Projecting Large Fires in the Western US With an Interpretable and Accurate Hybrid Machine Learning Method

Earth s Future, Journal Year: 2024, Volume and Issue: 12(10)

Published: Oct. 1, 2024

Abstract More frequent and widespread large fires are occurring in the western United States (US), yet reliable methods for predicting these fires, particularly with extended lead times a high spatial resolution, remain challenging. In this study, we proposed an interpretable accurate hybrid machine learning (ML) model, that explicitly represented controls of fuel flammability, availability, human suppression effects on fires. The model demonstrated notable accuracy F 1 ‐score 0.846 ± 0.012, surpassing process‐driven fire danger indices four commonly used ML models by up to 40% 9%, respectively. importantly, showed remarkably higher interpretability relative other models. Specifically, demystifying “black box” each using explainable AI techniques, identified substantial structural differences across models, even among those similar accuracy. relationships between their drivers, our were aligned closer established physical principles. discrepancy led diverse predictions exhibited greater consistency actual occurrence. With highly revealed strong compound from multiple climate variables related evaporative demand, energy release component, temperature, wind speed, dynamics megafires US. Our findings highlight importance assessing integrity addition They also underscore critical need address rise extremes linked wildfires.

Language: Английский

---

Long-term tracking of recovery of built infrastructure after wildfires with deep network topologies

Neural Computing and Applications, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 29, 2025

Language: Английский

---

Learning, Catastrophic Risk, and Ambiguity in the Climate-Change Era

Environmental and Energy Policy and the Economy, Journal Year: 2025, Volume and Issue: 6(1), P. 140 - 168

Published: Jan. 1, 2025

Language: Английский

---

Responses of leaf-level physiological traits and water use characteristics to drought of a xerophytic shrub in northern China

Journal of Hydrology, Journal Year: 2025, Volume and Issue: unknown, P. 133204 - 133204

Published: March 1, 2025

Language: Английский

---

Evaluating building-level tree cover change in Southern California wildland-urban interface using high-resolution satellite imagery

Journal of Environmental Management, Journal Year: 2025, Volume and Issue: 380, P. 125160 - 125160

Published: April 1, 2025

Language: Английский

---