Fine-scale remote sensing monitoring and interpretation of large-scene vegetation health in the Jiuzhai Valley biosphere reserve: A case study of the Changhai pilot zone
GAO Sheng1,2(), CHEN Fulong1,3(), SHI Pilong1,3, ZHOU Wei1,3, ZHU Meng1,3, LUO Yansong1,2, YANG Qingxia4, WANG Qin4
1. Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China 2. University of Chinese Academy of Sciences, Beijing 100049, China 3. International Research Center of Big Data for Sustainable Development Goals, Beijing 100094, China 4. Jiuzhaigou Valley Scenic Area Administration, Jiuzhaigou 623402, China
Under the intertwined effects of natural processes, geological disasters, and human disturbances, the health risks of vegetation in biosphere reserves have increased. Accurately extracting and identifying vegetation health information from complex large scenes faces technical challenges. This study investigated the Changhai pilot zone of the Jiuzhai Valley biosphere reserve by leveraging the macro, objective, and quantitative advantages of remote sensing technology. It proposed a fine-scale remote sensing monitoring method integrated with feature extraction and random forest for large-scene vegetation health, achieving the information extraction and target identification of unhealthy trees in typical biosphere reserves. The results show that: ① The random forest classification method combined with spectral and texture features can accurately extract unhealthy trees scattered in forests from high-resolution remote sensing images; ② The red-green ratio index, normalized difference vegetation index, correlation between red-edge and red bands, and corrected soil-adjusted vegetation index constitute typical features for extracting vegetation health information from remote sensing images; ③ The Changhai pilot zone exhibits a generally fair vegetation health status, with unhealthy trees accounting for 0.23%, and geological disasters exert positive effects on the spatial distribution of unhealthy trees. This study provides primary scientific data for vegetation health diagnosis of the Jiuzhai Valley biosphere reserve while showing generalization value for the remote sensing monitoring of ecological security in other biosphere reserves of China.
Sheng GAO,Fulong CHEN,Pilong SHI, et al. Fine-scale remote sensing monitoring and interpretation of large-scene vegetation health in the Jiuzhai Valley biosphere reserve: A case study of the Changhai pilot zone[J]. Remote Sensing for Natural Resources,
2024, 36(2): 188-197.
Fig.2 Comparison of spectral curves between healthy and unhealthy spruce-fir
Fig.3 Disease-impacted trees in WorldView images and UAV image
Fig.4 A flowchart of the analysis procedure for vegetation health monitoring
类型
光谱指数
计算公式①
描述
归一化植被 指数
NDVI57
NDVI57
传统用来识别植被的指数,与叶绿素浓度相关
NDVI58
NDVI58
红绿指数
RGI34
RGI34
凸显叶片变黄趋势
RGI35
RGI35
RGI45
RGI45
修正的土壤调节植被指数
MSAVI57
MSAVI57
弱化土壤对植被的影响
MSAVI58
MSAVI58
绿色归一化植被指数
GNDVI37
GNDVI37
与叶绿素浓度相关,且对于叶绿素a浓度相比于NDVI更为敏感
GNDVI38
GNDVI38
归一化差异红边指数
NDRE67
NDRE67
反映植被受胁迫时早期的红边异常
NDRE68
NDRE68
Tab.1 Spectral indices calculated from WorldView-2
纹理度量
计算公式①
描述
均值MEA
灰度共生矩阵窗口的灰度均值,反映图像的明暗深浅
相异性DIS
反映图像灰度的相异性
角二阶矩
反映图像灰度分布均匀程度和纹理粗细度
对比度CON
反映图像的清晰度和纹理的强度深浅
相关性COR
度量图像的灰度级在行或列方向上的相似程度
反差VAR
反映图像的局部差异性
同质性HOM
反映图像纹理的同质性
熵ENT
图像包含信息量的随机性度量
Tab.2 Texture measures calculated from GLCM
指标
8波段
8波段+植 被指数
8波段+灰度 共生矩阵
8波段+植被指 数+灰度共生矩阵
准确率
53.8
77.8
84.4
92.7
召回率
11.3
22.6
43.5
61.3
F1分数
18.7
35.0
57.4
73.8
Tab.3 Model’s accuracies with different feature sets(%)
Fig.5 Accuracies using GLCM with different window sizes
Fig.6 Cluster dendrogram
距离阈值
特征 个数
准确 率/%
召回 率/%
F1分 数/%
相对分类 能力/%
2.00
4
80.0
45.2
57.7
62.5
1.30
6
91.7
70.9
80.0
86.7
0.75
8
94.0
75.8
83.9
90.1
0.55
12
92.7
82.3
87.2
94.4
0.45
15
92.9
85.5
89.1
96.5
0.25
21
96.4
85.5
90.5
98.0
0.20
24
96.4
87.1
91.5
99.1
Tab.4 Relative capability of the classifier with new feature sets
Fig.7 Features with permutation importance greater than 0.02
Fig.8 Results of identifying unhealthy trees
Fig.9 Occurrence probability of unhealthy trees at different elevations
Fig.10 Causes of three different types of unhealthy trees
不健康树木成因
林区类型
不健康树木发生概率/%
人为扰动
人工林
0.17
自然过程
天然林
0.21
地质灾害
天然林
0.40
Tab.5 Occurrence probabilities of unhealthy trees with three causes
[1]
Batisse M. Action plan for biosphere reserves[J]. Environmental Conservation, 1985, 12(1):17-27.
[2]
Ishwaran N, Persic A, Tri N H. Concept and practice:The case of UNESCO biosphere reserves[J]. International Journal of Environment and Sustainable Development, 2008, 7(2):118-131.
[3]
Wang J, Feng L, Palmer P I, et al. Large Chinese land carbon sink estimated from atmospheric carbon dioxide data[J]. Nature, 2020, 586:720-723.
Tao H, Li C J, Cheng C, et al. Progress in remote sensing monitoring for pine wilt disease induced tree mortality:A review[J]. Forest Research, 2020, 33(3):172-183.
[5]
Dash J P, Watt M S, Pearse G D, et al. Assessing very high resolution UAV imagery for monitoring forest health during a simulated disease outbreak[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2017, 131:1-14.
Huang W J, Zhang J C, Shi Y, et al. Progress in monitoring and forecasting of crop pests and diseases by remote sensing[J]. Journal of Nanjing University of Information Science and Technology(Natural Science Edition), 2018, 10(1):30-43.
[7]
Meddens A J H, Hicke J A, Vierling L A, et al. Evaluating methods to detect bark beetle-caused tree mortality using single-date and multi-date Landsat imagery[J]. Remote Sensing of Environment, 2013, 132:49-58.
[8]
Wang H, Zhao Y, Pu R, et al. Mapping Robinia pseudoacacia forest health conditions by using combined spectral,spatial,and textural information extracted from IKONOS imagery and random forest classifier[J]. Remote Sensing, 2015, 7(7):9020-9044.
[9]
Waser L T, Küchler M, Jütte K, et al. Evaluating the potential of WorldView-2 data to classify tree species and different levels of ash mortality[J]. Remote Sensing, 2014, 6(5):4515-4545.
[10]
Näsi R, Honkavaara E, Lyytikäinen-Saarenmaa P, et al. Using UAV-based photogrammetry and hyperspectral imaging for mapping bark beetle damage at tree-level[J]. Remote Sensing, 2015, 7(11):15467-15493.
[11]
Yu R, Luo Y, Zhou Q, et al. Early detection of pine wilt disease using deep learning algorithms and UAV-based multispectral imagery[J]. Forest Ecology and Management, 2021, 497:119493.
[12]
Zhou Q, Yu L, Zhang X, et al. Fusion of UAV hyperspectral imaging and LiDAR for the early detection of EAB stress in ash and a new EAB detection index—NDVI(776,678)[J]. Remote Sensing, 2022, 14(10):2428.
Xue J, Yu L F, Lin Q N, et al. Using Sentinel-1 multi-temporal InSAR data to monitor the damage degree of shoot beetle in Yunnan pine forest[J]. Remote Sensing for Land and Resources, 2018, 30(4):108-114.doi: 10.6046/gtzyyg.2018.04.17.
[14]
He Y N, Chen G, Potter C, et al. Integrating multi-sensor remote sensing and species distribution modeling to map the spread of emerging forest disease and tree mortality[J]. Remote Sensing of Environment, 2019, 231:111238.
[15]
Wulder M A, Dymond C C, White J C, et al. Surveying mountain pine beetle damage of forests:A review of remote sensing opportunities[J]. Forest Ecology and Management, 2006, 221(1-3):27-41.
[16]
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(1):309.
[17]
Gitelson A A, Kaufman Y J, Merzlyak M N. Use of a green channel in remote sensing of global vegetation from EOS-MODIS[J]. Remote Sensing of Environment, 1996, 58(3):289-298.
[18]
Wójtowicz A, Piekarczyk J, Czernecki B, et al. A random forest model for the classification of wheat and rye leaf rust symptoms based on pure spectra at leaf scale[J]. Journal of Photochemistry and Photobiology B:Biology, 2021, 223:112278.
[19]
Eitel J U, Vierling L A, Litvak M E, et al. Broadband,red-edge information from satellites improves early stress detection in a New Mexico conifer woodland[J]. Remote Sensing of Environment, 2011, 115(12):3640-3646.
[20]
Chen G, He Y N, Santis A D, et al. Assessing the impact of emerging forest disease on wildfire using Landsat and KOMPSAT-2 data[J]. Remote Sensing of Environment, 2017, 195:218-229.
[21]
Wang H, Zhong Y, Pu R, et al. Dynamic analysis of Robinia pseudoacacia forest health levels from 1995 to 2013 in the Yellow River Delta,China using multitemporal Landsat imagery[J]. International Journal of Remote Sensing, 2018, 39(12):4232-4253.
[22]
Ferreira M P, Wagner F H, Aragão L E O C, et al. Tree species classification in tropical forests using visible to shortwave infrared WorldView-3 images and texture analysis[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2019, 149:119-131.
[23]
Eugenio F C, Silva S D P, Fantinel R A, et al. Remotely piloted aircraft systems to identify pests and diseases in forest species:The global state of the art and future challenges[J]. IEEE Geoscience and Remote Sensing Magazine, 2021, 10(1):320-333.
[24]
Dash J P, Pearse G D, Watt M S. UAV multispectral imagery can complement satellite data for monitoring forest health[J]. Remote Sensing, 2018, 10(8):1216.
[25]
Liakos K G, Busato P, Moshou D, et al. Machine learning in agriculture:A review[J]. Sensors, 2018, 18(8):2674.
[26]
Park H G, Yun J P, Kim M Y, et al. Multichannel object detection for detecting suspected trees with pine wilt disease using multispectral drone imagery[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14:8350-8358.
[27]
Hu G S, Wang T X, Wan M Z, et al. UAV remote sensing monitoring of pine forest diseases based on improved Mask R-CNN[J]. International Journal of Remote Sensing, 2022, 43(4):1274-1305.
[28]
Han Z M, Hu W J, Peng S L, et al. Detection of standing dead trees after pine wilt disease outbreak with airborne remote sensing imagery by multi-scale spatial attention deep learning and Gaussian kernel approach[J]. Remote Sensing, 2022, 14(13):3075.
[29]
Mantas V, Fonseca L, Baltazar E, et al. Detection of tree decline (Pinus pinaster Aiton) in European forests using Sentinel-2 data[J]. Remote Sensing, 2022, 14(9):2028.
[30]
Haralick R M, Shanmugam K, Dinstein I H. Textural features for image classification[J]. IEEE Transactions on Systems,Man,and Cybernetics, 1973, 6:610-621.
[31]
Janitza S, Strobl C, Boulesteix A L. An AUC-based permutation variable importance measure for random forests[J]. BMC Bioinformatics, 2013, 14:119.
doi: 10.1186/1471-2105-14-119
pmid: 23560875