Skip to content
DraconiticResearch
Search Sources API
194,161 studies31 sourcesUpdated 29 min ago
Peer-reviewedCC BY 4.0PLOS One

Physically aligned forest fire risk prediction: A deep learning framework coupling fuel and climate multivariate factors

  • Bojie Chen1
  • Anping Zeng2
  • Yu Xie
  • Xinjie Wang
  • Lihong Zhu
  • Qina Xie
  • Guoqing Long
  1. 1School of Computer Science and Technology, Yibin University, Yibin, China
  2. 2School of Business English, Chengdu International Studies University (Yibin Campus), Yibin, China
Open at sourcePDFDOICite

Abstract

Existing forest fire risk prediction methods often focus on single-element analysis, which makes it difficult to effectively capture the underlying mechanisms of the “fuel-climate” interaction. This paper proposes a physically aligned prediction framework named FWI-MSNet. Using 18 years of synchronized observation data from the Huitong Ecological Station in China, the framework constructs a multi-scale feature system, selecting 21 physically relevant key features, including core indicators of the Forest Fire Weather Index (FWI). A parallel multiscale 1-Dimensional Convolutional Neural Network (1D-CNN) extracts dynamic features across multiple temporal scales (from daily to seasonal). Subsequently, a synergistic mechanism integrating Gated Recurrent Units (GRU) and Transformer achieves deep integration of temporal evolution and global correlation. Experimental results demonstrate that compared to seven baseline models (including XGBoost, LSTM, CNN, and four ablation variants), the mean of RMSE, MAE, and MAPE decreases by 56.1%, and R2 improves from 0.223 (baseline average) to 0.9251. Furthermore, it accurately captures the trajectory of fire risk evolution in the 2020 catastrophic forest fires in Australia. This study provides a valuable reference for forest fire risk prediction.

Related

  1. Clinical-Anchored Incremental Survival Learning: A Stability-Constrained Framework for Molecularly Enhanced Survival Prediction in Lung Adenocarcinoma
    P. Shringarpure, S. Divakar, S. Khobragade
    bioRxivPreprintAll rights reservedFull textBioinformatics & computational biologybioRxiv
  2. sORF-Trans2MS: a two-module deep learning framework for sORF translation and MS-supported microprotein prediction
    J. He, J. Sui, Q. Chen +2
    bioRxivPreprintAll rights reservedFull textBioinformatics & computational biologybioRxiv
  3. MarkerScout: A Disease-Agnostic Machine Learning Framework for Biomarker Prediction from Multi-Scale Mechanistic Models
    Robert Moore, Frank Agayie-Ntim, Lindsey B. Crawford +5
    arXivPreprintCC BY-SA 4.0Full textGeneral biologyarXiv
  4. scDEFT: A deep learning framework for drug-effect prediction and counterfactual reasoning
    Murthy Devarakonda
    arXivPreprintCC BY-NC-ND 4.0Full textBioinformatics & computational biologyarXiv
  5. MetExPred: A Comprehensive Prediction Framework with Protein-Context-Aware Multi-view Learning for Drug Metabolism and Excretion
    Y. Huang, J. Wan, G. Wu +4
    bioRxivPreprintCC BY-NCFull textBioinformatics & computational biologybioRxiv

Cite

BibTeXRIS
BibTeX
@article{chen2026physically,
  title = {{Physically aligned forest fire risk prediction: A deep learning framework coupling fuel and climate multivariate factors}},
  author = {Bojie Chen and Anping Zeng and Yu Xie and Xinjie Wang and Lihong Zhu and Qina Xie and Guoqing Long},
  journal = {PLOS One},
  year = {2026},
  month = sep,
  doi = {10.1371/journal.pone.0355829},
  url = {https://doi.org/10.1371/journal.pone.0355829},
  note = {CC BY 4.0},
}
RIS
TY  - JOUR
TI  - Physically aligned forest fire risk prediction: A deep learning framework coupling fuel and climate multivariate factors
AU  - Bojie Chen
AU  - Anping Zeng
AU  - Yu Xie
AU  - Xinjie Wang
AU  - Lihong Zhu
AU  - Qina Xie
AU  - Guoqing Long
PY  - 2026
DA  - 2026/09/01
JO  - PLOS One
DO  - 10.1371/journal.pone.0355829
UR  - https://doi.org/10.1371/journal.pone.0355829
AB  - Existing forest fire risk prediction methods often focus on single-element analysis, which makes it difficult to effectively capture the underlying mechanisms of the “fuel-climate” interaction. This paper proposes a physically aligned prediction framework named FWI-MSNet. Using 18 years of synchronized observation data from the Huitong Ecological Station in China, the framework constructs a multi-scale feature system, selecting 21 physically relevant key features, including core indicators of the Forest Fire Weather Index (FWI). A parallel multiscale 1-Dimensional Convolutional Neural Network (1D-CNN) extracts dynamic features across multiple temporal scales (from daily to seasonal). Subsequently, a synergistic mechanism integrating Gated Recurrent Units (GRU) and Transformer achieves deep integration of temporal evolution and global correlation. Experimental results demonstrate that compared to seven baseline models (including XGBoost, LSTM, CNN, and four ablation variants), the mean of RMSE, MAE, and MAPE decreases by 56.1%, and R2 improves from 0.223 (baseline average) to 0.9251. Furthermore, it accurately captures the trajectory of fire risk evolution in the 2020 catastrophic forest fires in Australia. This study provides a valuable reference for forest fire risk prediction.
SN  - 1932-6203
ER  -