Prediction of Forest Fire Occurrence in Daxing’an Mountains Based on Logistic Regression Model

doi:10.13466/j.cnki.lyzygl.2019.01.018

Abstract

Abstract:

Based on the satellite data of forest fire from 2000 to 2016,this study selected such factors as meteorology,topography,vegetation,fuels,human activities and other factors as forest fire prediction variables,used Logistic regression model to analyze the main driving factors of forest fire occurrence,and established a forest fire prediction model in Daxing'an mountains.The results showed that the Logistic regression model has a high prediction accuracy of 80.6%,and the goodness-of -fit of the model is 0.868.In general,the fire risk is high in the south and east but low in the north and west.The residual analysis results showed that there large areas were underestimated in the south and southeast,indicating that the prediction ability of the model for these areas was low.

Key words: Daxing'an mountains, logistic regression model, fire prediction, fire model, driving factors

CLC Number:

S762

CHEN Dai. Prediction of Forest Fire Occurrence in Daxing’an Mountains Based on Logistic Regression Model[J]. FOREST RESOURCES WANAGEMENT, 2019, (1): 116-122.

Figures/Tables 8

Fig.1

Tab.1

Tab.2

Tab.3

Tab.4

Tab.5

Tab.6

Fig.2

References 24

[1]	Mckenzie D, Shankar U, Keane R E, et al. Smoke consequences of new wildfire regimes driven by climate change[J]. Earths Future, 2014,2(2):35-59.
[2]	Flannigan M D, Krawchuk M A, Groot W J D, et al.Implications of changing climate for global wildlandfire[J]. International Journal of Wildland Fire, 2009,18(18):483-507.
[3]	Loepfe L, Rodrigo A, Lloret F. Two thresholds determine climatic control of forest fire size in Europe and northern Africa[J]. Regional Environmental Change, 2014,14(4):1395-1404.
[4]	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]. International Journal of Wildland Fire, 2016,25(5):505-519.
[5]	Guo FT, Su ZW, Wang GY, et al. Understanding fire drivers and relative impacts in different Chinese forest ecosystems[J]. Science of the Total Environment, 2017,605:411-425. pmid: 28672230
[6]	石晶晶. 浙江省林火发生格局及预测模型研究[D]. 临安:浙江农林大学, 2014.
[7]	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]. Forest Ecology and Management, 2012,275(4):117-129.
[8]	Pradhan B, Suliman M D H B, Awang M A B. Forest fire susceptibility and risk mapping using remote sensing and geographical information systems(GIS)[J]. Disaster Prevention and Management, 2007,16(3):344-352.
[9]	Vilar L, Woolford D G, Martell D L, et al. A model for predicting human-caused wildfire occurrence in the region of Madrid,Spain[J]. International Journal of Wildland Fire, 2010,19(3):325-337.
[10]	Justice C O, Giglio L, Korontzi, S, et al. The MODIS fire products[J]. Remote Sensing of Environment, 2002,83(1):244-262.
[11]	Chang Y, Zhu Z L, Bu R C, et al. Predicting fire occurrence patterns with logistic regression in Heilongjiang Province,China[J]. Landscape Ecology, 2013,28(10):1989-2004.
[12]	胡海清. 林火生态与管理[M]. 北京: 中国林业出版社, 2005.
[13]	Chandler C, Cheney P, Trabaud L, et al. Fire in Forestry.Volume I.Forest fire behavior and effects[M]. Florida: Krieger Publishing Company, 1991.
[14]	Daily Climate Data Set of China International Exchange Station,China Meteorological Data and Sharing Network[EB/OL].(2016-05-15)[2019-01-01]. http://data.cma.cn/site/index.html.
[15]	Geographic Information Resources Service.National Administration of Surveying,Mapping and Geo-information of China.2002[EB/OL].(2016-05-15)[2019-01-01]. http://www.webmap.cn/main.do?method=index(accessed on 15 May 2016)
[16]	Gutman G, Ignatov A. The derivation of the green vegetation fraction from NOAA/AVHRR data for use in numerical weather prediction models[J]. International Journal ofRemote Sensing, 1998,19(8):1533-1543.
[17]	Ran Y H, Li X. Plant Functional Types Map in China.Cold and Arid Regions Science Data Center at Lanzhou,2011[EB/OL].(2016-05-15)[2019-01-01].Available online: http://westdc.westgis.ac.cn/.
[18]	Liu H, Jiang D, Yang X, et al. Spatialization Approach to 1km Grid GDP Supported by Remote Sensing[J]. Geo-information Science, 2005,7(2):120-123.
[19]	许吟隆, 薛峰, 林一骅. 不同温室气体排放情景下中国21世纪地面气温和降水变化的模拟分析[J].气候与环境研究, 2003(2):209-217.
[20]	邓欧, 李亦秋, 冯仲科, 等. 基于空间Logistic的黑龙江省林火风险模型与火险区划[J]. 农业工程学报, 2012,28(8):200-205.
[21]	S Guriş E, Çaĝlayan T Un. Estimating of probability of home-ownership in rural and urban areas:Logit,probit and Gompitmodel[J]. European Journal of Social Sciences, 2011,21(3):405-411.
[22]	Minab M, Ola M, Asl M G. Analysis of the effect of corporate governance on the financial health of companies Using Logit,Probit and Gompit models [C]//The 12th Iranian National Conference on Accounting,Iran, 2014.
[23]	Littell J S, McKenzie D, Peterson D L, et al.Climate and wildfire area burned in western U.S.ecoprovinces[J]. Ecological Applications, 2009,19(4):1003-1021. pmid: 19544740
[24]	Pourghasemi H R. GIS-based forest fire susceptibility mapping in Iran:a comparison between evidential belief function and binary logistic regression models[J]. Scandinavian Journal of Forest Research, 2016,31(1):80-98. doi: 10.1080/02827581.2015.1052750

变量类型	变量说明	变量名	分辨率/尺度	数据来源
气象(防火期)	日平均风速	WIN_avg	0.1m/s	中国气象数据共享网络^[14]
	日累计降水量	Pre	0.1mm
	日最高气温	Tmax	0.1℃
	日平均相对湿度	RH_avg	%
地形	高程	Elev	1m	地理空间数据云^[15]
地形	坡度	Slope	1°	地理空间数据云^[15]
植被	植被覆盖度	FVC	%	中国科学院计算机网络信息中心国际科学技术数据镜像网站^[17]
基础设施	到铁路的距离	Dis_railway	1m
	到公路的距离	Dis_road	1m	国家测绘和地质信息管理局^[11]
	到居民点的距离	Dis_residential	1m	国家测绘和地质信息管理局^[11]
社会经济因素	人均GDP	CGDP	元/km	地球系统科学的数据共享基础设施^[19]
社会经济因素	人口密度	POP	人/km	地球系统科学的数据共享基础设施^[19]

变量	1∶1			1∶2					相关性
变量	P_max	P_min	显著的样本数	P_max	P_min		显著的样本数		相关性
坡向	<0.0001	<0.0001	5	<0.0001		<0.0001		5	—
到公路距离	<0.0001	<0.0001	5	<0.0001		<0.0001		5	+
到铁路距离	<0.0001	<0.0001	5	<0.0001		<0.0001		5	+
到居民区距离	<0.0001	<0.0001	5	<0.0001		<0.0001		5	+
人口密度	0.0027	0.0002	5	0.0001		<0.0001		5	+
GDP	<0.0001	<0.0001	5	<0.0001		<0.0001		5	+
植被覆盖度	0.023	0.0002	5	0.0267		0.0008		3	+
日平均风速	0.0036	<0.0001	5	0.0469		0.0006		4	-
日累计降水	<0.0001	<0.0001	5	<0.0001		<0.0001		5	-
日最高气温	<0.0001	<0.0001	5	<0.0001		<0.0001		5	+
日相对湿度	<0.0001	<0.0001	5	<0.0001		<0.0001		5	-

样本	临界值	ROC曲线下面积	预测准确率/%
样本	临界值	ROC曲线下面积	训练样本	测试样本
样本1	0.506	0.872	79	78.5
样本2	0.527	0.874	78.7	78.1
样本3	0.511	0.874	78.9	77.8
样本4	0.526	0.866	78.1	80
样本5	0.526	0.869	78.6	79.1

样本	临界值	ROC曲线下面积	预测准确率/%
样本	临界值	ROC曲线下面积	训练样本	测试样本
样本1	0.364	0.866	80.6	80.8
样本2	0.333	0.865	80.2	79.2
样本3	0.338	0.872	80.9	77.9
样本4	0.328	0.868	80.8	78.6
样本5	0.363	0.866	80.8	79.1

参数	估计系数	标准误差	卡方值	显著水平
常量	-0.2121	0.1376	2.3756	0.1232
坡向	-0.1827	0.00673	737.7285	<0.0001
到公路距离	0.0542	0.00369	215.6088	<0.0001
到铁路距离	0.00971	0.00113	74.1665	<0.0001
到居民区距离	0.0177	0.00151	136.6398	<0.0001
人口密度	0.000521	0.000095	30.1968	<0.0001
GDP	0.00212	0.000222	91.3339	<0.0001
植被覆盖度	0.4312	0.1516	8.0864	0.0045
日平均风速	-0.0506	0.0188	7.2715	0.007
日累计降水	-0.3598	0.0282	163.1947	<0.0001
日最高气温	0.033	0.00296	124.1604	<0.0001
日相对湿度	-0.0314	0.00175	323.4969	<0.0001