基于机器学习算法的森林火灾风险评估研究

doi:10.12302/j.issn.1000-2006.202202004

国家林草科技领军期刊
中国精品科技期刊
中国高校百佳科技期刊
江苏省新闻出版政府奖期刊奖
RCCSE林学权威期刊（A+）
CSCD核心期刊
Scopus数据库收录期刊
中文核心期刊
SCD核心期刊

作者加群：102861116

微信公众号：南京林业大学学报

高级检索

南京林业大学学报（自然科学版） ›› 2023, Vol. 47 ›› Issue (5): 49-56.doi: 10.12302/j.issn.1000-2006.202202004

所属专题：林草计算机应用研究专题

• 专题报道：林草计算机应用研究专题（执行主编李凤日） • 上一篇下一篇

基于机器学习算法的森林火灾风险评估研究

李史欣(), 张福全^*(), 林海峰

南京林业大学信息科学技术学院,江苏南京 210037

收稿日期:2022-02-11 修回日期:2022-07-12 出版日期:2023-09-30 发布日期:2023-10-10
作者简介:李史欣(2501972185@qq.com)。
基金资助:
江苏省重点研发计划(BE2021716)

Research on forest fire risk evaluation based on machine learning algorithm

LI Shixin(), ZHANG Fuquan^*(), LIN Haifeng

College of Information Science and Technology,Nanjing Forestry University,Nanjing 210037,China

Received:2022-02-11 Revised:2022-07-12 Online:2023-09-30 Published:2023-10-10

摘要/Abstract

摘要：

【目的】利用森林火灾风险图可提高有效巡护,优化有限防火资源,基于地形、人类活动、植被和气象因素数据,采用基于机器学习算法构建了林火发生预测模型,对林火防护提供一定的参考。【方法】以安徽省滁州韭山为研究对象,提取林区的坡度、海拔、坡向、到居住点的距离、到道路的距离、地形湿度指数、归一化植被指数和温度驱动因素,评估火灾发生驱动因子,将潜在驱动因子分成地形、人类活动、植被与气象因素等4类;使用哨兵火灾产品,提取林区内的历史火点,然后采用机器学习算法建立林火发生的预测模型;最后利用混淆矩阵评估指标和接受者操作特征曲线(ROC)进行精度评价。【结果】植被、温度和到道路的距离是研究区域火灾发生的主要驱动因素。两种模型的ROC曲线表明,逻辑回归预测模型准确度为71.07%,曲线下面积(AUC)值为0.717 2;随机森林模型具有较好的准确性,准确度达到84.91%,曲线下面积值为0.850 1。【结论】随机森林模型表现出比逻辑回归模型更好的预测能力。森林火灾风险图表明,随机森林模型预测下,研究区11.91% (29.36 km²)位于高、极高风险级别。森林火灾风险图可有效协助林火防护管理者采取适当的林火预防措施,保护森林资源。

关键词: 火灾风险, 机器学习, 预测模型, 森林植被, 气象因素

Abstract:

【Objective】 Forest fire risk maps are necessary to improve effective patrols and optimize the scientific layout of limited fire prevention resources. This study uses machine learning algorithms to construct forest fire occurrences based on terrain, human activities, vegetation, and meteorological factor data.【Method】With Jiushan Mountain in Chuzhou city in Anhui province as the research object, we extracted the following potential driving factors: slope, elevation, aspect, distance to settlement, distance to road, topographic wetness index, normalized difference vegetation index and temperature of the study area, evaluated the driving factors of fire occurrence, and then divided the potential driving factors into topography, human activities, vegetation, meteorological factors, and other four categories. Historical fire points in the forest area were extracted from the sentinel fire products. A prediction model for forest fire occurrence was then constructed using a machine learning algorithm. Finally, the accuracy of the models was evaluated using a confusion matrix and receiver operating characteristic curves. 【Result】We found that vegetation, temperature and distance to the road are the main driving factors of forest fires in the study area. The ROC curves of the two models showed that the Logistic regression prediction model had an accuracy of 71.07%, where the area under the curve was 0.717 2. Meanwhile, the random forest model had a better accuracy, with an accuracy of 84.91% and an area under the curve of 0.850 1.【Conclusion】The random forest model exhibits a better predictive ability than that does the logistic regression model. Furthermore, the generated forest fire risk map shows that 11.91% (29.36 km²) of the study area is at a high or extremely a high risk. If utilized, this forest fire risk map can effectively help forest fire protection managers implement appropriate measures to protect forest resources in Jiushan Mountain.

Key words: fire risk, machine learning, forecasting model, forest vegetation, meteorological factors

中图分类号:

S771.8

李史欣,张福全,林海峰. 基于机器学习算法的森林火灾风险评估研究[J]. 南京林业大学学报（自然科学版）, 2023, 47(5): 49-56.

LI Shixin, ZHANG Fuquan, LIN Haifeng. Research on forest fire risk evaluation based on machine learning algorithm[J].Journal of Nanjing Forestry University (Natural Science Edition), 2023, 47(5): 49-56.DOI: 10.12302/j.issn.1000-2006.202202004.

图/表 8

图1

表1

图2

表2

表3

图3

图4

图5

参考文献 33

[1]	MORENO M V, CHUVIECO E. Characterising fire regimes in Spain from fire statistics[J]. Int J Wildland Fire, 2013, 22(3):296.DOI: 10.1071/wf12061.
[2]	HUSSIN Y A, MATAKALA M, ZAGDAA N. The applications of remote sensing and GIS in modeling forest fire hazard in Mongolia[J]. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2008(5):289-294.
[3]	GUO F T, WANG G Y, SU Z W, et al. What drives forest fire in Fujian,China? evidence from Logistic regression and Random Forests[J]. Int J Wildland Fire, 2016, 25(5):505.DOI: 10.1071/wf15121.
[4]	ESTES B L, KNAPP E E, SKINNER C N, et al. Factors influencing fire severity under moderate burning conditions in the Klamath Mountains,northern California,USA[J]. Ecosphere, 2017, 8(5):e01794.DOI: 10.1002/ecs2.1794.
[5]	ROGEAU M P, ARMSTRONG G W. Quantifying the effect of elevation and aspect on fire return intervals in the Canadian Rocky Mountains[J]. For Ecol Manag, 2017, 384:248-261.DOI: 10.1016/j.foreco.2016.10.035.
[6]	SCHILLACI C, BRAUN A, KROPÁCEK J. Terrain analysis and landform recognition[M]. Geomorphological Techniques,London: British Society for Geomorphology, 2015.
[7]	KRUEGER E S, OCHSNER T E, ENGLE D M, et al. Soil moisture affects growing-season wildfire size in the southern great plains[J]. Soil Sci Soc Am J, 2015, 79(6):1567-1576.DOI: 10.2136/sssaj2015.01.0041.
[8]	RICOTTA C, BAJOCCO S, GUGLIETTA D, et al. Assessing the influence of roads on fire ignition:does land cover matter?[J]. Fire, 2018, 1(2):24.DOI: 10.3390/fire1020024.
[9]	KALANTAR B, UEDA N, IDREES M O, et al. Forest fire susceptibility prediction based on machine learning models with resampling algorithms on remote sensing data[J]. Remote Sens, 2020, 12(22):3682.DOI: 10.3390/rs12223682.
[10]	周振伟, 于成龙. 层次分析法在森林火险预测中的应用[J]. 黑龙江气象, 2008, 25(增刊1):17-19.
	ZHOU Z W, YU C L. Application of analytic hierarchy process in forest fire risk prediction[J]. Heilongjiang Meteorol, 2008, 25(S1):17-19.
[11]	许志卿, 苏喜友, 张颐. 基于支持向量机方法的森林火险预测研究[J]. 中国农学通报, 2012, 28(13):126-131.
	XU Z Q, SU X Y, ZHANG Y. Forest fire prediction based on support vector machine[J]. Chin Agric Sci Bull, 2012, 28(13):126-131.DOI: 10.3969/j.issn.1000-6850.2012.13.023.
[12]	崔亮, 张继权, 包玉龙, 等. 呼伦贝尔草原火灾风险预警研究[J]. 草业学报, 2012, 21(4):282-292.
	CUI L, ZHANG J Q, BAO Y L, et al. A study on early warning of grassland fire disaster risk in Hulunbeier[J]. Acta Prataculturae Sin, 2012, 21(4):282-292.DOI: 10.11686/cyxb20120434.
[13]	MALLINIS G, PETRILA M, MITSOPOULOS I, et al. Geospatial patterns and drivers of forest fire occurrence in Romania[J]. Appl Spatial Analysis, 2019, 12(4):773-795.DOI: 10.1007/s12061-018-9269-3.
[14]	高超, 林红蕾, 胡海清, 等. 我国林火发生预测模型研究进展[J]. 应用生态学报, 2020, 31(9):3227-3240. doi: 10.13287/j.1001-9332.202009.014
	GAO C, LIN H L, HU H Q, et al. A review of models of forest fire occurrence prediction in China[J]. Chin J Appl Ecol, 2020, 31(9):3227-3240.DOI: 10.13287/j.1001-9332.202009.014.
[15]	OLIVEIRA S, OEHLER F, SAN-MIGUEL-AYANZ J, et al. Modeling spatial patterns of fire occurrence in Mediterranean Europe using Multiple Regression and Random Forest[J]. For Ecol Manag, 2012, 275:117-129.DOI: 10.1016/j.foreco.2012.03.003.
[16]	贾南, 陈悦, 康可霖, 等. 基于RF的森林火灾风险评价模型及其应用研究[J]. 安全与环境学报, 2020, 20(4):1236-1240.
	JIA N, CHEN Y, KANG K L, et al. Improved forest fire risk assessment model and its application based on the RF algorithm[J]. J Saf Environ, 2020, 20(4):1236-1240.DOI: 10.13637/j.issn.1009-6094.2019.0899.
[17]	王雪娟, 张雪平. 韭山国家森林公园观赏植物资源及其保护与发展对策[J]. 安徽科技学院学报, 2010, 24(5):5-8.
	WANG X J, ZHANG X P. Protection and development strategy of ornamental plant resources in Jiushan National Forest Park[J]. J Anhui Sci Technol Univ, 2010, 24(5):5-8.DOI: 10.3969/j.issn.1673-8772.2010.05.002.
[18]	王佳辰, 冯雨馨, 郝志尚, 等. 安徽省森林火灾预测及分析[J]. 大众标准化, 2021(13):198-200.
	WANG J C, FENG Y X, HAO Z S, et al. Forecast and ana-lysis of forest fire in Anhui Province[J]. Pop Stand, 2021(13):198-200.DOI: 10.3969/j.issn.1007-1350.2021.13.067.
[19]	杨春梅, 侯玉宁, 刘九庆. 基于森林火灾数据的余火清理机器人模块化设计[J]. 森林工程, 2022, 38(2):105-111.
	YANG C M, HOU Y N, LIU J Q. Modular design of embers clearing robot based on forest fire data[J]. Forest Engineering, 2022, 38(2):105-111.
[20]	MA W Y, FENG Z K, CHENG Z X, et al. Identifying forest fire driving factors and related impacts in China using random forest algorithm[J]. Forests, 2020, 11(5):507.DOI: 10.3390/f11050507.
[21]	MCDONNELL R A, LLOYD C, BURROUGH P. Principles of geographical information systems[M].3rd ed. London: Oxford University Press, 2015.
[22]	NOVO A, LORENZO H, LORENZO H, et al. Mapping forest fire risk-a case study in Galicia (Spain)[J]. Remote Sensing (Basel, Switzerland), 2020, 12(22): 3705.
[23]	SETIAWAN I, MAHMUD A R, MANSOR S, et al. GIS-grid-based and multi-criteria analysis for identifying and mapping peat swamp forest fire hazard in Pahang,Malaysia[J]. Disaster Prev And Management, 2004, 13(5):379-386.DOI: 10.1108/09653560410568507.
[24]	NIKHIL S, DANUMAH J H, SAHA S, et al. Application of GIS and AHP method in forest fire risk zone mapping:a study of the parambikulam tiger reserve,Kerala,India[J]. J Geovis Spat Anal, 2021, 5(1):14.DOI: 10.1007/s41651-021-00082-x.
[25]	EUGENIO F C, DOS SANTOS A R, FIEDLER N C, et al. Applying GIS to develop a model for forest fire risk:a case study in Espírito Santo,Brazil[J]. J Environ Manag, 2016, 173:65-71.DOI: 10.1016/j.jenvman.2016.02.021.
[26]	POURTAGHI Z S, POURGHASEMI H R, ARETANO R, et al. Investigation of general indicators influencing on forest fire and its susceptibility modeling using different data mining techniques[J]. Ecol Indic, 2016, 64:72-84.DOI: 10.1016/j.ecolind.2015.12.030.
[27]	GHESHLAGHI H A. Using GIS to develop a model for forest fire risk mapping[J]. J Indian Soc Remote Sens, 2019, 47(7):1173-1185. doi: 10.1007/s12524-019-00981-z
[28]	ZHAO P C, ZHANG F Q, LIN H F, et al. GIS-based forest fire risk model:a case study in Laoshan National Forest Park,Nanjing[J]. Remote Sens, 2021, 13(18):3704.DOI: 10.3390/rs13183704.
[19]	DIGAVINTI J, MANIKIAM B. Satellite monitoring of forest fire impact and regeneration using NDVI and LST[J]. J Appl Rem Sens, 2021, 15(4):042412.DOI: 10.1117/1.jrs.15.042412.
[30]	GARCÍA M J L, CASELLES V. Mapping burns and natural reforestation using thematic Mapper data[J]. Geocarto Int, 1991, 6(1):31-37.DOI: 10.1080/10106049109354290.
[31]	顾先丽, 吴志伟, 张宇婧, 等. 气候变化背景下江西省林火空间预测[J]. 生态学报, 2020, 40(2):667-677.
	GU X L, WU Z W, ZHANG Y J, et al. Prediction research of the forest fire in Jiangxi Province in the background of climate change[J]. Acta Ecol Sin, 2020, 40(2):667-677.DOI: 10.5846/stxb201901030022.
[32]	TIEN BUI D, LE K T, NGUYEN V, et al. Tropical forest fire susceptibility mapping at the cat Ba National Park area,Hai phong city,Vietnam,using GIS-based kernel logistic regression[J]. Remote Sens, 2016, 8(4):347.DOI: 10.3390/rs8040347.
[33]	GOLEIJI E, HOSSEINI S M, KHORASANI N, et al. Forest fire risk assessment-an integrated approach based on multicriteria evaluation[J]. Environ Monit Assess, 2017, 189(12):612.DOI: 10.1007/s10661-017-6225-7. pmid: 29110111

变量类型 variable type	变量名 variable name	符号 symbol	数据来源 data sources
地形 terrain	坡度/(°) slope	s	DEM http://www. gscloud.cn
	坡向 aspect	a
	地形湿度指数 TWI	I_TW
	海拔/m altitude	h
人类活动 human activity	到道路的距离/m DTR	d_TR	谷歌地图 google map
人类活动 human activity	到居民点的距离/m DTS	d_TS
植被 vegetation	归一化植被指数NDVI	I_NDV	Landsat-8 OLI http://www.gscloud.cn
气象 meteorological	温度/℃ temperature	T

相关因素 relevant factors	R	P
坡度slope	-0.267	<0.001
坡向aspect	-0.031	<0.05
地形湿度指数TWI	0.070	<0.05
海拔altitude	-0.323	<0.001
归一化植被指数NDVI	-0.357	<0.001
到道路的距离DTR	-0.401	<0.001
到居民点的距离DTS	0.169	<0.001
温度temperature	-0.286	<0.001

参数 parameter	逻辑回归 LR	随机森林 FR
特异性/% specificity	62.35	83.53
灵敏度/% sensitivity	81.08	86.49
阳性预测值/% positive predictive value	65.22	82.05
阴性预测值/% negative predictive value	79.10	87.65
总体准确度/% accuracy total	71.07	84.91

基于机器学习算法的森林火灾风险评估研究

Research on forest fire risk evaluation based on machine learning algorithm

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 33

相关文章 15

编辑推荐

Metrics

本文评价

作者

读者

审稿

[1]	王琬钧, 于颖, 杨曦光. 基于叶绿素荧光的叶绿素含量估算[J]. 南京林业大学学报（自然科学版）, 2024, 48(6): 157-165.
[2]	李慧, 张婉, 常益豪, 杨霞, 肖祥伟, 朱景乐. 幼龄栓皮栎叶面积预测模型建立及应用[J]. 南京林业大学学报（自然科学版）, 2024, 48(5): 246-254.
[3]	赵凌霄, 李智扬, 屈磊磊. 基于EMD和CatBoost算法的改进时间序列模型——以大连市PM_2.5预测为例[J]. 南京林业大学学报（自然科学版）, 2024, 48(3): 268-274.
[4]	陈建坤, 牟凤云, 张用川, 田甜, 王俊秀. 基于多机器学习模型的逐小时PM_2.5浓度预测对比[J]. 南京林业大学学报（自然科学版）, 2022, 46(5): 152-160.
[5]	郭天威, 陆春锋, 王君櫹, 刘瑞程, 周生路. 基于三生空间耦合的生态安全格局构建与优化——以扬州市为例[J]. 南京林业大学学报（自然科学版）, 2021, 45(5): 133-142.
[6]	郝泉龄, 徐国祺, 王立海, 时小龙, 许明贤, 纪文文, 张广晖. 基于Logistic回归模型的红松立木腐朽分级预测[J]. 南京林业大学学报（自然科学版）, 2020, 44(2): 150-158.
[7]	贾炜玮, 梁玉钊, 李凤日. 落叶松人工林树皮厚度预测模型[J]. 南京林业大学学报（自然科学版）, 2019, 43(6): 97-104.
[8]	刘智丽, 岳德鹏, 杨慧, 张启斌, 于强, 徐小东. 基于城市化进程的宜兴市景观演变与预测[J]. 南京林业大学学报（自然科学版）, 2019, 43(6): 105-112.
[9]	李楠,李龙伟,陆灯盛,张银龙,吴明. 杭州湾滨海湿地生态安全动态变化及趋势预测[J]. 南京林业大学学报（自然科学版）, 2019, 43(03): 107-115.
[10]	朱若柠,沈文娟,张亚丽,李明诗. 基于时间序列MODIS-VCF数据的云南省森林覆盖变化及破碎化分析[J]. 南京林业大学学报（自然科学版）, 2019, 43(02): 184-190.
[11]	薛羿,孟昭军,董效文,牛豪杰,严善春. 5种林分中性信息素对亚洲型舞毒蛾的诱捕作用[J]. 南京林业大学学报（自然科学版）, 2018, 42(05): 71-76.
[12]	郝建,蒙明君,黄德卫,韦菊玲,李忠国,唐继新,徐大平. 格木人工林节子的分布特征及预测模型[J]. 南京林业大学学报（自然科学版）, 2017, 41(03): 100-104.
[13]	王磊,黄利斌,万欣,张亚楠,王火. 城市森林对大气颗粒物(尤其PM_2.5)调控作用研究进展[J]. 南京林业大学学报（自然科学版）, 2016, 40(05): 148-154.
[14]	王岽,徐京京,周亮广,吴见. 安徽省生态系统服务价值变化及其趋势预测[J]. 南京林业大学学报（自然科学版）, 2015, 39(04): 88-94.
[15]	龚文峰,孙海,智长贵,张静,范文义. 额尔古纳国家级自然保护区主要森林植被地形梯度特征定量分析[J]. 南京林业大学学报（自然科学版）, 2015, 39(01): 79-85.