基于NOAA-AVHRR数据的中国东部地区植被遥感分类研究

doi:10.17521/cjpe.2005.0058

摘要/Abstract

摘要：

该文采用 19幅 (时间跨 8个月 ) 时间序列的NOAA_AVHRR的归一化植被指数 (NDVI) 最大值合成影像遥感数据, 经过主分量分析 (Principlecomponentanalysis, PCA) 处理后, 用非监督分类方法的ISODATA算法, 对中国东部地区的 (五省一市 ) 植被进行分类, 结果可以分出 2 8种土地覆盖类型, 除了两种类型为水体和城市或裸地外, 其余 2 6种类型均为植被类型, 根据中国植被分类系统, 这 2 6类可以归并为 6大植被类型 :1) 常绿阔叶林 ;2 ) 针叶林 ;3) 竹林 ;4 ) 灌草丛 ;5 ) 水生植被 ;6 ) 农业植被。用 1∶10 0 0 0 0 0数字化《中国植被图集》的植被类型检验遥感分类结果表明, 针叶林、灌草丛、常绿阔叶林和农业植被的分类具有较高的位置精度和面积精度, 位置精度分别为 79.2 %、91.3%、6 8.2 %和 95.9%, 面积精度分别达到 92.1%、95.9%、6 3.8%和 90.5 %。这 6大植被类型在地理空间上的分布规律与中国东部常绿阔叶林区植被的地带性分布基本一致。

关键词: 植被分类, 遥感, NDVI, 主分量分析, 中国东部

Abstract:

East China lies in the subtropical monsoon climatic zone and is dominated by subtropical evergreen broad-leaved forests, a unique vegetation type on earth mainly distributed in East Asia with the largest and most concentrated distribution in China. It is important to be able to monitor and estimate forest biomass and production, regional carbon storage, and global climate change impacts of these important vegetation types. In this paper, we used coarse resolution remote sensing data to identify the vegetation types in East China and develop a map of the spatial distribution of vegetation types in this region. Nineteen maximum normalized difference vegetation index (NDVI) composite images (Acquisition time span 7 months from February through August), which were derived from 10-days of National Oceanographic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) channel 1 and channel 2 observations, and an unsupervised classification method, the ISODATA algorithm, was employed to identify the vegetation types. The image was processed using principal component analysis (PCA) to reduce the dimensions of the dataset resulting in a total of 28 spectral clusters of land cover of which 2 clusters were urban/bare soil and water. The 26 remaining spectral clusters were merged into 6 vegetation types: evergreen broad-leaved forest, coniferous forest, bamboo forest, shrub-grass, aquatic vegetation and agricultural vegetation using the Chinese vegetation taxonomy system. The spatial distribution and areal extent calculated for the coniferous forests, shrub-grass, evergreen broad-leaved forests and agricultural vegetation compared well with the Vegetation Atlas of China at a 1∶1 000 000 scale. The spatial accuracy for coniferous forests, shrub-grass, evergreen broad-leaved forests and agricultural vegetation was 79.2%, 91.3%, 68.2% and 95.9%, respectively, and the area accuracy was 92.1%, 95.9%, 63.8% and 90.5%, respectively. The spatial and area accuracy of the bamboo forest was 28.7% and 96.5%, the spatial accuracy of aquatic vegetation was 69.6%, but there is great error in its area accuracy because image acquisition did not cover the full year. Our research demonstrated the feasibility of using NOAA-AVHRR to identify the different vegetation types in the subtropical evergreen broad-leaved forest zone in East China. The spatial location of the 6 identified vegetation types coincided with the actual geographical distribution of the actual vegetation types in East China.

Key words: Vegetation classification, Remote sensing, NDVI, Principal component analysis, East China

李俊祥, 达良俊, 王玉洁, 宋永昌. 基于NOAA-AVHRR数据的中国东部地区植被遥感分类研究. 植物生态学报, 2005, 29(3): 436-443. DOI: 10.17521/cjpe.2005.0058

LI Jun-Xiang, DA Liang-Jun, WANG Yu-Jie, SONG Yong-Chang. VEGETATION CLASSIFICATION OF EAST CHINA USING MULTI-TEMPORAL NOAA-AVHRR DATA. Chinese Journal of Plant Ecology, 2005, 29(3): 436-443. DOI: 10.17521/cjpe.2005.0058

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.plant-ecology.com/CN/10.17521/cjpe.2005.0058

https://www.plant-ecology.com/CN/Y2005/V29/I3/436

图/表 5

图1 研究地区的地理位置

Fig.1 Location of the research area and its coverage

表1 19幅NDVI最大值合成影像经PCA处理后各波段的统计特征

Table 1 Statistical traits of 19 maximum NDVI composite images after PCA processing

	协相关特征向量 Correlation eigenvector	累积贡献率 Cumulative contribution	协方差特征向量 Correlation eigenvector	累积贡献率 Cumulative contribution
主成分1 PC1	18.620	98.000	247 763.44	98.058
主成分2 PC2	0.151	98.795	1 952.86	98.831
主成分3 PC3	0.058	99.100	741.93	99.124
主成分4 PC4	0.034	99.279	434.73	99.296
主成分5 PC5	0.026	99.416	329.55	99.427
主成分6 PC6	0.020	99.521	243.69	99.523
主成分7 PC7	0.014	99.595	191.44	99.599
主成分8 PC8	0.012	99.658	151.07	99.659
主成分9 PC9	0.011	99.716	132.53	99.711
主成分10 PC10	0.009	99.763	123.20	99.760
主成分11 PC11	0.008	99.805	107.16	99.823
主成分12 PC12	0.007	99.842	91.18	99.839
主成分13 PC13	0.006	99.874	85.65	99.873
主成分14 PC14	0.005	99.900	69.38	99.900
主成分15 PC15	0.005	99.926	68.20	99.927
主成分16 PC16	0.004	99.947	57.38	99.950
主成分17 PC17	0.004	99.968	50.92	99.970
主成分18 PC18	0.003	99.984	38.43	99.985
主成分19 PC19	0.003	100.000	37.69	100.000

图2 中国东部植被分类图 1:水体Water 2:水生植被Aquatic vegetation 3:水生/沼泽植被Aquatic or swamp vegetation 4:沼泽/滩涂植被类型ⅠSwamp orbeach vegetationⅠ5:沼泽/滩涂植被类型ⅡSwamp or beach vegetationⅡ6:常绿阔叶林类型ⅠEvergreen broad_leaved forest typeⅠ7:常绿阔叶林类型ⅡEvergreen broad_leaved forest typesⅡ8:针叶林Ⅰ杉木林ConiferousforestⅠ (Cunninghamia lanceolata) 9:针叶林Ⅱ马尾松林Needle_leaved forest typeⅡ (Pinus massoniana) 10:针叶林Ⅲ马尾松占优势Needle_leaved forest typeⅢ (Pinus dominated) 11:针阔叶混交林Needle_leaved broad_leaved mixed forest type 12:常绿果树或常绿灌丛Evergreen orchards or evergreen shrub types 13:针叶林ⅣConiferous forest typeⅣ14:矮灌草丛类型ⅠLow shrub grass typeⅠ15:矮灌草丛类型ⅡLow shrub grass typeⅡ16:矮灌草丛类型ⅢLow shrub grass typeⅢ17:矮灌草丛类型ⅣLow shrub grasstypeⅣ18:一年两熟旱作农业植被小麦+玉米One yeartwo ripe crop in dry farmland (wheat+corn) 19:城市/裸露沙滩Urban area or bare sand beach 20:一年两熟/二年三熟水稻+油菜/小麦:农业植被One year two or two year three ripe crop (rice+wheat/rape) 21:竹林Bamboo forest type 22:一年两熟/三熟水稻+油菜:农业植被One year two or three ripe crop (rice+rape) 23:一年两熟/三熟水稻为主:农业植被One year two or three ripe crop (rice dominated) 24:一年两熟水稻+水稻为主:农业植被One year two ripe crop (rice+rice) 25:一年两熟/三熟水稻需灌溉:农业植被One year two or three crop in irrigated land (rice) 26:落叶灌丛Deciduous shrub 27:一年两熟水稻占优势:农业植被One year two ripe crop (rice dominant) 28:山地灌草丛Montane brushwood

Fig.2 Map of vegetation classification in East China

表2 中国东部植被遥感分类结果类型精度评估

Table 2 Type accuracy assessment of the classified vegetation types in East China

	针叶林^* NLF	常绿阔叶林 EBLF	竹林 BF	灌草丛 SHG	水生植被 AV	农业植被 CV	落叶阔叶林 DLF	常绿落叶阔叶混交林 EDMF
遥感分类 (3×3滤波) 斑块数 Patches identified (3×3 filtered)	1 654	727	553	1 902	2 143	1 425	-	-
中国植被数字化斑块数 Patches digitalized	3 388	765	453	1 251	69	2 904	109	51
重合的斑块数 Patches overlaid	2 684	522	130	1 142	48	2 786	-	-
分类精度 Accuracy (%)	79.2	68.2	28.7	91.3	69.6	95.9

表3 中国东部植被遥感分类结果面积精度评估

Table 3 Area accuracy assessment of the classified vegetation types in East China

遥感分类类型 Identified vegetation type	遥感分类面积 Area of identified types (km²)	中国植被图类型 Digitalized vegetation types from Vegetation Atlas of China	数字化中国植被图类型面积 Area of digitalized types (km²)	面积误差 Errors of area (km²)	面积精度 Area accuracy (%)
针叶林NLF	150 292.0	针叶林NLF	163 184.5	-12 892.5	92.1
常绿阔叶林EBLF	35 427.0	常绿阔叶林EBLF	26 010.4	9 417.0	63.8
竹林BF	16 243.0	竹林BF	16 823.5	-580.0	96.5
灌草丛SHG	85 462.0	灌草丛SHG	89 067.0	-3 605.0	95.9
水生植被AV	5 555.0	水生植被AV	1 713.0	3 842.0	324.3
农业植被CV	286 328.0	农业植被CV	316 212.1	-29 884.0	90.5
水面^* Water	55 005.0	水面^* Water	22 034.4	32 971.0	249.6
城区/裸地* Urban/bare soil	4 288.0
		亚热带常绿落叶阔叶混交林EDMF	1108.9
		落叶阔叶林DBF	4563.8
合计 Total	638 600.0		639 000.0	-400.0	99.9

参考文献 30

[1]	Achard F, Estreguil C (1995). ForestclassificationofSoutheastA siausingNOAAAVHRRdata. RemoteSensingofEnvironment, 54,198-208.
[2]	Achard F, Eva H, Mayaux P (2001). Tropicalforestmappingfromcoarsespatialresolutionsatellitedata:productionandaccuracyassessmentissues. InternationalJournalofRemoteSensing, 22,2741-2762.
[3]	Boyd DS, Ripple WJ (1997). Potentialvegetationindicesfordeter miningglobalforestcover. InternationalJournalofRemoteSens ing, 18,1395-1401.
[4]	Cihlar J, Ly H, Xiao Q (1996). LandcoverclassificationwithAVHRRmultichannelcompositioninNortherenvironments. RemoteSensingofEnvironment, 58,36-51.
[5]	DeFries R, Townshend JRG (1994). NDVI_derivedlandcoverclassificationsonaglobalscale. InternationalJournalofRemoteSensing, 15,3567-3586.
[6]	Editorial Board of The Forest of Jiangxi Province (江西森林编辑委员会) (1986). TheForestofJiangxiProvince (江西森林). JiangxiScience&TechnologyPress, Nanchang. (inChinese).
[7]	Editorial Board of Vegetation of Anhui Province (安徽植被编辑委员会) (1983). VegetationofAnhuiProvince (安徽植被). An huiScienceandTechnologyPress, Hefei. (inChinese).
[8]	Ehrlich D, Estes JE, Singh A (1994). ApplicationofNOAA_AVHRR 1kmdataforenvironmentalmonitoring. InternationalJournalofRemoteSensing, 15,145-161.
[9]	Eidenshink JC, Faundeen JL (1994). The1kmAVHRRgloballanddataset:firststagesinimplementation. InternationalJour nalofRemoteSensing, 15,3443-3462.
[10]	Goward SN, Huemmrich KF (1992). VegetationcanopyPARab sorbanceandthenormalizeddifferencevegetationindex:anas sessmentusingtheSAILmodel. RemoteSensingofEnvironment, 39,119-140.
[11]	Hou XY (侯学煜) (2000). VegetationAtlasofChina (中国植被图集). SciencePress, Beijing. (inChinese).
[12]	Li XB (李晓兵), Shi PJ (史培军) (1999). Researchonregula tionofNDVIChangeofChineseprimaryvegetationtypesbasedonNOAA/AVHRRdata. ActaBotanicaSinica (植物学报), 41,314-324. (inChinesewithEnglishabstract).
[13]	Lin P (林鹏) (1990). VegetationofFujianProvince (福建植被). FujianScience&TechnologyPress, Fuzhou. (inChinese).
[14]	Mayaux P, Gond V, Bartholome E (2000). Anear_realtimeforestcovermapofMadagascarderivedfromSPOT -4VEGETATION (VGT) data. InternationalJournalofRemoteSensing, 21,3139-3144.
[15]	Moody A, Strahler AH (1994). CharacteristicsofcompositeAVHRRdataandproblemsintheirclassification. InternationalJournalofRemoteSensing, 15,3473-3491.
[16]	Running SW, Hunt ER, Nemani R, Glassy J (1994). MODISLAI (leafareaindex) andFPAR (fractionphotosyntheticallyactivera diation). MODISalgorithmdocument.NASA, 19.
[17]	Scepan J (1999). Thematicvalidationofhigh_resolutiongloballand_coverdatasets. PhotogrammetricEngineering&RemoteSensing, 65,1051-1060.
[18]	Sellers PJ (1987). Canopyreflectance, photosynthesisandtranspi rationⅡ.Theroleofbiophysicsinthelinearityoftheirinterde pendence. RemoteSensingofEnvironment, 21,143-183.
[19]	Sellers PJ, Turker CJ, Collatz GJ, Los SO, Justice CO, Dazlich DA, Randall DA (1994). Aglobal1°by1°NDVIdatasetforclimatestudies.Part2, Thegenerationofglobalfieldsofterres trialbiophysicalparametersfromtheNDVI. InternationalJournalRemoteSensing, 15,3519-3545.
[20]	Sheng YW (盛永伟), Chen WY (陈维英), Xiao QG (肖乾广) GuoL (郭亮) (1995). Chinesevegetationclassificationwithme teorologicalsatellitevegetationindexdata. ChineseScienceBul letin (科学通报), 40,68-71. (inChinese).
[21]	Song YC (宋永昌) (1999). PerspectiveofthevegetationzonationofforestregionineasternChina. ActaBotanicaSinica (植物学报), 41,541-552. (inChinesewithEnglishabstract).
[22]	The Forest of Zhejiang Province Compilation Committee Edition (浙江森林编辑委员会) (1993). TheForestofZhejiangProvince (浙江森林). ChinaForestryPublishingHouse, Beijing. (inChinese).
[23]	Townshend JRG (1994). GlobaldatasetsforlandapplicationfromAdvancedVeryHighResolutionRadiometer:anintroduction. InternationalJournalofRemoteSensing, 15,3319-3332.
[24]	Townshend JRG, Justice CO, Skole D, Malingreau JP, Cihlar J, Teiliet P, Sadowski F, Ruttenberg S (1994). The1kmresolu tionglobaldataset:needsofTheInternationalGeosphereBio sphereProgramme. InternationalJournalofRemoteSensing, 15,3417-3441.
[25]	Tucker CJ, Townshend JRG, Goff TE (1985a). Africaland_coverclassificationusingsatellitedata. Science, 227,369-375. DOI URL
[26]	Tucker CJ, Vanpraet CL, Sharman MJ, vanIttersum G (1985b). SatelliteremotesensingoftotalherbaceousbiomassproductionintheSenegaleseSahel:1980-1984. RemoteSensingofEnviron ment, 17,233-249.
[27]	USGS (2004). Globallandcovercharacteristicsdataset. Http://edcdaac.usgs.gov/glcc/globdoc2_0.html.Cited21October2004.
[28]	Wu ZY (吴征镒) (1980). VegetationofChina (中国植被). SciencePress, Beijing. (inChinese).
[29]	Xiao X, Boles S, Liu J, Zhuang D, Liu M (2002). Characteriza tionofforesttypesinNortheasternChina, usingmulti_temporalSPOT_4VEGETATIONsensordata. RemoteSensingofEnviron ment, 82,335-348.
[30]	Zhu Z, Evans DL (1994). U.S.foresttypesandpredictedpercentforestcoverfromAVHRRdata. PhotogrammetricEngineeringandRemoteSensing, 60,525-531.