LAND COVER CLASSIFICATION IN CHINA BASED ON THE NDVI-T<sub>S</sub> FEATURE SPACE

doi:10.17521/cjpe.2005.0115

Abstract

Abstract:

Mapping and quantifying land use and land cover changes are important for evaluating recent changes in the regional and global environment and for simulating future changes under different climate change and human land use scenarios. The normalized difference vegetation index (NDVI) and surface temperature (T_s) are two important parameters used to describe the characteristics of land cover. The NDVI-T_s feature space combines these two parameters into one variable. By and large, T_s/NDVI is synchronized to the growing season of vegetation so it can approximate the different phases and status of vegetation growth. Compared to NDVI and T_s, NDVI-T_s contains more land cover information and should be more suitable for characterizing the distribution of vegetation or land cover.
The aim of this paper was to discuss the feasibility of using the NDVI-T_s feature space to better characterize the current distribution and changes in vegetation and land cover in China. We used several classification methods, including Principal Component Analysis (PCA), unsupervised classification and post-classification sustained based on digital elevation models (DEM). The results indicated that T_s/NDVI was highly sensitive and could discriminate different vegetation cover categories in China at large-scales. The accuracy of vegetation classification based on T_s/NDVI was 72.0%, a 3.3% improvement in accuracy as compared to NDVI images, using unsupervised classification, and the Kappa value increased 0.020 2. Moreover, because of the simplexity of remote sensing information, the classification based on seasonal T_s/NDVI data could not avoid completely the mixed classification phenomena. It was necessary to add other information, reflecting vegetation characteristic and its environment to implement post-classification, such as DEM.
When the inversing accuracy of T_s improved, T_s/NDVI data can precisely describe the status of vegetation or land cover in China and improve monitoring of land use changes. This technique has great research potential and practical value.

Key words: T_s, NDVI, T_s/NDVI, Land cover classification

YU Feng, LI Xiao-Bing, WANG Hong, YU Hong-Jing, CHEN Yun-Hao. LAND COVER CLASSIFICATION IN CHINA BASED ON THE NDVI-T<sub>S</sub> FEATURE SPACE[J]. Chin J Plant Ecol, 2005, 29(6): 934-944.

Add to citation manager EndNote|Ris|BibTeX

URL: https://www.plant-ecology.com/EN/10.17521/cjpe.2005.0115

https://www.plant-ecology.com/EN/Y2005/V29/I6/934

Figures/Tables 6

References 29

[1]	Becker F, Li ZL (1990). Towards a local split window method over land surface. International Journal of Remote Sensing, 11,369-393.
[2]	Carpenter GA, Gjaja MN, Gopal S, Woodcock CE (1997). ART neural networks for remote sensing: vegetation classification from Landsat TM and terrain data. IEEE Transactions on Geoscience and Remote Sensing, 35,308-325.
[3]	Chen YH (陈云浩), Li XB (李晓兵), Shi PJ (史培军), Dou W (窦闻), Li X (李霞) (2003). Intra-annual vegetation change characteristics in the NDVI-T _s space application to Farming-Pastoral Zone in North China. Acta Botanica Sinica (植物学报), 45,1139-1145. (in Chinese with English abstract)
[4]	Cihlar J, Xiao QH, Chen J, Beaubien J, Fung K, Latifovic R (1998). Classification by progressive generalization: a new automated methodology for remote sensing multichannel data. International Journal of Remote Sensing, 19,2685-2704.
[5]	Du MY (杜明义), Wu WB (武文波), Guo DZ (郭达志) (2002). Research on multi-source geographic information based classification of desertification. Journal of Image and Graphics (中国图象图形学报), 7,740-743. (in Chinese with English abstract)
[6]	Ehrlich D, Estes JE, Singh A (1994). Applications of NOAA-AVHRR 1 km data for environmental monitoring. International Journal of Remote Sensing, 15,145-161.
[7]	Foody GM (1998). Sharpening fuzzy classification output to refine the representation of sub-pixel land cover distribution. International Journal of Remote Sensing, 19,2593-2599.
[8]	Hansen M, Dubayah R, DeFries R (1996). Classification trees: an alternative to traditional land cover classifiers. International Journal of Remote Sensing, 17,1075-1081.
[9]	Holben BN (1986). Characteristics of maximum-value composite images for temporal AVHRR data. International Journal of Remote Sensing, 7,1435-1445.
[10]	Josef C, Ly H, Li QZ, Chen J, Hartely P, Huang FT (1997). Multitemporal, mulitichannel AVHRR data sets for land biosphere studies-artifacts and corrections. Remote Sensing of Environment, 60,35-57.
[11]	Lambin EF, Ehrlich D (1995). Combing vegetation indices and surface temperature for land-cover mapping at broad spatial scales. International Journal of Remote Sensing, 16,573-579.
[12]	Lambin EF, Ehrlich D (1996). The surface temperature-vegetation index space for land cover and land-cover change analysis. International Journal of Remote Sensing, 17,463-487.
[13]	Lambin EF, Ehrlich D (1997). Land-cover change in Sub-Saharan Africa (1982-1991): application of a change index based on remotely sensed surface temperature and vegetation indices at a continental scale. Remote Sensing of Environment, 61,181-200.
[14]	Land Processes Distributed Active Archive Center LPDAAC (2004). Global land cover characterization database. http://edcdaac.usgs.gov/glcc/.Cited 20 Jun 2004.
[15]	Lark RM (1995). A reappraisal of unsupervised classification Ⅱ. Optimal adjustment of the map legend and a neigbourhood approach for mapping legend units. International Journal of Remote Sensing, 16,1445-1460.
[16]	Li DF (李道峰), Tian Y (田英), Hao FH (郝芳华) (2003). NDVI data based study on complex classification of vegetation cover of Yellow River Basin. Research of Soil and Water Conservation (水土保持研究), 10,88-91. (in Chinese with English abstract)
[17]	Li XB (李晓兵), Chen YH (陈云浩), Yu F (喻锋) (2004). Global and regional cover mapping from remote sensing data: status quo, strategies and trends. Advance in Earth Sciences, 19(1),71-80. (in Chinese with English abstract)
[18]	Loveland TR, Merchant JW, Ohlen DO, Brown J (1991). Development of a land-cover database for the conterminous U.S. Photogrammetric Engineering and Remote Sensing, 57,1453-1463.
[19]	Meer F (1995). Spectral unmixing of landsat the matci mapper data. International Journal of Remote Sensing, 16,3189-3194.
[20]	NASA. Pathfinder Advanced Very High Resolution Radiometer(AVHRR) Land FTP Data (2004). http://daac.gsfc.nasa.gov/. Cited 20 Jun 2004.
[21]	Pan YZ (潘耀忠), Li XB (李晓兵), He CY (何春阳) (2000). Research on comprehensive land cover classification in China: Based on NOAA/AVHRR and Holdridge PE index. Quaternary Sciences (第四纪研究), 20,270-281. (in Chinese with English abstract)
[22]	Qin ZH (覃志豪), Arnon K (2001). Split window algorithms for retrieving land surface temperature from NOAA-AVHRR data. Remote Sensing for Land & Resources (国土资源遥感), 2,33-42. (in Chinese with English abstract)
[23]	Running SW, Loveland TR, Pierce LL, Nemani RR, Hunt ER (1995). A remote sensing based vegetation classification logic for global land cover analysis. Remote Sensing of Environment, 51,39-48.
[24]	Saunders RW, Kriebel KT (1988). An improved method for detecting clear sky and cloudy radiances from AVHRR data. International Journal of Remote Sensing, 9,123-150.
[25]	Townshend JR, Justice CO, Kalb V (1987). Characterization and classification of South American land cover types using satellite data. International Journal of Remote Sensing, 8,1189-1207.
[26]	Tucker CJ, Townshend JR, Goff TE (1985). African land-cover classification using satellite data. Science, 227,369-375. URL PMID
[27]	Wu ZY (吴征缢) (1980). Vegetation of China (中国植被). Science Press, Beijing. (in Chinese)
[28]	Yan H (延昊), Wang CY (王长耀), Niu Z (牛铮), Jiang XG (姜小光), Wang W (王汶) (2001). Bio-implication of Principal Component Analysis to land cover using multitemporal AVHRR data. Remote Sensing Technology and Application (遥感技术与应用), 16,209-213. (in Chinese with English abstract)
[29]	Yang ZY (杨正宇), Zhou GS (周广胜), Yang DA (杨奠安) (2003). Comparison of simulated vegetation distribution in China produced by four popular climate-vegetation classification models. Acta Phytoecologica Sinica (植物生态学报), 27,587-593. (in Chinese with English abstract)

主成分 Principal component		NDVI、T_s和T_s/NDVI主成分变换特征向量 Eigenvectors of principal components for NDVI,T_s and T_s/NDVI												方差(%) Expressed variance
主成分 Principal component		1	2	3	4	5	6	7	8	9	10	11	12	方差(%) Expressed variance
NDVI	PC₁	0.26	0.26	0.27	0.27	0.29	0.31	0.32	0.32	0.31	0.29	0.28	0.27	81
	PC₂	0.25	0.32	0.36	0.13	-0.12	-0.33	-0.40	-0.37	-0.26	0.05	0.25	0.37	14.7
	PC₃	-0.44	-0.15	0.07	-0.54	-0.08	0.31	-0.05	-0.17	0.01	0.36	0.36	0.29	1.5
	PC₄	-0.05	-0.14	-0.16	0.29	0.67	0.35	-0.22	-0.41	-0.25	0.04	0.03	-0.13	1.4
T_s	PC₁	0.13	0.19	0.25	0.32	0.36	0.37	0.36	0.36	0.33	0.28	0.20	0.15	96.5
	PC₂	-0.52	-0.41	-0.21	0.01	0.17	0.23	0.23	0.20	0.11	-0.04	-0.29	-0.49	3.0
T_s/NDVI	PC₁	0.13	0.19	0.27	0.33	0.37	0.37	0.36	0.35	0.32	0.28	0.20	0.15	96.3
	PC₂	-0.54	-0.43	-0.20	-0.01	0.15	0.21	0.24	0.22	0.13	-0.01	-0.27	-0.47	2.8

主成分 Principal component		NDVI、T_s和T_s/NDVI主成分变换特征向量 Eigenvectors of principal components for NDVI,T_s and T_s/NDVI												方差(%) Expressed variance
主成分 Principal component		1	2	3	4	5	6	7	8	9	10	11	12	方差(%) Expressed variance
NDVI	PC₁	0.26	0.26	0.27	0.27	0.29	0.31	0.32	0.32	0.31	0.29	0.28	0.27	81
	PC₂	0.25	0.32	0.36	0.13	-0.12	-0.33	-0.40	-0.37	-0.26	0.05	0.25	0.37	14.7
	PC₃	-0.44	-0.15	0.07	-0.54	-0.08	0.31	-0.05	-0.17	0.01	0.36	0.36	0.29	1.5
	PC₄	-0.05	-0.14	-0.16	0.29	0.67	0.35	-0.22	-0.41	-0.25	0.04	0.03	-0.13	1.4
T_s	PC₁	0.13	0.19	0.25	0.32	0.36	0.37	0.36	0.36	0.33	0.28	0.20	0.15	96.5
	PC₂	-0.52	-0.41	-0.21	0.01	0.17	0.23	0.23	0.20	0.11	-0.04	-0.29	-0.49	3.0
T_s/NDVI	PC₁	0.13	0.19	0.27	0.33	0.37	0.37	0.36	0.35	0.32	0.28	0.20	0.15	96.3
	PC₂	-0.54	-0.43	-0.20	-0.01	0.15	0.21	0.24	0.22	0.13	-0.01	-0.27	-0.47	2.8

分类指标 Index for classification	GPS取样区 Sampling area for GPS spots	总点数 Number of total spots	正确点 Number of correct	错(混)分点 Number of incongruous (mixed)	精度 Accuracy(%)
T_s/NDVI	海南地区 Hainan region	79	50	29	63.3
	内蒙古地区 Inner Mongolia region	219	163	56	74.4
	新疆地区 Xinjiang region	282	201	81	71.3
	东北地区 Northeast region	87	66	21	75.9
	整个研究区 The whole study region	667	480	187	72.0
NDVI	海南地区 Hainan region	79	54	25	68.4
	内蒙古地区 Inner Mongolia region	219	152	67	69.4
	新疆地区 Xinjiang region	282	192	90	68.1
	东北地区 Northeast region	87	60	27	69.0
	整个研究区 The whole study region	667	458	209	68.7

分类指标 Index for classification	GPS取样区 Sampling area for GPS spots	总点数 Number of total spots	正确点 Number of correct	错(混)分点 Number of incongruous (mixed)	精度 Accuracy(%)
T_s/NDVI	海南地区 Hainan region	79	50	29	63.3
	内蒙古地区 Inner Mongolia region	219	163	56	74.4
	新疆地区 Xinjiang region	282	201	81	71.3
	东北地区 Northeast region	87	66	21	75.9
	整个研究区 The whole study region	667	480	187	72.0
NDVI	海南地区 Hainan region	79	54	25	68.4
	内蒙古地区 Inner Mongolia region	219	152	67	69.4
	新疆地区 Xinjiang region	282	192	90	68.1
	东北地区 Northeast region	87	60	27	69.0
	整个研究区 The whole study region	667	458	209	68.7