Inversion Research of Forest Stock Volume Using the Red Edge Bands of Sentinel-2A

doi:10.13466/j.cnki.lyzygl.2022.02.017

Abstract

Abstract:

Accurate and efficient estimation of forest stock volume is useful for measuring forest health and evaluating the carbon sequestration capacity of forests. The red-edge band is sensitive to vegetation chlorophyll changes,but its validity in forest stock volume estimation needs further verification. To explore the feasibility of red-edge band in estimating forest stock volume,the Xingning district of Nanning City was used as the study area,and different sets of modeling variables were constructed based on Sentinel-2A images to extract common band reflectance,red-edge band reflectance,common vegetation index and red-edge vegetation index,and the forest stock volume was estimated by multiple linear regression and random forest algorithm. The forest resources planning and design survey results data was used as the actual measurements for model accuracy evaluation. By comparing the modeling effects of the models and variable sets,the influence of the red-edge band on the estimation accuracy of the forest stock volume was analyzed. The results showed that the red-edge vegetation index was significantly correlated with the forest stock volume (P<0.01),and the forest stock volume estimation accuracy of the red-edge vegetation index variable set was significantly better than the other variable sets in different variable sets; among the two estimation models,the random forest model was more effective,and the accuracy of the random forest model was better than that of the multiple linear regression model in all variable sets. The random forest model using the variable set 4 achieved the highest estimation accuracy with R²,RMSE,and RRMSE of 0.66,28.63,and 23.54%,respectively. It can be concluded from the study that the red-edge band information of Sentinel-2A can be effectively used for remote sensing estimation of forest stock volume,which can provide a reference for efficient monitoring and management of forest resources by remote sensing.

Key words: forest stock volume, random forest, Sentinel-2A, red edge band, vegetation index

CLC Number:

LONG Zhihao, LUO Peng, XU Dengping, LI Zhen, DAI Huabin. Inversion Research of Forest Stock Volume Using the Red Edge Bands of Sentinel-2A[J]. FOREST RESOURCES WANAGEMENT, 2022, (2): 126-134.

Figures/Tables 9

Fig.1

Tab.1

Tab.2

Calculation formula of vegetation indexes

植被指数	表达式
ARVI^[18]	$b a n d 8 - b a n d 4 + γ b a n d 2 - b a n d 4$ / $b a n d 8 + b a n d 4 + γ b a n d 2 - b a n d 4$
MSAVI^[19]	[2band8+1- $2 b a n d 8 + 1 2 - 8 b a n d 8 - b a n d 4$ ]/2
NDVI^[20]	$b a n d 8 - b a n d 4$ / $b a n d 8 + b a n d 4$
RECI^[21]	$b a n d 8 / b a n d 5$ -1
RENDVI^[22]	$b a n d 8 - b a n d 5$ / $b a n d 8 + b a n d 5$
RESR^[23]	band8/band5

Tab.2

Fig.2

Tab.3

Fig.3

Fig.4

Tab.4

Fig.5

References 28

[1]	Nichol J E, Sarker M L R. Improved biomass estimation using the texture parameters of two high-resolution optical sensors[J]. IEEE Transactions on Geoscience and Remote Sensing, 2010, 49(3):930-948. doi: 10.1109/TGRS.2010.2068574
[2]	孙雪莲, 舒清态, 欧光龙. 基于随机森林回归模型的思茅松人工林生物量遥感估测[J]. 林业资源管理, 2015(1):71-76.
[3]	Gleason C J, Im J. A review of remote sensing of forest biomass and biofuel:options for small-area applications[J]. GIScience & Remote Sensing, 2011, 48(2):141-170.
[4]	闾妍宇, 李超, 欧光龙, 等. 基于地理加权回归模型的思茅松生物量遥感估测[J]. 林业资源管理, 2017(1):82-90.
[5]	王佳, 宋珊芸, 刘霞, 等. 结合影像光谱与地形因子的森林蓄积量估测模型[J]. 农业机械学报, 2014, 45(5):216-220.
[6]	蒋馥根, 孙华, 李成杰, 等. 联合GF-6和Sentinel-2红边波段的森林地上生物量反演[J]. 生态学报, 2021, 41(20):8222-8236.
[7]	刘欢, 王雅倩, 王晓明, 等. 基于近红外高光谱成像技术的小麦不完善粒检测方法研究[J]. 光谱学与光谱分析, 2019, 39(1):223-229.
[8]	吴静, 吕玉娜, 李纯斌, 等. 基于多时相Sentinel-2A的县域农作物分类[J]. 农业机械学报, 2019, 50(9):194-200.
[9]	蔡文婷, 赵书河, 王亚梅, 等. 结合Sentinel-2光谱与纹理信息的冬小麦作物茬覆盖度估算[J]. 遥感学报, 2020, 24(9):1108-1119.
[10]	谢巧云. 考虑红边特性的多平台遥感数据叶面积指数反演方法研究[D]. 北京: 中国科学院大学(中国科学院遥感与数字地球研究所), 2017.
[11]	Filho M G, Kuplich T M, Quadros F L F D. Estimating natural grassland biomass by vegetation indices using Sentinel 2 remote sensing data[J]. International Journal of Remote Sensing, 2020, 41(8):2861-2876. doi: 10.1080/01431161.2019.1697004
[12]	罗亚, 徐建华, 岳文泽, 等. 植被指数在城市绿地信息提取中的比较研究[J]. 遥感技术与应用, 2006(3):212-219.
[13]	龙依, 蒋馥根, 孙华, 等. 基于HLS数据的森林蓄积量遥感反演[J]. 森林与环境学报, 2021, 41(6):620-628.
[14]	Dong Taifeng, Meng Jihua, Shang Jiali, et al. Evaluation of chlorophyll-related vegetation indices using simulated Sentinel-2 data for estimation of crop fraction of absorbed photosynthetically active radiation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2015, 8(8):4049-4059. doi: 10.1109/JSTARS.2015.2400134
[15]	蒋馥根. 植被叶面积指数kNN优化方法反演研究[D]. 长沙: 中南林业科技大学, 2020.
[16]	刘茜, 杨乐, 柳钦火. 森林地上生物量遥感反演方法综述[J]. 遥感学报, 2015, 19(1):62-74.
[17]	张雷, 王琳琳, 张旭东, 等. 随机森林算法基本思想及其在生态学中的应用——以云南松分布模拟为例[J]. 生态学报, 2014, 34(3):650-659.
[18]	Kaufman Y J, Tanre D. Atmospherically resistant vegetation index (ARVI) for EOS-MODIS[J]. IEEE transactions on Geoscience and Remote Sensing, 1992, 30(2):261-270. doi: 10.1109/36.134076
[19]	Qi J, Chehbouni A, Huete A R, et al. A modified soil adjusted vegetation index[J]. Remote sensing of environment, 1994, 48(2):119-126. doi: 10.1016/0034-4257(94)90134-1
[20]	Rouse J W, Haas R H, Schell J A, et al. Monitoring vegetation systems in the Great Plains with ERTS[J]. NASA special publication, 1974, 351(1974):309.
[21]	Gitelson A A, Gritz Y, Merzlyak M N. Relationships between leaf chlorophyll content and spectral reflectance and algorithms for non-destructive chlorophyll assessment in higher plant leaves[J]. Journal of plant physiology, 2003, 160(3):271-282. pmid: 12749084
[22]	张磊, 宫兆宁, 王启为, 等. Sentinel-2影像多特征优选的黄河三角洲湿地信息提取[J]. 遥感学报, 2019, 23(2):313-326.
[23]	Erdle K, Mistele B, Schmidhalter U. Comparison of active and passive spectral sensors in discriminating biomass parameters and nitrogen status in wheat cultivars[J]. Field Crops Research, 2011, 124(1):74-84. doi: 10.1016/j.fcr.2011.06.007
[24]	郑阳, 吴炳方, 张淼. Sentinel-2数据的冬小麦地上干生物量估算及评价[J]. 遥感学报, 2017, 21(2):318-328.
[25]	陈妙金, 汪小钦, 吴思颖. 基于随机森林算法的水土流失影响因子重要性分析[J]. 自然灾害学报, 2019, 28(4):209-219.
[26]	Breiman L. Random Forests[J]. Machine Learning, 2001, 45:5-32. doi: 10.1023/A:1010933404324
[27]	朱婉雪, 李仕冀, 张旭博, 等. 基于无人机遥感植被指数优选的田块尺度冬小麦估产[J]. 农业工程学报, 2018, 34(11):78-86.
[28]	Lin Shangrong, Li Jing, Liu Qinhuo, et al. Evaluating the effectiveness of using vegetation indices based on red-edge reflectance from Sentinel-2 to estimate gross primary productivity[J]. Remote Sensing, 2019, 11(11):1303. doi: 10.3390/rs11111303

波段	描述	中心波长/nm	空间分辨率/m
蓝波段band2	Blue	490	10
绿波段band3	Green	560	10
红波段band4	Red	665	10
红边波段band5	Red Edge1	705	20
红边波段band6	Red Edge2	740	20
红边波段band7	Red Edge3	783	20
近红外波段band8	NIR	842	10
红边波段band8A	Red Edge4	865	20
短波红外band11	SWIR1	1610	20
短波红外band12	SWIR2	2190	20

变量	相关系数	变量	相关系数
band2	0.499^**	band11	0.426^**
band3	0.438^**	band12	0.579^**
band4	0.509^**	ARVI	-0.637^**
band5	0.481^**	MSAVI	-0.525^**
band6	-0.13	NDVI	-0.629^**
band7	-0.213^**	RECI	-0.565^**
band8	-0.251^**	RENDVI	-0.616^**
band8A	-0.213^**	RESR	-0.594^**

变量集	模型	决定系数 (R²)	均方根误差 (RMSE)/ (m³/hm²)	相对均方根误差 (RRMSE) /%
变量集1	多元线性回归	0.31	37.23	30.37
变量集2	多元线性回归	0.36	36.51	29.78
变量集3	多元线性回归	0.40	35.82	29.23
变量集4	多元线性回归	0.40	35.82	29.23
变量集1	随机森林	0.22	38.59	33.06
变量集2	随机森林	0.59	31.19	25.65
变量集3	随机森林	0.62	30.28	24.90
变量集4	随机森林	0.66	28.63	23.54