基于GEE与Sentinel-2影像的落叶针叶林提取

王春玲; 樊怡琳; 庞勇; 荚文

doi:10.12171/j.1000-1522.20220422

基于GEE与Sentinel-2影像的落叶针叶林提取

doi: 10.12171/j.1000-1522.20220422

王春玲^{1, 2,},
樊怡琳^{1, 2},
庞勇^3, ,,
荚文³

1.
北京林业大学信息学院，北京 100083
2.
国家林业草原林业智能信息处理工程技术研究中心，北京 100083
3.
中国林业科学研究院资源信息研究所，北京 100091

基金项目: “十三五”国家重点研发计划（2017YFD0600404）

详细信息

作者简介:
王春玲，博士，副教授。主要研究方向：大数据技术与人工智能。Email：wangchl@bjfu.edu.cn　地址：100083 北京市海淀区清华东路35号北京林业大学信息学院

责任作者:
庞勇，博士，研究员。主要研究方向：林业遥感机理模型、激光雷达信号处理。 Email：pangy@ifrit.ac.cn　地址：100091 北京市海淀区颐和园后中国林业科学研究院资源信息研究所

中图分类号: S757.3
计量
- 文章访问数: 106
- HTML全文浏览量: 22
- PDF下载量: 36
- 被引次数: 0
出版历程
- 收稿日期: 2022-10-22
- 修回日期: 2022-11-07
- 网络出版日期: 2023-07-19
- 刊出日期: 2023-08-25

Extraction of deciduous coniferous forest based on Google earth engine (GEE) and Sentinel-2 image

Wang Chunling^{1, 2
,},
Fan Yilin^{1, 2},
Pang Yong^{3
, ,},
Jia Wen³

1.
School of Information Science and Technology, Beijing Forestry University, Beijing 100083, China
2.
Engineering Research Center for Forestry-Oriented Intelligent Information Processing ofNational Forestry and Grassland Administration, Beijing 100083, China
3.
Research Institute of Forest Resource Information Techniques , Chinese Academy of Forestry, Beijing 100091, China

摘要

摘要: 目的针对森林资源精细监测评价的需求，探索多时相、多特征的Sentinel-2影像在落叶针叶林识别中的应用潜力，根据落叶针叶林的物候特征构建分类模型，为大范围落叶针叶林识别提供方法参考。方法基于GEE平台，以黑龙江省孟家岗林场为研究区，分析不同季节落叶针叶林与其他森林之间的差异。研究使用2020年春季（5月7日和5月27日）、夏季（8月9日）和秋季（10月19日）的4景Sentinel-2影像，提取光谱特征、纹理特征和地形特征构建多特征数据集，根据特征重要性得分进行特征优选，最后使用随机森林分类器得到落叶针叶林识别的最佳模型，实现孟家岗林场落叶针叶林的精确提取。结果试验结果表明落叶针叶林具有明显的植被光谱特征和季相特性，多时相影像数据包含落叶针叶林更多物候期，春季和秋季的影像更有利于区分落叶针叶林与其他森林。此外，近红外、短波红外波段的光谱信息对识别落叶针叶林有较大帮助。利用 GEE平台和多时相Sentinel-2影像可以高效快速地提取植被信息，落叶针叶林提取总体精度与Kappa系数分别达到 91.20%，0.82。结论基于GEE平台和Sentinel-2影像构建的分类模型对落叶针叶林信息的快速提取有一定的可行性和适用性，研究结果对大面积落叶针叶林的空间位置分布提取具有一定的参考价值。
- 落叶针叶林 /
- Sentinel-2影像 /
- 物候特征 /
- 随机森林 /
- GEE平台
Abstract: Objective In view of fine monitoring and evaluation needs of forest resources, the application potential of multi-temporal and multi-feature Sentinel-2 images in the identification of deciduous coniferous forests was exploratively studied, and a classification model was built according to the phenological characteristics of deciduous coniferous forests to provide method reference for identifying deciduous coniferous forests on a large scale. Method Based on the GEE platform, the difference between deciduous coniferous forests and other forests in different seasons was analyzed by taking Mengjiagang Forest Farm in Heilongjiang Province of northeastern China as the research area. In this study, four seasonal Sentinel-2 images of spring (May 7 and May 27), summer (August 9), and autumn (October 19) in 2020 were used to construct a multi-feature dataset by extracting spectral features, texture features, and topographic features. Feature optimization was carried out according to feature importance scores. Finally, the optimal model for identifying deciduous coniferous forests was established by a random forest classifier to achieve rapid extraction of deciduous coniferous forest in Mengjiagang Forest Farm. Result The experimental results showed that deciduous coniferous forests displayed obvious vegetation spectral features and seasonal characteristics. The multi-temporal image data contained more phenological period information on deciduous coniferous forests, and the images in spring and autumn can enable better differentiation between deciduous coniferous forests and other forests. In addition, near-infrared and short-wave infrared spectral information can greatly help identify deciduous coniferous forests. By the GEE platform and multi-temporal Sentinel-2 images, it is possible to extract vegetation information efficiently and quickly. The overall extraction accuracy and Kappa coefficient of deciduous coniferous forest reached 91.20% and 0.82, respectively. Conclusion The classification model constructed based on the GEE platform and Sentinel-2 image has certain feasibility and applicability for the rapid extraction of deciduous coniferous forest information, and the research results provide a certain reference value for the large-scale extraction of spatial location distribution information of deciduous coniferous forest.
- deciduous coniferous forest /
- Sentinel-2 image /
- phenological characteristics /
- random forest /
- Google earth engine (GEE) platform

HTML全文

图 1 样本分布

Figure 1. Sample plot distribution

下载: 全尺寸图片幻灯片

图 2 技术流程图

Figure 2. Technique flowchart

下载: 全尺寸图片幻灯片

图 3 森林分布

Figure 3. Forest distribution

下载: 全尺寸图片幻灯片

图 4 落叶针叶林分布

Figure 4. Deciduous coniferous forest distribution

下载: 全尺寸图片幻灯片

图 5 本文局部 Google earth影像及其对应的分类结果

Figure 5. Local Google earth images and their corresponding classification results of this paper

下载: 全尺寸图片幻灯片

图 6 不同特征数量的分类结果精度

Figure 6. Classification accuracy of different feature number

下载: 全尺寸图片幻灯片

图 7 特征重要性得分

Figure 7. Score of feature importance

下载: 全尺寸图片幻灯片

图 8 特征变量相关系数

Figure 8. Correlation coefficient of feature variables

下载: 全尺寸图片幻灯片

图 9 时间序列曲线

Figure 9. Time series curve

下载: 全尺寸图片幻灯片

图 10 不同时期不同地物波谱图

Figure 10. Spectral curves of different objects in different periods

下载: 全尺寸图片幻灯片

表 1 Sentinel-2光谱波段信息

Table 1. Sentinel-2 spectral band information

波段 Band	描述 Description	中心波长 Central wavelength/nm	分辨率 Resolution/m
B2	蓝光 Blue light	497	10
B3	绿光 Green light	560	10
B4	红光 Red light	665	10
B5	红边1 Red edge 1	704	20
B6	红边2 Red edge 2	740	20
B7	红边3 Red edge 3	783	20
B8	近红外 Near-infrared（NIR）	835	10
B8A	窄波近红外 Narrow NIR	865	20
B11	短波红1 Short-wave infrared 1 （SWIR1）	1 614	20
B12	短波红2 Short-wave infrared 2 （SWIR2）	2 202	20

下载: 导出CSV

表 2 土地覆盖产品收集情况

Table 2. Collection of land cover products

产品分类 Product classification	发布机构 Issuing authority	数据来源 Data source	产品年份 Product year
GLC_FCS 30	中国科学院 Chinese Academy of Sciences	Landsat TM/ETM+	2000、2015、2020
ChinaCover	中国科学院 Chinese Academy of Sciences	Landsat TM/ETM+HJ-1A/B	2000、2010、2015

下载: 导出CSV

表 3 样本情况

Table 3. Sample plot situation

类型 Type	训练样本 Training sample	验证样本 Validation sample	描述 Description
森林 Forest	230	133
非森林 Non-forest	216	80
落叶针叶林 Deciduous coniferous forest	150	98	实地 In the field
落叶针叶林 Deciduous coniferous forest	150	88	机载高光谱 Airborne hyperion （CAF-LiCHy）
其他森林 Other forest	163	84	实地 In the field
其他森林 Other forest	163	88	机载高光谱 Airborne hyperion （CAF-LiCHy）
合计 Total	759	571

下载: 导出CSV

表 4 光谱特征

Table 4. Spectral characteristics

光谱指数 Spectral index	公式 Formula	参考文献 Reference
NDVI	（B8 − B4）/（B8 + B4）	[20]
SAVI	（B8 − B4）/（B8 + B4 + 0.5） × 1.5	[21]
EVI	2.5 × （B8 − B4）/（B8 + 6.0 × B4 − 7.5 × B2 + 1.0）	[22]
RVI	B8/B4	[23]
DVI	B8 − B4	[24]
注：NDVI. 归一化植被指数；SAVI. 土壤调整植被指数；EVI. 增强型植被指数；RVI. 比值植被指数；DVI.差值植被指数。Notes: NDVI, normalized difference vegetation index; SAVI, soil-adjusted vegetation index; EVI, enhanced vegetation index; RVI, ratio vegetation index; DVI, difference vegetation index.

下载: 导出CSV

表 5 特征统计

Table 5. Statistic of features

特征类型 Feature type	特征变量 Feature variables	数量 Number
光谱波段 Spectral band	B2_5，B3_5，B4_5，B5_5，B6_5，B7_5，B8_5，B8A_5，B11_5，B12_5	30
	B2_8，B3_8，B4_8，B5_8，B6_8，B7_8，B8_8，B8A_8，B11_8，B12_8
	B2_10，B3_10，B4_10，B5_10，B6_10，B7_10，B8_10，B8A_10，B11_10，B12_10
光谱指数 Spectral index	NDVI_5， DVI_5， RVI_5， EVI_5， SAVI_5	15
	NDVI_8， DVI_8， RVI_8， EVI_8， SAVI_8
	NDVI_10， DVI_10， RVI_10， EVI_10， SAVI_10
纹理特征 Textural feature	corr， var， idm，con，ent，asm	6
地形特征 Topographic feature	海拔 Elevation ，坡度 Slope ，坡向 Aspect	3
合计 Total		54
注：corr. 相关性；var. 方差；idm. 逆差矩；con. 对比度；ent. 熵；asm.角二阶矩。Notes: corr, correlation; var, variance; idm, inverse different moment; con, contrast; ent, entropy; asm, angular second-order moment.

下载: 导出CSV

表 6 混淆矩阵

Table 6. Confusion matrix

类别 Type	非森林 Non-forest	森林 Forest	生产精度 Production accuracy/%
非森林 Non-forest	75	5	93.75
森林 Forest	3	130	97.74
用户精度 User accuracy/%	96.15	96.29
注：总精度96.20%；Kappa系数0.93。Notes: overall accuracy is 96.20%; Kappa coefficient is 0.93.

下载: 导出CSV

表 7 机载高光谱数据验证样本混淆矩阵

Table 7. CAF-LiCHy validation of sample confusion matrix

类别 Type	其他森林 Other forest	落叶针叶林 Deciduous coniferous forest	生产精度 Production accuracy/%
其他森林 Other forest	78	6	92.85
落叶针叶林 Deciduous coniferous forest	11	75	87.20
用户精度 User accuracy/%	87.64	92.59
注：总精度90.00%；Kappa系数0.80。Notes: overall accuracy is 90%; Kappa coefficient is 0.80.

下载: 导出CSV

表 8 实地验证样本混淆矩阵

Table 8. Field validation of sample confusion matrix

类别 Type	其他森林 Other forest	落叶针叶林 Deciduous coniferous forest	生产精度 Production accuracy/%
其他森林 Other forest	78	6	92.85
落叶针叶林 Deciduous coniferous forest	10	88	89.79
用户精度 User accuracy/%	88.86	93.61
注：总精度91.20%；Kappa系数0.82。Notes: overall accuracy is 91.20%； Kappa coefficient is 0.82.

下载: 导出CSV

参考文献(29)

[1]	周以良. 中国的几种植被类型(Ⅵ)落叶针叶林[J]. 生物学通报, 1988(5): 6−10. Zhou Y L. Several types of vegetation in China (Ⅵ) deciduous coniferous forest[J]. Bulletin of Biology, 1988(5): 6−10.
[2]	刘旭升, 张晓丽. 森林植被遥感分类研究进展与对策[J]. 林业资源管理, 2004(1): 61−64. Liu X S, Zhang X L. Research advances and counter measures of remote sensing classification of forest vegetation[J]. Forest Resources Management, 2004(1): 61−64.
[3]	Teluguntla P, Thenkabail P S, Oliphant A, et al. A 30-m landsat-derived cropland extent product of Australia and China using random forest machine learning algorithm on Google Earth Engine cloud computing platform[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 144: 325−340. doi: 10.1016/j.isprsjprs.2018.07.017
[4]	Gorelick N, Hancher M, Dixon M, et al. Google Earth Engine: planetary-scale geospatial analysis for everyone[J]. Remote Sensing of Environment, 2017, 202: 18−27. doi: 10.1016/j.rse.2017.06.031
[5]	李斌, 李崇贵, 李煜. 基于Sentinel-2数据的塞罕坝机械林场落叶松人工林提取[J]. 林业资源管理, 2021(2): 117−123. Li B, Li C G, Li Y. Research on larch extraction in Saihanba Mechanical Forest Farm based on Sentinel-2 data[J]. Forest Resources Management, 2021(2): 117−123.
[6]	Immitzer M, Vuolo F, Atzberger C. First experience with Sentinel-2 data for crop and tree species classifications in central Europe[J/OL]. Remote Sensing, 2016, 8(3): 166[2023−01−22]. https://doi.org/10.3390/rs8030166.
[7]	Persson M, Lindberg E, Reese H. Tree species classification with multi-temporal Sentinel-2 data[J/OL]. Remote Sensing, 2018, 10(11): 1794[2023−01−02]. https://doi.org/10.3390/rs10111794.
[8]	姚茂林, 江洪, 张丽玉. 基于Sentinel-2影像红边光谱指数与特征优选的竹林提取研究[J]. 海南大学学报(自然科学版), 2022, 40(4): 373−381. Yao M L, Jiang H, Zhang L Y. Bamboo information extraction from Sentinel-2 image based on improved spectral indices and random forest Gini index[J]. Natural Science Journal of Hainan University, 2022, 40(4): 373−381.
[9]	Immitzer M, Neuwirth M, Böck S, et al. Optimal input features for tree species classification in central Europe based on multi-temporal Sentinel-2 data[J/OL]. Remote Sensing, 2019, 11(22): 2599[2023−01−02]. https://doi.org/10.3390/rs11222599.
[10]	Grabska E, Frantz D, Ostapowicz K. Evaluation of machine learning algorithms for forest stand species mapping using Sentinel-2 imagery and environmental data in the Polish Carpathians[J/OL]. Remote Sensing of Environment, 2020, 251: 112103[2023−01−23]. https://doi.org/10.1016/j.rse.2020.112103.
[11]	郑杨, 董利虎, 李凤日. 黑龙江省红松人工林枝条分布数量模拟[J]. 应用生态学报, 2016, 27(7): 2172−2180. Zheng Y, Dong L H, Li F R. Branch quantity distribution simulation for Pinus koraiensis plantation in Heilongjiang Province, China[J]. Chinese Journal of Applied Ecology, 2016, 27(7): 2172−2180.
[12]	Shoko C, Mutanga O. Examining the strength of the newly-launched Sentinel 2 MSI sensor in detecting and discriminating subtle differences between C3 and C4 grass species[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2017, 129: 32−40. doi: 10.1016/j.isprsjprs.2017.04.016
[13]	吴炳方, 苑全治, 颜长珍, 等. 21世纪前十年的中国土地覆盖变化[J]. 第四纪研究, 2014, 34(4): 723−731. Wu B F, Yuan Q Z, Yan C Z, et al. Land cover changes of China from 2000 to 2010[J]. Quaternary Sciences, 2014, 34(4): 723−731.
[14]	吴炳方. 中华人民共和国土地覆被地图集(1∶1 000 000)[M]. 北京: 中国地图出版社, 2017. Wu B F. Land cover atlas of the People’s Republic of China (1∶1 000 000)[M]. Beijing: China Cartographic Publishing House, 2017.
[15]	Zhang X, Liu L, Chen X, et al. Fine land-cover mapping in China using Landsat datacube and an operational SPECLib-based approach[J/OL]. Remote Sensing, 2019, 11(9): 1056[2023−01−02]. https://doi.org/10.3390/rs11091056.
[16]	Zhang X, Liu L, Chen X, et al. GLC_FCS30: Global Land-Cover product with fine classification system at 30m using time-series Landsat imagery[J]. Earth System Science Data, 2021, 13(6): 2753−2776. doi: 10.5194/essd-13-2753-2021
[17]	庞勇, 蒙诗栎, 史锴源, 等. 中国天然林保护工程区森林覆盖遥感监测[J]. 生态学报, 2021, 41(13): 5080−5092. Pang Y, Meng S L, Shi K Y, et al. Forest coverage monitoring in the Natural Forest Protection Project area of China[J]. Acta Ecologica Sinica, 2021, 41(13): 5080−5092.
[18]	Pang Y, Li Z, Ju H, et al. LiCHy: the CAF’s LiDAR, CCD and hyperspectral integrated airborne observation system[J/OL]. Remote Sensing, 2016, 8(5): 398[2023−01−02]. https://doi.org/10.3390/rs8050398.
[19]	池毓锋, 赖日文, 余莉莉, 等. 基于Landsat 8 OLI数据的树种类型分布提取[J]. 自然资源学报, 2017, 32(7): 1193−1203. Chi Y F, Lai R W, Yu L L, et al. Extracting tree species distribution with Landsat 8 OLI data[J]. Journal of Natural Resources, 2017, 32(7): 1193−1203.
[20]	Tucker C J. Red and photographic infrared linear combinations for monitoring vegetation[J]. Remote Sensing of Environment, 1979, 8(2): 127−150. doi: 10.1016/0034-4257(79)90013-0
[21]	Huete A R. A soil-adjusted vegetation index (SAVI)[J]. Remote Sensing of Environment, 1988, 25(3): 295−309. doi: 10.1016/0034-4257(88)90106-X
[22]	董灵波, 梁凯富, 张一帆,等. 基于Landsat 8时间序列数据的翠岗林场森林类型划分[J]. 应用生态学报, 2022, 33(9): 2339−2346. Dong L B, Liang K F, Zhang Y F, et al. Classification of forest types in Cuigang Forest Farm based on time series data of Landsat 8[J]. Chinese Journal of Applied Ecology, 2022, 33(9): 2339−2346.
[23]	尹芬, 丁美青. 运用 ENVI 实现 SPOT-5 卫星影像的 RVI 提取[J]. 测绘与空间地理信息, 2016, 39(1): 47−48. Yin F, Ding M Q. RVI extraction of SPOT-5 satellite images using ENVI[J]. Geomatics & Spatial Information Technology, 2016, 39(1): 47−48.
[24]	周晓双. 基于高光谱的稻麦叶面积指数监测研究[D]. 南京: 南京农业大学, 2015. Zhou X S. Monitoring leaf area index in rice and wheat with hyperspectral remote sensing[D]. Nanjing: Nanjing Agricultural University, 2015.
[25]	李智峰, 朱谷昌, 董泰锋. 基于灰度共生矩阵的图像纹理特征地物分类应用[J]. 地质与勘探, 2011, 47(3): 456−461. Li Z F, Zhu G C, Dong T F. Application of GLCM-based texture features to remote sensing image classification[J]. Geology and Exploration, 2011, 47(3): 456−461.
[26]	Huang D, Xu S, Sun J, et al. Accuracy assessment model for classification result of remote sensing image based on spatial sampling[J/OL]. Journal of Applied Remote Sensing, 2017, 11(4): 046023[2023−01−02]. https://doi.org/10.1117/1.JRS.11.046023.
[27]	Qu L, Chen Z, Li M, et al. Accuracy improvements to pixel-based and object-based lulc classification with auxiliary datasets from Google earth engine[J/OL]. Remote Sensing, 2021, 13(3): 453[2023−01−02]. https://doi.org/10.3390/rs13030453.
[28]	Hill R A, Wilson A K, George M, et al. Mapping tree species in temperate deciduous woodland using time-series multi-spectral data[J]. Applied Vegetation Science, 2010, 13(1): 86−99. doi: 10.1111/j.1654-109X.2009.01053.x
[29]	Nelson M. Evaluating multitemporal sentinel-2 data for forest mapping using random forest[J/OL]. Environmental Science, Mathematics, 2017[2023−01−02]. https://core.ac.uk/display/132537655.

施引文献

资源附件(0)

访问统计

点击查看大图

图(10) / 表(8)

计量

文章访问数: 106
HTML全文浏览量: 22
PDF下载量: 36
被引次数: 0

姓名
邮箱
手机号码
标题
留言内容
验证码

留言板

基于GEE与Sentinel-2影像的落叶针叶林提取

doi: 10.12171/j.1000-1522.20220422

作者简介:
王春玲，博士，副教授。主要研究方向：大数据技术与人工智能。Email：wangchl@bjfu.edu.cn　地址：100083 北京市海淀区清华东路35号北京林业大学信息学院

责任作者:
庞勇，博士，研究员。主要研究方向：林业遥感机理模型、激光雷达信号处理。 Email：pangy@ifrit.ac.cn　地址：100091 北京市海淀区颐和园后中国林业科学研究院资源信息研究所

计量

Extraction of deciduous coniferous forest based on Google earth engine (GEE) and Sentinel-2 image

计量

目录

留言板

基于GEE与Sentinel-2影像的落叶针叶林提取

doi: 10.12171/j.1000-1522.20220422

作者简介: 王春玲，博士，副教授。主要研究方向：大数据技术与人工智能。Email：wangchl@bjfu.edu.cn 地址：100083 北京市海淀区清华东路35号北京林业大学信息学院

责任作者: 庞勇，博士，研究员。主要研究方向：林业遥感机理模型、激光雷达信号处理。 Email：pangy@ifrit.ac.cn 地址：100091 北京市海淀区颐和园后中国林业科学研究院资源信息研究所

计量

出版历程

Extraction of deciduous coniferous forest based on Google earth engine (GEE) and Sentinel-2 image

计量

出版历程

目录

作者简介:
王春玲，博士，副教授。主要研究方向：大数据技术与人工智能。Email：wangchl@bjfu.edu.cn　地址：100083 北京市海淀区清华东路35号北京林业大学信息学院

责任作者:
庞勇，博士，研究员。主要研究方向：林业遥感机理模型、激光雷达信号处理。 Email：pangy@ifrit.ac.cn　地址：100091 北京市海淀区颐和园后中国林业科学研究院资源信息研究所