Integrating deep learning with physics-based model for predicting grassfire spread

Abstract

Shrublands and grasslands, which constitute approximately 70% of Australia’s vegetation, play a critical role in global wildfire-prone regions. To advance the understanding of grass fire spread, a three-dimensional, physics-based fire model provides valuable insights into fire dynamics. However, such models are computationally intensive and time-consuming. To address these challenges, we constructed an extensive numerical database comprising 64,000 high-fidelity wildfire simulation cases and implemented a Long Short-Term Memory neural network architecture. The model demonstrates strong predictive performance, achieving a coefficient of determination (R²) of 0.96 on training data, indicating excellent agreement with the physics-based simulation outputs. By utilizing coordinates from five reference points to predict fire front movement, this approach offers a novel method for analysing fire dynamics in homogeneous fuel beds with an average deviation of less than 2.5%. Combining the strengths of physics-based modelling and deep learning, our research enhances fire spread prediction accuracy of over 95% while significantly reducing computational demands. Future efforts will focus on refining the model, expanding the dataset, and incorporating additional variables to improve predictive capabilities and operational applicability.

Keywords

Fire propagation / Long short-term memory / Artificial intelligence (AI) / Numerical simulation / Fire dynamics behaviour

References

[1] Allaire F, Mallet V, Filippi JB. Emulation of wildland fire spread simulation using deep learning. Neural Netw, 2021, 141: 184-198.

[2] Bakhshaii A, Johnson EA. A review of a new generation of wildfire–atmosphere modeling. Can J for Res, 2019, 49(6): 565-574.

[3] Bartoli P, Simeoni A, Biteau H, Torero JL, Santoni PA. Determination of the main parameters influencing forest fuel combustion dynamics. Fire Saf J, 2011, 46(1–2): 27-33.

[4] Cheney P, Sullivan A. Grassfires: fuel, weather and fire behaviour. CSIRO Publ, 2008.

[5] Cheney NP, Gould JS, Catchpole WR. The influence of fuel, weather and fire shape variables on fire-spread in grasslands. Int J Wildland Fire, 1993, 3(1): 31.

[6] Chetehouna K, El Tabach E, Bouazaoui L, Gascoin N. Predicting the flame characteristics and rate of spread in fires propagating in a bed of Pinus pinaster using artificial neural networks. Process Saf Environ Prot, 2015, 98: 50-56.

[7] Coen J. Some requirements for simulating wildland fire behavior using insight from coupled weather: wildland fire models. Fire, 2018, 1(1): 6.

[8] Cruz MG, Alexander ME. Uncertainty associated with model predictions of surface and crown fire rates of spread. Environ Model Softw, 2013, 47: 16-28.

[9] Cruz MG, Gould JS, Alexander ME, Sullivan AL, McCaw WL, Matthews S. Empirical-based models for predicting head-fire rate of spread in Australian fuel types. Aust for, 2015, 78(3): 118-158.

[10] Cruz MG, Gould JS, Alexander ME, Sullivan AL, McCaw L, Matthews S (2017) A guide to rate of fire spread models for Australian vegetation. CSIRO Land and Water Flagship, Canberra, ACT and AFAC, Melbourne, VIC

[11] Dabrowski JJ, Pagendam DE, Hilton J, Sanderson C, MacKinlay D, Huston C, Bolt A, Kuhnert P. Bayesian physics informed neural networks for data assimilation and spatio-temporal modelling of wildfires. Spatial Stat, 2023, 55: 100746.

[12] Denham M, Wendt K, Bianchini G, Cortés A, Margalef T. Dynamic data-driven genetic algorithm for forest fire spread prediction. J Comput Sci, 2012, 3(5): 398-404.

[13] Eftekharian E, Ghodrat M, He YP, Ong RH, Kwok KCS, Zhao M. Numerical analysis of wind velocity effects on fire-wind enhancement. Int J Heat Fluid Flow, 2019, 80: 108471.

[14] Eftekharian E, Ghodrat M, He YP, Ong RH, Kwok KCS, Zhao M. Correlations for fire-wind enhancement flow characteristics based on LES simulations. Int J Heat Fluid Flow, 2020, 82: 108558.

[15] Finney MA (1998) FARSITE, Fire Area Simulator--model development and evaluation. US Department of Agriculture, Forest Service, Rocky Mountain Research Station

[16] Hilton JE, Miller C, Sharples JJ, Sullivan AL. Curvature effects in the dynamic propagation of wildfires. Int J Wildland Fire, 2016, 25(12): 1238.

[17] Hodges JL, Lattimer BY. Wildland fire spread modeling using convolutional neural networks. Fire Technol, 2019, 55(6): 2115-2142.

[18] Hodges JL, Lattimer BY, Luxbacher KD. Compartment fire predictions using transpose convolutional neural networks. Fire Saf J, 2019, 108: 102854.

[19] Huang XY, Wu XQ, Usmani A (2022) Perspectives of using artificial intelligence in building fire safety. In: Handbook of cognitive and autonomous systems for fire resilient infrastructures. Springer International Publishing, pp 139–159. https://doi.org/10.1007/978-3-030-98685-8_6

[20] Innocent J, Sutherland D, Moinuddin K. Field-scale physical modelling of grassfire propagation on sloped terrain under low-speed driving wind. Fire, 2023, 6(10): 406.

[21] Islam SMN, Thakur N, Garg K, Gupta A (2022) A recent survey on LSTM techniques for time-series data forecasting. In: Applications of artificial intelligence, big data and Internet of Things in sustainable development. CRC Press, pp 123–132. https://doi.org/10.1201/9781003245469-8

[22] Jackson W, Argent R, Bax N, Clark G, Coleman S, Cresswell I, Emmerson K, Evans K, Hibberd M, Johnston E. Australia state of the environment 2016. Indep Rep Aust Gov Minist Environ Energy, 2017.

[23] Jain P, Coogan SCP, Subramanian SG, Crowley M, Taylor S, Flannigan MD. A review of machine learning applications in wildfire science and management. Environ Rev, 2020, 28(4): 478-505.

[24] Jenkins MA, Sun R, Krueger SK, Charney JJ, Zulauf MA (2007) Effect of vertical wind shear on the grassfire evolution. Seventh symposium on fire and forest meteorology

[25] Khan N, Sutherland D, Wadhwani R, Moinuddin K. Physics-based simulation of heat load on structures for improving construction standards for bushfire prone areas. Front Mech Eng, 2019, 5: 35.

[26] Khanmohammadi S, Arashpour M, Golafshani EM, Cruz MG, Rajabifard A, Bai Y. Prediction of wildfire rate of spread in grasslands using machine learning methods. Environ Model Softw, 2022, 156: 105507.

[27] Lareau NP, Clements CB. Environmental controls on pyrocumulus and pyrocumulonimbus initiation and development. Atmos Chem Phys, 2016, 16(6): 4005-4022.

[28] Lareau NP, Clements CB. The mean and turbulent properties of a wildfire convective plume. J Appl Meteorol Climatol, 2017, 56(8): 2289-2299.

[29] Lautenberger C, Rein G, Fernandez-Pello C. The application of a genetic algorithm to estimate material properties for fire modeling from bench-scale fire test data. Fire Saf J, 2006, 41(3): 204-214.

[30] Li YZ, Wang ZL, Huang XY. Super real-time forecast of wildland fire spread by a dual-model deep learning method. J Environ Inform, 2024.

[31] Li YF, Wu H, Hou B, Liu T, Wang AS, Tu JZ, Zhang HT, Lv JS. A spatio-temporal attention–enhanced LSTM model for critical fault-set identification under wildfire conditions. IET Gener Transm Distrib, 2025, 19(1): e70138.

[32] Linn R, Reisner J, Colman JJ, Winterkamp J. Studying wildfire behavior using FIRETEC. Int J Wildland Fire, 2002, 11(4): 233.

[33] Linn R.R., Canfield J.M., Cunningham P., Edminster C., Dupuy J.-L., Pimont F.. Using periodic line fires to gain a new perspective on multi-dimensional aspects of forward fire spread. Agricultural and Forest Meteorology, 2012, 157: 60-76.

[34] Liu NA, Lei J, Gao W, Chen HX, Xie XD. Combustion dynamics of large-scale wildfires. Proc Combust Inst, 2021, 38(1): 157-198.

[35] McArthur AG (1966) Weather and grassland fire behaviour. Forestry and Timber Bureau, Department of National Development, Commonwealth of Australia

[36] McGrattan K, Hostikka S, Floyd J, McDermott R, Vanella M. (2022) Fire dynamics simulator (Sixth Edition) user’s guide

[37] McGrattan K, Hostikka S, Floyd J, McDermott R, Vanella M (2022) Fire dynamics simulator technical reference guide volume 1: mathematical model. U.S. Department of Commerce

[38] Mell W, Jenkins MA, Gould J, Cheney P. A physics-based approach to modelling grassland fires. Int J Wildland Fire, 2007, 16(1): 1.

[39] Mell W, Charney J, Jenkins MA, Cheney P, Gould J (2013) Numerical simulations of grassland fire behavior from the LANL-FIRETEC and NIST-WFDS models. In: Remote sensing and modeling applications to wildland fires. Springer, pp 209–225. https://doi.org/10.1007/978-3-642-32530-4_15

[40] Moinuddin KAM, Sutherland D, Mell W. Simulation study of grass fire using a physics-based model: striving towards numerical rigour and the effect of grass height on the rate of spread. Int J Wildland Fire, 2018, 27(12): 800.

[41] Morvan D, Hoffman C, Rego F, Mell W. Numerical simulation of the interaction between two fire fronts in grassland and shrubland. Fire Saf J, 2011, 46(8): 469-479.

[42] Morvan D (2020) Validation of wildfire spread models. In: Encyclopedia of wildfires and wildland-urban interface (WUI) fires. Springer International Publishing, pp 1031–1037. https://doi.org/10.1007/978-3-319-52090-2_59

[43] Naser MZ. Mechanistically informed machine learning and artificial intelligence in fire engineering and sciences. Fire Technol, 2021, 57(6): 2741-2784.

[44] Perez-Ramirez Y, Mell WE, Santoni PA, Tramoni JB, Bosseur F. Examination of WFDS in modeling spreading fires in a furniture calorimeter. Fire Technol, 2017, 53(5): 1795-1832.

[45] Ronchi E, Gwynne SMV, Rein G, Intini P, Wadhwani R. An open multi-physics framework for modelling wildland-urban interface fire evacuations. Saf Sci, 2019, 118: 868-880.

[46] Ronchi E, Gwynne S, Rein G, Wadhwani R, Intini P, Bergstedt A (2017) e-Sanctuary: Open Multi-Physics Framework for Modelling Wildfire Urban Evacuation (FPRF-2017-22). Fire Protection Research Foundation Quincy, MA

[47] Rothermel RC, Deeming JE (1980) Measuring and interpreting fire behavior for correlation with fire effects. Intermoutain Forest and Range Experiment Station

[48] Rothermel RC (1972) A mathematical model for predicting fire spread in wildland fuels. Res. Pap. INT-115. Ogden, UT: U.S. Department of Agriculture, Intermountain Forest and Range Experiment Station, p 40

[49] Shinohara M. Effects of wind speed and heat release rate on the vortex strength and size of fire whirls without an inner core of flame. Fire Saf J, 2021, 120: 103045.

[50] Singh H, Ang LM, Lewis T, Paudyal D, Acuna M, Srivastava PK, Srivastava SK. Trending and emerging prospects of physics-based and ML-based wildfire spread models: a comprehensive review. J for Res, 2024, 35(1): 135.

[51] Sullivan AL. Convective froude number and Byram’s energy criterion of Australian experimental grassland fires. Proc Combust Inst, 2007, 31(2): 2557-2564.

[52] Sullivan AL. Wildland surface fire spread modelling, 1990–2007.1: physical and quasi-physical models. Int J Wildland Fire, 2009, 18(4): 349.

[53] Sullivan AL. Inside the inferno: fundamental processes of wildland fire behaviour: part 1: combustion chemistry and heat release. Curr Forestry Rep, 2017, 3(2): 132-149.

[54] Sullivan AL. Inside the inferno: fundamental processes of wildland fire behaviour. Curr for Rep, 2017, 3(2): 150-171.

[55] Tolhurst K, Shields B, Chong D. Phoenix: development and application of a bushfire risk management tool. Aust J Emerg Manag, 2008, 23(4): 47-54

[56] Vakalis D, Sarimveis H, Kiranoudis C, Alexandridis A, Bafas G. A GIS based operational system for wildland fire crisis management i. mathematical modelling and simulation. Appl Math Model, 2004, 28(4): 389-410.

[57] Valero MM, Jofre L, Torres R. Multifidelity prediction in wildfire spread simulation: modeling, uncertainty quantification and sensitivity analysis. Environ Model Softw, 2021, 141: 105050.

[58] Vanella M, McGrattan K, McDermott R, Forney G, Mell W, Gissi E, Fiorucci P. A multi-fidelity framework for wildland fire behavior simulations over complex terrain. Atmosphere, 2021, 12(2): 273.

[59] Wu XQ, Zhang XN, Huang XY, Xiao F, Usmani A. A real-time forecast of tunnel fire based on numerical database and artificial intelligence. Build Simul, 2022, 15(4): 511-524.

[60] Zeng YF, Zhang XN, Su LC, Wu XQ, Huang XY. Artificial intelligence tool for fire safety design (IFETool): demonstration in large open spaces. Case Stud Therm Eng, 2022, 40: 102483.

[61] Zhang XN, Wu XQ, Huang XY. Smart real-time forecast of transient tunnel fires by a dual-agent deep learning model. Tunn Undergr Space Technol, 2022, 129: 104631.

Funding

The Hong Kong Polytechnic University

RIGHTS & PERMISSIONS

The Author(s)