基于Sentinel-2B的油松冠层可燃物含水率反演研究

doi:10.13466/j.cnki.lyzygl.2023.04.017

摘要/Abstract

摘要：

森林火灾的发生与植被冠层可燃物含水率的大小有着密切联系。利用高精度、大尺度、高效率的遥感影像反演获取植被冠层可燃物含水率对于有效防治森林火灾具有重要意义。油松由于其自身理化性质成为引发森林火灾的主要树种之一,以张家口崇礼区的油松为研究对象,基于Sentinel-2B遥感影像和油松含水率实测数据,建立了多个油松冠层可燃物含水率反演模型:一元线性回归模型、一元非线性回归模型和多元非线性回归模型,并利用决定系数(R²)和均方根误差(RMSE)进行模型精度评价。结果表明,非线性模型总体上要优于线性模型;通过多个自变量因子建立的多元非线性模型能够更好地反映油松冠层可燃物含水率情况,模型反演精度更高,可以为植被冠层可燃物含水率反演模型方法选择提供一定的理论依据。

关键词: Sentinel-2B, 油松, 冠层可燃物含水率, 线性回归, 多元非线性回归

Abstract:

The occurrence of forest fires is closely related to the moisture content of vegetation canopy combustibles.Using high-precision,large-scale,and high-efficiency remote sensing image inversion to obtain the moisture content of vegetation canopy combustibles is of great significance for effective prevention and control of forest fires.Pinus tabulaeformis is one of the main tree species causing forest fires due to its physical and chemical properties.This study takes Pinus tabulaeformis in Chongli District,Zhangjiakou as the research object.Based on Sentinel 2B remote sensing images and measured moisture content dataof Pinus tabulaeformis,multiple linear regression models,nonlinear regression models and multiple nonlinear regression models were established for the moisture content of Pinus tabulaeformis canopy combustibles.Using the coefficient of determination(R²)and root mean square error(RMSE)to evaluate model accuracy.The results indicated that the nonlinear model was generally superior to the linear model;The multivariate nonlinear model established through multiple independent variable factors better reflected the moisture content of Pinus tabulaeformis canopy combustibles,and the model had higher inversion accuracy,which provided a certain theoretical basis for the selection of vegetation canopy fuel moisture inversion model methods.

Key words: Sentinel-2B, Pinus tabulaeformis, moisture content of combustibles in the canopy, principal component analysis, multiple nonlinear regression

中图分类号:

S762
S771.8

刘鸿升, 欧阳文欣, 魏英杰, 谢亦秋, 李建军. 基于Sentinel-2B的油松冠层可燃物含水率反演研究[J]. 林业资源管理, 2023,(4): 141-149.

LIU Hongsheng, OUYANG Wenxin, WEI Yingjie, XIE Yiqiu, LI Jianjun. Research on Inversion of Combustible Moisture Content in the Pinus Tabulaeformis Canopy Based on Sentinel-2B[J]. FOREST RESOURCES WANAGEMENT, 2023,(4): 141-149.

图/表 13

图1

图2

表1

植被指数计算

指数	公式	变量释义
NDVI	NDVI= $(ρ N I R - ρ R) ρ N I R + ρ R$	NIR:近红外波段反射率 SWIR:短红外波段反射率 R:红波段反射率 B:蓝光波段反射率 L:土壤调节因子
EVI	EVI=2.5× $ρ N I R - ρ R ρ N I R + 6 ρ R - 7.5 ρ B + 1$
RVI	RVI= $ρ N I R ρ R$
DVI	DVI=ρ_NIR-ρ_R
SAVI	SAVI= $ρ N I R - ρ R ρ N I R + ρ R + ρ L$ (1+ρ_L)
NDWI	NDWI= $ρ N I R - ρ S W I R ρ N I R + ρ S W I R$

表1

表2

表3

表4

图3

表5

图4

表6

表7

表8

图5

参考文献 27

[1]	Sandra D, Hector A, Wardle D A. Biodiversity in forest carbon sequestration initiatives:Not just a side benefit[J]. Current Opinion in Environmental Sustainability, 2009, 1(1):55-60. doi: 10.1016/j.cosust.2009.08.001
[2]	Yang Xiguang, Yu Ying, Hu Haiqing, et al. Moisture content estimation of forest litter based on remote sensing data[J]. Environmental monitoring and assessment, 2018, 190:1-10. doi: 10.1007/s10661-017-6336-1
[3]	Wu Chao, Venevsky S, Sitch S, et al. Present-day and future contribution of climate and fires to vegetation composition in the boreal forest of China[J]. Ecosphere, 2017, 8(8):e01917. doi: 10.1002/ecs2.2017.8.issue-8
[4]	Mohajane M, Costache R, Karimi F, et al. Application of remote sensing and machine learning algorithms for forest fire mapping in a Mediterranean area[J]. Ecological Indicators, 2021, 129:107869. doi: 10.1016/j.ecolind.2021.107869
[5]	Moritz M A, Batllori E, Bradstock R A, et al. Learning to coexist with wildfire[J]. Nature, 2014, 515:58-66. doi: 10.1038/nature13946
[6]	王秋华. 森林火灾燃烧过程中的火行为研究[D]. 北京: 中国林业科学研究院, 2010.
[7]	Loehman R A, Reinhardt E, Riley K L. Wildland fire emissions,carbon,and climate:Seeing the forest and the trees-A cross-scale assessment of wildfire and carbon dynamics in fire-prone,forested ecosystems[J]. Forest Ecology and Management, 2014, 317:9-19. doi: 10.1016/j.foreco.2013.04.014
[8]	全兴文, 何彬彬, 刘向茁, 等. 多模型耦合下的植被冠层可燃物含水率遥感反演[J]. 遥感学报, 2019, 23(1):62-77.
[9]	张志, 许文浩. 澳大利亚 2019—2020 森林火灾对我国应急管理体系建设的启示[J]. 中国应急救援, 2020(2):18-22.
[10]	赵月. 基于无人机的林火监测系统设计与实现[D]. 济南: 山东大学, 2019.
[11]	刘昕, 邸雪颖. 三种方法对森林地表可燃物含水率的预测评价[J]. 森林工程, 2013, 29(2):8-13.
[12]	孙龙, 刘祺, 胡同欣. 森林地表死可燃物含水率预测模型研究进展[J]. 林业科学, 2021, 57(4):142-152.
[13]	李海洋, 胡海清, 孙龙. 我国森林死可燃物含水率与气象和土壤因子关系模型研究[J]. 森林工程, 2016, 32(3):1-6.
[14]	Gao Bocai. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space[J]. Remote Sensing of Environment, 1996, 58(3):257-266. doi: 10.1016/S0034-4257(96)00067-3
[15]	Ceccato P, Gobron N, Flasse S, et al. Designing a spectral index to estimate vegetation water content from remote sensing data:Part 1 theoretical approach[J]. Remote Sensing of Environment, 2002, 82(2-3):188-197. doi: 10.1016/S0034-4257(02)00037-8
[16]	王长青. 结合Sentinel-1B和Landsat8 OLI数据的针叶林冠层叶片含水量反演研究[D]. 哈尔滨: 东北林业大学, 2018.
[17]	刘曦, 金森. 平衡含水率法预测死可燃物含水率的研究进展[J]. 林业科学, 2007(12):126-133.
[18]	张恒, 刘鑫源, 李星汉, 等. 我国气象要素回归法预测地表可燃物含水率研究进展[J]. 森林防火, 2018(3):25-31.
[19]	Luo Ke, Wei Yufeng, Du Jie, et al. Machine learning-based estimates of aboveground biomass of subalpine forests using Landsat 8 OLI and Sentinel-2B images in the Jiuzhaigou National Nature Reserve,Eastern Tibet Plateau[J]. Journal of Forestry Research, 2022, 33(4):1329-1340. doi: 10.1007/s11676-021-01421-w
[20]	冯舒芮, 成莉莉, 张珊珊, 等. 崇礼区森林植被和乔木生长的空间变化[J]. 中央民族大学学报:自然科学版, 2023, 32(2):5-9.
[21]	郜鹏, 赵海明, 解旭东, 等. 崇礼冬奥风景廊道生态修复数字监测体系构建[J]. 现代园艺, 2023, 46(8):151-153.
[22]	李贾, 兰岚, 张佐忠, 等. 基于深度学习的森林可燃物含水率反演技术[J]. 林业科学, 2022, 58(10):47-58.
[23]	Maffei C, Leone A P, Vella M, et al. Retrieval of vegetation moisture indicators for dynamic fire risk assessment with simulated MODIS radiance[C]. 2007 IEEE International Geoscience and Remote Sensing Symposium,Barcelona,Spain, 2007:4648-4651.
[24]	张峰, 周广胜. 植被含水量高光谱遥感监测研究进展[J]. 植物生态学报, 2018, 42(5):517-525. doi: 10.17521/cjpe.2017.0313
[25]	陈宗杰, 杨武年, 张琬琳, 等. 基于植被冠层含水量反演数据的森林易燃程度评价[J]. 测绘通报, 2023, 553(4):99-105.
[26]	Argañaraz J P, Landi M A, Bravo S J, et al. Estimation of live fuel moisture content from MODIS images for fire danger assessment in southern Gran Chaco[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(12):5339-5349. doi: 10.1109/JSTARS.4609443
[27]	Yebra M, Chuvieco E, Riaño D. Estimation of live fuel moisture content from MODIS images for fire risk assessment[J]. Agricultural and Forest Meteorology, 2008, 148(4):523-536. doi: 10.1016/j.agrformet.2007.12.005

	含水率	B4	NDVI	RVI	SAVI	NDWI
含水率	1
B4	-0.844	1
NDVI	0.882	-0.932	1
RVI	0.790	-0.792	0.922	1
SAVI	0.723	-0.663	0.881	0.935	1
NDWI	0.840	-0.754	0.830	0.744	0.733	1

成分	特征值	初始特征值方差百分比/%	累积百分比/%	总计/%	提取载荷平方和方差百分比/%	累积百分比/%
1	4.282	85.637	85.637	4.82	85.637	85.637
2	0.390	7.807	93.444
3	0.273	5.469	98.913
4	0.050	1.007	99.919
5	0.004	0.081	100.000

变量	主成分1
B4	-0.896
NDVI	0.988
RVI	0.952
SAVI	0.912
NDWI	0.875

自变量因子	一元线性回归模型	R²	RMSE
B4	y=0.395-1.390x	0.712	0.014
NDVI	y=0.161+0.234x	0.778	0.012
RVI	y=0.279+0.007x	0.624	0.016
SAVI	y=0.235+0.220x	0.523	0.018
NDWI	y=0.247+0.268x	0.706	0.014
Y	y=0.273+0.007x	0.643	0.016

自变量因子	二次多项式回归模型	R²	RMSE
B4	y=0.445+18.453x²-3.44x	0.748	0.013
NDVI	y=0.115-0.101x²+0.372x	0.780	0.012
RVI	y=0.219-0.001x²+0.024x	0.770	0.013
SAVI	y=0.032-0.998x²+1.135x	0.633	0.016
NDWI	y=0.258-0.261x²+0.362x	0.711	0.014
Y	y=0.207-0.001x²+0.024x	0.779	0.013