ShearFuse-UNet: Hadamard, DCT, and Shearlet Transform Fusion for Next-Day Wildfire Spread Prediction

2026-06-11

AI Summary

This work proposes a lightweight frequency-domain fusion U-Net architecture for next-day wildfire spread prediction. The method integrates three fixed spectral transforms—two-dimensional Walsh-Hadamard transform, discrete cosine transform, and cone-adapted digital shearlet transform—in parallel within the encoder, and fuses their outputs with residual pathways via a learnable SpectralFusion gating mechanism to effectively capture fireline edge structures. By replacing learnable projections with fixed mathematical transforms, the model avoids reliance on self-attention mechanisms. On the WildfireSpreadTS dataset, it achieves an F1 score of 0.596 with only 267k parameters, outperforming a ResNet18-U-Net variant with 14 million parameters (F1 = 0.589). Strong generalization is further demonstrated on the Google Next-Day Wildfire dataset.

Abstract

We propose ShearFuse-UNet, a lightweight and computationally efficient deep learning model for next-day wildfire spread prediction from multi-modal satellite data. The model integrates three complementary transform-domain branches inside each encoder block of a U-Net backbone: a 2D Fast Walsh-Hadamard Transform (WHT) branch, a 2D Discrete Cosine Transform (DCT) branch, and a cone-adapted digital Shearlet residual branch. The WHT and DCT branches establish orthogonal latent spaces with learnable spectral scaling and fixed soft-thresholding, while the Shearlet branch provides anisotropic, multi-directional feature decomposition that explicitly encodes the elongated edge structures characteristic of fire fronts. A learned SpectralFusion gate adaptively combines the WHT and DCT responses, and the Shearlet reconstruction is added as a residual. This three-branch design bears a loose structural analogy to transformer self-attention: the WHT and DCT branches provide complementary spectral representations that are adaptively fused, while the Shearlet branch contributes directional content through a residual pathway. Unlike self-attention, the proposed design relies on fixed mathematical transforms rather than learned projection operators, reducing parameter count and computational cost. Evaluated on the WildfireSpreadTS dataset, ShearFuse-UNet achieves an F1 score of 0.596 with only 267k parameters, outperforming a ResNet18-based U-Net (14M parameters, F1 = 0.589) and demonstrating a highly favorable accuracy-efficiency trade-off. Results on the Google Next-Day Wildfire Spread dataset further validate these findings across a different benchmark.

Problem

• wildfire spread prediction
• multi-modal satellite data
• next-day forecasting
• computational efficiency
• deep learning

Innovation

• ShearFuse-UNet
• transform-domain fusion
• Shearlet transform
• lightweight wildfire prediction
• spectral feature fusion