence.
• Figure 3: Model performance at a global scale with different time-series length $ts \in \{6, 12, 24\}$ periods (8-days).
• Figure 4: Performance of FireCastNet model at a global scale with timeseries length $24$ for different time overlap windows.
• Figure 5: Feature attributions for fire-danger predictions with the global model across forecasting horizons (1, 2, 4, 8, 16, 24 8-days). Each column sums to 1.0, indicating the relative importance of each variable for the respective horizon. The results where obtained using Integrated Gradients sundararajan2017axiomatic.