Prediction and driving factors of forest fire occurrence in Jilin Province, China

Abstract

Forest fires are natural disasters that can occur suddenly and can be very damaging, burning thousands of square kilometers. Prevention is better than suppression and prediction models of forest fire occurrence have developed from the logistic regression model, the geographical weighted logistic regression model, the Lasso regression model, the random forest model, and the support vector machine model based on historical forest fire data from 2000 to 2019 in Jilin Province. The models, along with a distribution map are presented in this paper to provide a theoretical basis for forest fire management in this area. Existing studies show that the prediction accuracies of the two machine learning models are higher than those of the three generalized linear regression models. The accuracies of the random forest model, the support vector machine model, geographical weighted logistic regression model, the Lasso regression model, and logistic model were 88.7%, 87.7%, 86.0%, 85.0% and 84.6%, respectively. Weather is the main factor affecting forest fires, while the impacts of topography factors, human and social-economic factors on fire occurrence were similar.

References

AndrewsPL. Current status and future needs of the behaveplus fire modeling system. Int J Wildland Fire, 2014, 23(1): 21-33

AndrewsPL. The Rothermel surface fire spread model and associated developments: a comprehensive explanation. Gen Tech Rep, 2018, 121: 371

AndrewsPL, LoftsgaardenDO, BradshawLS. Evaluation of fire danger rating indexes using logistic regression and percentile analysis. Int J Wildland Fire, 2003, 12(2): 213-226

BisquertM, CasellesE, SánchezJM, CasellesV. Application of artificial neural networks and logistic regression to the prediction of forest fire danger in Galicia using MODIS data. Int J Wildland Fire, 2012, 21(8): 1025-1029

BoubetaM, LombardíaMJ, Marey-PérezM, MoralesD. Poisson mixed models for predicting number of fires. Int J Wildland Fire, 2019, 28(3): 237-253

BreimanL. Random forests. Mach Learn, 2001, 45: 5-32

CardilleJA, VenturaSJ, TurnerMG. Environmental and social factors influencing wildfires in the Upper Midwest. United States Ecol Appl, 2001, 11(1): 111-127

CatryFX, RegoFC, BaçãoFL, MoreiraF. Modeling and mapping wildfire ignition risk in Portugal. Int J Wildland Fire, 2009, 18(8): 921-931

CawsonJG, DuffTJ. Forest fuel bed ignitability under marginal fire weather conditions in Eucalyptus forests. Int J Wildland Fire, 2019, 28(3): 198-204

CortesC, VapnikV. Support-vector networks. Mach Learn, 1995, 20: 273-297

CoughlanR, Di GiuseppeF, VitoloC, BarnardC, LopezP, DruschM. Using machine learning to predict fire-ignition occurrences from lightning forecasts. Meteorol Appl, 2021, 28(1): e1973

Díaz-AvalosC, PetersonDL, AlvaradoE, FergusonSA, BesagJE. Space time modelling of lightning-caused ignitions in the Blue Mountains. Oregon Can J Forest Res, 2001, 31(9): 1579-1593

EliaM, D'EsteM, AscoliD, GiannicoV, SpanoG, GangaA, ColangeloG, LafortezzaR, SanesiG. Estimating the probability of wildfire occurrence in Mediterranean landscapes using artificial neural networks. Environ Impact Assess Rev, 2020, 85: 106474

EslamiR, AzarnoushM, KialashkiA, KazemzadehF. Gis-based forest fire susceptibility assessment by random forest, artificial neural network and logistic regression methods. J Trop for Sci, 2021, 33(2): 173-184

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

GhoshD, ChinnaiyanAM. Classification and selection of biomarkers in genomic data using LASSO. J Biomed Biotechnol, 2005, 2: 147

GibsonRK, BradstockRA, PenmanT, KeithDA, DriscollDA. Climatic, vegetation and edaphic influences on the probability of fire across mediterranean woodlands of south-eastern Australia. J Biogeogr, 2015, 42(9): 1750-1760

GigovićL, PourghasemiHR, DrobnjakS, BaiS. Testing a new ensemble model based on SVM and random forest in forest fire susceptibility assessment and its mapping in Serbia’s Tara National park. Forests, 2019, 10(5): 408

GuoF, SuZ, WangG, SunL, LinF, LiuA. Wildfire ignition in the forests of southeast China: Identifying drivers and spatial distribution to predict wildfire likelihood. Appl Geogr, 2016, 66: 12-21

GuoF, WangG, SuZ, LiangH, WangW, LinF, LiuA. What drives forest fire in Fujian, China? Evidence from logistic regression and random forests. Int J Wildland Fire, 2016, 25(5): 505-519

GuoF, SuZ, WangG, SunL, TigabuM, YangX, HuH. Understanding fire drivers and relative impacts in different Chinese forest ecosystems. Sci Total Environ, 2017, 605: 411-425

GuoE, ZhangJ, WangY, AluS, WangR, LiD, HaS. Assessing non-linear variation of temperature and precipitation for different growth periods of maize and their impacts on phenology in the Midwest of Jilin Province. China Theor Appl Climtol, 2018, 132(3): 685-699

HardyCC, HardyCE. Fire danger rating in the United States of America: an evolution since 1916. Int J Wildland Fire, 2007, 16(2): 217-231

JollyWM, CochraneMA, FreebornPH, HoldenZA, BrownTJ, WilliamsonGJ, BowmanDM. Climate-induced variations in global wildfire danger from 1979 to 2013. Nat Commun, 2015, 6(1): 1-11

KeeleyJE, SyphardAD. Large California wildfires: 2020 fires in historical context. Fire Ecol, 2021, 17(1): 1-11

KimSJ, LimCH, KimGS, LeeJ, GeigerT, RahmatiO, SonY, LeeWK. Multi-temporal analysis of forest fire probability using socio-economic and environmental variables. Remote Sens, 2019, 11(1): 86

KreyeJK, HiersJK, VarnerJM, HornsbyB, DrukkerS, O’brien JJ, . Effects of solar heating on the moisture dynamics of forest floor litter in humid environments: composition, structure, and position matter. Can J Forest Res, 2018, 48(11): 1331-1342

LiY, WangYF, SunYJ, LeiYC, ShaoWC, LiJ. Temporal-spatial characteristics of NPP and its response to climate change of Larix forests in Jilin Province. Acta Ecol Sin, 2022, 42: 947-959

LuB, HarrisP, CharltonM, BrunsdonC. The GWmodel R package: further topics for exploring spatial heterogeneity using geographically weighted models. Geo-Spatial Inf Sci, 2014, 17(2): 85-101

MaW, FengZ, ChengZ, ChenS, WangF. Identifying forest fire driving factors and related impacts in China using random forest algorithm. Forests, 2020, 11(5): 507

Martínez-FernándezJ, ChuviecoE, KoutsiasN. Modelling long-term fire occurrence factors in Spain by accounting for local variations with geographically weighted regression. Nat Hazards Earth Sys, 2013, 13(2): 311-327

MilanovićS, MarkovićN, PamučarD, GigovićL, KostićP, MilanovićSD. Forest fire probability mapping in eastern Serbia: logistic regression versus random forest method. Forests, 2020, 12(1): 5

Miranda-AragónL, Treviño-GarzaE, Jiménez-PérezJ, Aguirre-CalderónO, González-TagleM, Pompa-GarcíaM, Aguirre-SaladoC. Modeling susceptibility to deforestation of remaining ecosystems in North Central Mexico with logistic regression. J Forestry Res, 2012, 23(3): 345-354

MoznyM, TrnkaM, BrázdilR. Climate change driven changes of vegetation fires in the Czech Republic. Theor Appl Clim, 2021, 143(1): 691-699

NadeemK, TaylorS, WoolfordDG, DeanC. Mesoscale spatiotemporal predictive models of daily human-and lightning-caused wildland fire occurrence in British Columbia. Int J Wildland Fire, 2019, 29(1): 11-27

NurdiatiS, SopaheluwakanA, JuliantoMT, SeptiawanP, RohimahastutiF. Modelling and analysis impact of El Nino and IOD to land and forest fire using polynomial and generalized logistic function: cases study in South Sumatra and Kalimantan. Indonesia Model Earth Syst Environ, 2022, 8(3): 3341-3356

NymanP, MetzenD, NoskePJ, LanePN, SheridanGJ. Quantifying the effects of topographic aspect on water content and temperature in fine surface fuel. Int J Wildland Fire, 2015, 24(8): 1129-1142

OrdóñezC, SaavedraA, Rodríguez-PérezJR, Castedo-DoradoF, CoviánE. Using model-based geostatistics to predict lightning-caused wildfires. Environ Model Softw, 2012, 29(1): 44-50

PetersonD, WangJ, IchokuC, RemerL. Effects of lightning and other meteorological factors on fire activity in the North American boreal forest: implications for fire weather forecasting. Atmos Chem Phys, 2010, 10(14): 6873-6888

PhamBT, JaafariA, AvandM, Al-AnsariN, Du DinhT, YenHPH, PhongTV, NguyenDH, LeHV, Mafi-GholamiD. Performance evaluation of machine learning methods for forest fire modeling and prediction. Symmetry, 2020, 12(6): 1022

PhelpsN, WoolfordDG. Guidelines for effective evaluation and comparison of wildland fire occurrence prediction models. Int J Wildland Fire, 2021, 30(4): 225-240

RanstamJ, CookJ. LASSO regression. Br J Surg, 2018, 105(10): 1348-1348

RijalB. Quantile regression: an alternative approach to modelling forest area burned by individual fires. Int J Wildland Fire, 2018, 27(8): 538-549

RodriguesM, de la RivaJ, FotheringhamS. Modeling the spatial variation of the explanatory factors of human-caused wildfires in Spain using geographically weighted logistic regression. Appl Geogr, 2014, 48: 52-63

RodriguesM, Costafreda-AumedesS, ComasC, Vega-GarcíaC. Spatial stratification of wildfire drivers towards enhanced definition of large-fire regime zoning and fire seasons. Sci Total Environ, 2019, 689: 634-644

RudinC. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat Mach Intell, 2019, 1(5): 206-215

Senande-RiveraM, Insua-CostaD, Miguez-MachoG. Spatial and temporal expansion of global wildland fire activity in response to climate change. Nat Commun, 2022, 13(1): 1-9

SharmaLK, GuptaR, FatimaN. Assessing the predictive efficacy of six machine learning algorithms for the susceptibility of Indian forests to fire. Int J Wildland Fire, 2022, 31(8): 735-758

StroblR, GrillE, MansmannU. Graphical modeling of binary data using the LASSO: a simulation study. BMC Med Res Method, 2012, 12(1): 1-13

ŠturmT, PodobnikarT. A probability model for long-term forest fire occurrence in the Karst forest management area of Slovenia. Int J Wildland Fire, 2017, 26(5): 399-412

SuZ, ZhengL, LuoS, TigabuM, GuoF. Modeling wildfire drivers in Chinese tropical forest ecosystems using global logistic regression and geographically weighted logistic regression. Nat Hazard, 2021, 108(1): 1317-1345

TaylorSW, AlexanderME. Science, technology, and human factors in fire danger rating: the Canadian experience. Int J Wildland Fire, 2006, 15(1): 121-135

TibshiraniR, SaundersM, RossetS, ZhuJ, KnightK. Sparsity and smoothness via the fused lasso. J R Stat Soc Ser B Stat Methodol, 2005, 67(1): 91-108

van der VeldeIR, van der WerfGR, HouwelingS, MaasakkersJD, BorsdorffT, LandgrafJ, TolP, van KempenTA, van HeesR, HoogeveenR. Vast CO₂ release from Australian fires in 2019–2020 constrained by satellite. Nature, 2021, 597(7876): 366-369

WangZ, LaiC, ChenX, YangB, ZhaoS, BaiX. Flood hazard risk assessment model based on random forest. J Hydrol, 2015, 527: 1130-1141

YangG, DiXY, ZengT, ShuZ, WangC, YuHZ. Prediction of area burned under climatic change scenarios: a case study in the Great Xing’an Mountains boreal forest. J Forestry Res, 2010, 21(2): 213-218

ZhuY, YuQ, LuoQ, ZhangH, ZhaoJ, JuZ, DuY, YangY. Impacts of climate change on suitability zonation for potato cultivation in Jilin Province. Northeast China Sci Rep, 2021, 11(1): 1-14

Funding

This research was funded by the National Natural Science Foundation of China(32271881)