Sub-hourly forecasting of fire potential using machine learning on time series of surface weather variables

Alberto Ardid; Andres Valencia; Anthony Power; Matthias M. Boer; Marwan Katurji; Shana Gross; David Dempsey

doi:10.1071/WF24113

RESEARCH ARTICLE (Open Access)

Sub-hourly forecasting of fire potential using machine learning on time series of surface weather variables

Alberto Ardid ^A ^* , Andres Valencia ^A , Anthony Power ^B , Matthias M. Boer ^C , Marwan Katurji ^D , Shana Gross ^E and David Dempsey ^A

+ Author Affiliations

- Author Affiliations

^A Civil and Natural Resources Engineering Department, University of Canterbury, New Zealand.

^B Covey Associates, Australia.

^C Hawkesbury Institute for the Environment, Western Sydney University, Australia.

^D School of Earth and Environment, University of Canterbury, New Zealand.

^E New Zealand Forest Research Institute, Scion, New Zealand.

^* Correspondence to: aardids@gmail.com

International Journal of Wildland Fire 34, WF24113 https://doi.org/10.1071/WF24113

Submitted: 2 July 2024 Accepted: 7 January 2025 Published: 30 January 2025

© 2025 The Author(s) (or their employer(s)). Published by CSIRO Publishing on behalf of IAWF. This is an open access article distributed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC)

Abstract

Background

Rapidly developing pre-fire weather conditions contributing to sudden fire outbreaks can have devastating consequences. Accurate short-term forecasting is important for timely evacuations and effective fire suppression measures.

Aims

This study aims to introduce a novel machine learning-based approach for forecasting fire potential and to test its performance in the Sunshine Coast region of Queensland, Australia, over a period of 15 years from 2002 to 2017.

Methods

By analysing real-time data from local weather stations at a sub-hourly temporal resolution, we aimed to identify distinct weather patterns occurring hours to days before fires. We trained random forest machine learning models to classify pre-fire conditions.

Key results

The models achieved high out-of-sample accuracy, with a 47% higher accuracy than the standard fire danger index for the region. When simulating real forecasting conditions, the model anticipated 75% of the fires (11 out of 15).

Conclusions

This method provides objective, quantifiable information, enhancing the precision and effectiveness of fire warning systems.

Implications

The proposed forecasting approach supports decision-makers in implementing timely evacuations and effective fire suppression measures, ultimately reducing the impact of fires.

Keywords: early warning, fire danger, fire potential, fire potential forecasting, fire potential probability, machine learning, surface weather variables, time series feature engineering, weather station data.

References

Abt KL, Prestemon JP, Gebert KM (2009) Fire suppression cost forecasts for the US Forest Service. Journal of Forestry 107(4), 173-178.
| Crossref | Google Scholar |

Archibald S, Roy DP, van Wilgen BW, Scholes RJ (2009) What limits fire? An examination of drivers of burnt area in southern Africa. Global Change Biology 15(3), 613-630.
| Crossref | Google Scholar |

Ardid A, Dempsey D, Caudron C, et al. (2022) Seismic precursors to the Whakaari 2019 phreatic eruption are transferable to other eruptions and volcanoes. Nature Communications 13, 2002.
| Crossref | Google Scholar | PubMed |

Ardid A, Dempsey D, Caudron C, Cronin S, Kennedy B, Girona T, et al. (2023) Generalized Eruption Forecasting Models using Machine Learning Trained on Seismic Data from 24 Volcanoes. https://www.researchsquare.com/article/rs-3483573/v1

Ardid A, Dempsey D, Corry J, Garrett O, Lamb OD, Cronin S (2024a) Multitimescale template matching: discovering eruption precursors across diverse volcanic settings. Seismological Research Letters 95, 2611-2621.
| Crossref | Google Scholar |

Ardid A, Valencia A, Power A, Dempsey D (2024b) Fire danger forecasting on an hourly basis via weather station-based machine learning in the Sunshine Coast, Australia. Preprint. 10.21203/rs.3.rs-3857623/v1

Bell N (2019) Development in Australian bushfire prone areas. Environment [5] 1-14.
| Google Scholar |

Benjamini Y, Yekutieli D (2001) The control of the false discovery rate in multiple testing under dependency. Annals of Statistics 29, 1165-1188.
| Crossref | Google Scholar |

Bergado JR, Persello C, Reinke K, Stein A (2021) Predicting fire burns from big geodata using deep learning. Safety Science 140, 105276.
| Crossref | Google Scholar |

Beverly JL, Bothwell P, Conner JCR, Herd EPK (2010) Assessing the exposure of the built environment to danger ignition sources generated from vegetative fuel. International Journal of Wildland Fire 19(3), 299-313.
| Crossref | Google Scholar |

Bjanes A, De La Fuente R, Mena P (2021) A deep learning ensemble model for fire susceptibility mapping. Ecological Informatics 65, 101397.
| Crossref | Google Scholar |

Boer MM, De Dios VR, Stefaniak EZ, Bradstock RA (2021) A hydroclimatic model for the distribution of fire on Earth. Environmental Research Communications 3, 035001.
| Crossref | Google Scholar |

Bond WJ, Keeley JE (2005) Fire as a global ‘herbivore’: the ecology and evolution of flammable ecosystems. Trends in Ecology & Evolution 20(7), 387-394.
| Crossref | Google Scholar | PubMed |

Bowman DMJS, Balch JK, Artaxo P, Bond WJ, Carlson JM, Cochrane MA, et al. (2009) Fire in the Earth system. Science 324(5926), 481-484.
| Crossref | Google Scholar | PubMed |

Bradstock RA (2010) A biogeographic model of fire regimes in Australia: current and future implications. Global Ecology and Biogeography 19, 145-158.
| Crossref | Google Scholar |

Brody-Heine S, Zhang J, Katurji M, Pearce HG, Kittridge M (2023) Wind vector change and fire weather index in New Zealand as a modified metric in evaluating fire danger. International Journal of Wildland Fire 32, 872-885.
| Crossref | Google Scholar |

Burke M, Driscoll A, Heft-Neal S, Xue J, Burney J, Wara M (2021) The changing risk and burden of fire in the United States. Proceedings of the National Academy of Sciences 118(2), e2011048118.
| Crossref | Google Scholar | PubMed |

Canadell JG, Meyer CP, Cook GD, et al. (2021) Multi-decadal increase of forest burned area in Australia is linked to climate change. Nature Communications 12, 6921.
| Crossref | Google Scholar | PubMed |

Christ M, Braun N, Neuffer J, Kempa-Liehr AW (2018) Time series feature extraction on basis of scalable hypothesis tests (tsfresh – a Python package). Neurocomputing 307, 72-77.
| Crossref | Google Scholar |

Dempsey DE, Cronin SJ, Mei S, et al. (2020) Automatic precursor recognition and real-time forecasting of sudden explosive volcanic eruptions at Whakaari, New Zealand. Nature Communications 11, 3562.
| Crossref | Google Scholar | PubMed |

Dempsey DE, Kempa-Liehr AW, Ardid A, et al. (2022) Evaluation of short-term probabilistic eruption forecasting at Whakaari, New Zealand. Bulletin of Volcanology 84, 91.
| Crossref | Google Scholar |

Farahmand A, Stavros EN, Reager JT, Behrangi A, Randerson JT, Quayle B (2020) Satellite hydrology observations as operational indicators of forecasted fire danger across the contiguous United States. Natural Hazards and Earth System Sciences 20(4), 1097-1106.
| Crossref | Google Scholar |

Fox-Hughes P (2015) Characteristics of some days involving abrupt increases in fire danger. Journal of Applied Meteorology and Climatology 54(12), 2353-2363.
| Crossref | Google Scholar |

Fujioka FM, Gill AM, Viegas DX, Wotton BM (2008) Fire danger and fire behavior modeling systems in Australia, Europe, and North America. Developments in Environmental Science 8, 471-497.
| Google Scholar |

Ghali R, Jmal M, Souidene Mseddi W, Attia R (2020) Recent advances in fire detection and monitoring systems: a review. In ‘Proceedings of the 8th International Conference on Sciences of Electronics, Technologies of Information and Telecommunications (SETIT’18). Vol. 1’. pp. 332–340. (Springer International Publishing)

Gong A, Huang Z, Liu L, Yang Y, Ba W, Wang H (2023) Development of an index for Forest Fire Risk Assessment considering hazard factors and the hazard-formative environment. Remote Sensing 15, 5077.
| Crossref | Google Scholar |

Griffiths D (1999) Improved formula for the drought factor in McArthur’s Forest Fire Danger Meter. Australian Forestry 62(2), 202-206.
| Crossref | Google Scholar |

Hollis JJ, Matthews S, Anderson WR, Cruz MG, Fox-Hughes P, Grootemaat S, et al. (2024) A framework for defining fire danger to support fire management operations in Australia. International Journal of Wildland Fire 33(3), WF23141.
| Crossref | Google Scholar |

Hong H, Tsangaratos P, Ilia I, Liu J, Zhu AX, Xu C (2018) Applying genetic algorithms to set the optimal combination of forest fire related variables and model forest fire susceptibility based on data mining models. The case of Dayu County, China. Science of The Total Environment 630, 1044-1056.
| Crossref | Google Scholar | PubMed |

Huot F, Hu RL, Ihme M, Wang Q, Burge J, Lu T, et al. (2020) Deep learning models for predicting fires from historical remote-sensing data. arXiv:2010.07445 [cs]. Retrieved from http://arxiv.org/abs/2010.07445

Jung M, Tautenhahn S, Wirth C, Kattge J (2013) Estimating basal area of spruce and fir in post-fire residual stands in Central Siberia using Quickbird, Feature Selection, and Random Forests. Procedia Computer Science 18, 2386-2395.
| Crossref | Google Scholar |

Kaur P (2023) Forest fire prediction using heterogeneous data sources and machine learning methods. Master’s Thesis, University of Waterloo, Canada.

Klausmeyer KR, Shaw MR (2009) Climate change, habitat loss, protected areas and the climate adaptation danger of species in Mediterranean ecosystems worldwide. PLoS One 4(7), e6392.
| Crossref | Google Scholar | PubMed |

Kondylatos S, Prapas I, Ronco M, Papoutsis I, Camps-Valls G, Piles M, et al. (2022) Fire danger prediction and understanding with deep learning. Geophysical Research Letters 49, e2022GL099368.
| Crossref | Google Scholar |

Le HV, Hoang DA, Tran CT, Nguyen PQ, Tran VHT, Hoang ND, et al. (2021) A new approach of deep neural computing for spatial prediction of fire danger at tropical climate areas. Ecological Informatics 63, 101300.
| Crossref | Google Scholar |

Leuenberger M, Parente J, Tonini M, Pereira MG, Kanevski M (2018) Fire susceptibility mapping: deterministic vs. stochastic approaches. Environmental Modelling & Software 101, 194-203.
| Google Scholar |

McCoy L, Field D (2022) Lessons from NSW RFS trial of the Australian fire danger rating system. The Australian Journal of Emergency Management 37(4), 55-58.
| Google Scholar |

Matthews S, Fox-Hughes P, Grootemaat, S, Hollis JJ, Kenny BJ, Sauvage S (2019) ‘Australian Fire Danger Rating System: research prototype.’ p. 384. (NSW Rural Fire Service: Lidcombe, Australia)

McKnight PE, Najab J (2010) Mann‐Whitney U Test. In ‘The Corsini encyclopedia of psychology’. (Eds IB Weiner, WE Craighead) p. 1. (John Wiley & Sons, Ltd)

Mohapatra A, Trinh T (2022) Early fire detection technologies in practice—a review. Sustainability 14(19), 12270.
| Crossref | Google Scholar |

Müller MM, Vilà-Vilardell L, Vacik H (2020) Towards an integrated forest fire danger assessment system for the European Alps. Ecological Informatics 60, 101151.
| Crossref | Google Scholar |

Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Duchesnay É (2011) Scikit-learn: Machine learning in Python. The Journal of machine Learning research 12, 2825-2830.
| Google Scholar |

Pettinari ML, Chuvieco E (2020) Fire danger observed from space. Surveys in Geophysics 41(6), 1437-1459.
| Crossref | Google Scholar |

Qayyum F, Jamil H, Alsboui T, Hijjawi M (2024) RETRACTED ARTICLE: Wildfire risk exploration: leveraging SHAP and TabNet for precise factor analysis. Fire Ecology 20(1), 10.
| Crossref | Google Scholar |

Scott AC, Glasspool IJ (2006) The diversification of Paleozoic fire systems and fluctuations in atmospheric oxygen concentration. Proceedings of the National Academy of Sciences 103(29), 10861-10865.
| Crossref | Google Scholar | PubMed |

Sharples JJ, Mills GA, McRae RHD, Weber RO (2009). Fire danger anomalies associated with Foehn-like winds in southeastern Australia. In ‘Proceedings of the 18th World IMAC/MODSIM Congress’.

Si L, Shu L, Wang M, Zhao F, Chen F, Li W, Li W (2022) Study on forest fire danger prediction in plateau mountainous forest area. Natural Hazards Research 2(1), 25-32.
| Crossref | Google Scholar |

Sorensen JH (2000) Hazard warning systems: review of 20 years of progress. Natural Hazards Review 1(2), 119-125.
| Crossref | Google Scholar |

Stocks BJ, Lawson BD, Alexander ME, Wagner CV, McAlpine RS, Lynham TJ, Dube DE (1989) The Canadian Forest Fire Danger Rating system: an overview. The Forestry Chronicle 65(6), 450-457.
| Crossref | Google Scholar |

Van Wagner CE (1974) ‘ Structure of the Canadian Forest Fire Weather Index. Vol. 1333.’ (Ottawa, ON, Canada: Environment Canada, Forestry Service)

Wang Z, He B, Chen R, Fan C (2023) Improving fire danger assessment using time series features of weather and fuel in the Great Xing’an Mountain Region, China. Forests 14(5), 986.
| Crossref | Google Scholar |

Worsnop RP, Scheuerer M, Di Giuseppe F, Barnard C, Hamill TM, Vitolo C (2021) Probabilistic fire danger forecasting: a framework for week-two forecasts using statistical postprocessing techniques and the Global ECMWF Fire Forecast system (GEFF). Weather and Forecasting 36(6), 2113-2125.
| Crossref | Google Scholar |

Yebra M, Dennison PE, Chuvieco E, Riaño D, Zylstra P, Hunt ER, Jr, et al. (2013) A global review of remote sensing of live fuel moisture content for fire danger assessment: moving towards operational products. Remote Sensing of Environment 136, 455-468.
| Crossref | Google Scholar |

Yebra M, Quan X, Riaño D, Larraondo PR, van Dijk AI, Cary GJ (2018) A fuel moisture content and flammability monitoring methodology for continental Australia based on optical remote sensing. Remote Sensing of Environment 212, 260-272.
| Crossref | Google Scholar |

Yebra M, Scortechini G, Adeline K, Aktepe N, Almoustafa T, Bar-Massada A, et al. (2024) Globe-LFMC 2.0, an enhanced and updated dataset for live fuel moisture content research. Scientific Data 11(1), 332.
| Crossref | Google Scholar | PubMed |

Zhang G, Wang M, Liu K (2019) Forest fire susceptibility modeling using a convolutional neural network for Yunnan province of China. International Journal of Disaster Risk Science 10(3), 386-403.
| Crossref | Google Scholar |

Zheng Z, Gao Y, Yang Q, Zou B, Xu Y, Chen Y, Yang S, Wang Y, Wang Z (2020) Predicting forest fire risk based on mining rules with ant-miner algorithm in cloud-rich areas. Ecological Indicators 118, 106772.
| Crossref | Google Scholar |

Zhang G, Wang M, Liu K (2021) Deep neural networks for global fire susceptibility modelling. Ecological Indicators 127, 107735.
| Crossref | Google Scholar |