基于植被指数的典型草原区生物量模型——以内蒙古锡林浩特市为例

doi:10.17521/cjpe.2007.0004

摘要/Abstract

摘要：

利用遥感估测地上生物量是国内外生态学与地理学的研究热点。但基于植被指数的生物量回归模型结果差异较大,究竟哪种植被指数与哪种模型更适合典型草原的生物量反演,是现代草地遥感急需解决的问题之一。该文基于TM影像数据的不同植被指数(VI)差异性,分别选取了RVI(比值植被指数)、NDVI(归一化差异植被指数)、SAVI(土壤调节植被指数)、MASVI(修改型土壤调整植被指数)和RSR(简化比率植被指数)5种植被指数,与同期的内蒙古典型草原区地面实测地上生物量做相关分析,分别建立了5种植被指数与地上生物量的线性及3种非线性(对数、二次多项式、三次多项式)回归模型。研究结果表明:对于中国北方典型草原区而言,地上生物量与5种植被指数(RVI、NDVI、SAVI、MSAVI和RSR)均呈现出显著相关,但地上生物量与后4种植被指数是正相关,与RVI为负相关;利用5种植被指数(RVI、NDVI、SAVI、MSAVI和RSR)监测草地植被生物量的复相关系数均大于0.6,充分说明利用植被指数检测典型草原生物量是一种简单可行的方法;NDVI建立的生物量回归模型,其复相关系数大于其它4类植被指数(RVI、SAVI、MSAVI和RSR),说明NDVI-生物量模型优于植被指数RVI、SAVI、MSAVI和RSR 模型,其模拟地表生物量的效果好;对于TM影像来说,植被生物量的线性模型与3种非线性模型(三次多项式生物量模型、二次多项式生物量模型、对数模型)都表现出较好的模拟效果,都通过了0.01的显著性检验,而且该研究的结果显示出三次多项式生物量回归模型最优,其次是二次多项式生物量模型,再次是线性模型,相对较差的是对数模型。通过NDVI-生物量三次多项式回归模型模拟锡林浩特草原的生物量,可以看出整个研究区的地上生物量基本上是东高西低、东南高西北低的趋势,这与研究区的地形、气候及土地利用等多种因素有关。

关键词: 植被指数, 典型草原, 地上生物量, 回归模型

Abstract:

Aims There is a crucial need in grassland study for a vegetation index (VI)-biomass model simulating steppe biomass based on remote sensing.

Methods Thematic mapper (TM) images (spatial resolution of 30 m× 30 m) for the research area in 2005 and 1991 were rectified so that geometric errors were less than one pixel, then extracted the image of the research region in the soft of ERDAS. We used five vegetation indexes:RVI (ratio vegetation index), NDVI (normalized difference vegetation index), SAVI (soil-adjusted vegetation index), MASVI (modified soil-adjusted vegetation index) and RSR (reduced simple ratio index). They were correlated to plant biomass sampled on the ground at the same time as the TM images. We developed four kinds of regression models: linear, logarithm, second-degree polynomial and cubic polynomial.

Important findings The correlations between sampled biomass and the five VIs were highly significant, with four (NDVI, SAVI, MSAVI, RSR) being positive and one (RVI) negative. Multiple correlation coefficients (R²) of the 15 regression models were >0.6, indicating that a VI-biomass regression model was a simple method to monitor the biomass of steppe grassland. The R² of the NDVI-biomass model was the highest, indicating that it was better suited to simulate the biomass of typical steppe than the other VIs. For TM image, all four kinds of models were significant at the 0.01 level, with the cubic polynomial model as the best to simulate the biomass, followed by the second-degree polynomial, linear and logarithm models. Therefore, the cubic polynomial regression model based on NDVI-biomass was the best model, and was used to simulate the biomass of the research region. Simulated biomass was higher in the east than in the west of the research region and higher in the southeast than in the northwest. Simulated biomass was consistent with sampled biomass in 2005.

Key words: vegetation index, typical steppe, biomass, regression model

李素英, 李晓兵, 莺歌, 符娜. 基于植被指数的典型草原区生物量模型——以内蒙古锡林浩特市为例. 植物生态学报, 2007, 31(1): 23-31. DOI: 10.17521/cjpe.2007.0004

LI Su-Ying, LI Xiao-Bing, YING Ge, FU Na. VEGETATION INDEXES-BIOMASS MODELS FOR TYPICAL SEMI-ARID STEPPE—A CASE STUDY FOR XILINHOT IN NORTHERN CHINA. Chinese Journal of Plant Ecology, 2007, 31(1): 23-31. DOI: 10.17521/cjpe.2007.0004

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.plant-ecology.com/CN/10.17521/cjpe.2007.0004

https://www.plant-ecology.com/CN/Y2007/V31/I1/23

图/表 8

图1 调查样地分布图

Fig.1 The map of research plot in typical steppe of Inner Mongolia

表1 植被指数表

Table 1 The vegetation index

中文名称 Chinese name	英文缩写 English abbreviation	计算公式 Formula	作者(年代) Author(year)
比值植被指数 Ratio vegetation index	RVI	R/NIR	Pearson et al., 1972
归一化差异植被指数 Normalized difference vegetation index	NDVI	(NIR-R)/(NIR+R)	Rouse et al., 1974
土壤调节植被指数 Soil-adjusted vegetation index	SAVI	$(NIR - R) (NIR + R + L)$ (1+L)	Huete, 1988
修改型土壤调整植被指数 Modified soil-adjusted vegetation index	MSAVI	$2 NIR + 1 - (2 NIR + 1) 2 - 8 (NIR - R)) 2$	Qi et al., 1994
简化比率指数 Reduced simple ratio	RSR	$NIR R$ (1- $MIR - MI R min MI R max - MI R min)$	Brown et al., 2000

表1 植被指数表

Table 1 The vegetation index

中文名称 Chinese name	英文缩写 English abbreviation	计算公式 Formula	作者(年代) Author(year)
比值植被指数 Ratio vegetation index	RVI	R/NIR	Pearson et al., 1972
归一化差异植被指数 Normalized difference vegetation index	NDVI	(NIR-R)/(NIR+R)	Rouse et al., 1974
土壤调节植被指数 Soil-adjusted vegetation index	SAVI	$(NIR - R) (NIR + R + L)$ (1+L)	Huete, 1988
修改型土壤调整植被指数 Modified soil-adjusted vegetation index	MSAVI	$2 NIR + 1 - (2 NIR + 1) 2 - 8 (NIR - R)) 2$	Qi et al., 1994
简化比率指数 Reduced simple ratio	RSR	$NIR R$ (1- $MIR - MI R min MI R max - MI R min)$	Brown et al., 2000

图2 地上生物量与TM植被指数的散点图 RVI、MDVI、SAVI、MAVI、RSR:见表1 See Table 1

Fig.2 The scatter map between the vegetation biomass and the vegetation index of TM

表2 地上生物量与TM影像植被指数的相关性表

Table 2 Correlations between biomass and vegetation index

	生物量 Biomass	比值植被指数 RVI	归一化差异植被指数 NDVI	土壤调节植被指数 SAVI	修改型土壤调整植被指数 MSAVI	简化比率植被指数 RSR
生物量Biomass	1	-0.786^**	0.797^**	0.788^**	0.787^**	0.782^**
比值植被指数RVI		1	-0.997^**	-0.994^**	-1.000^**	-0.938^**
归一化差异植被指数NDVI			1	0.996^**	0.998^**	0.951^**
土壤调节植被指数SAVI				1	0.994^**	0.945^**
修改型土壤调整植被指数MSAVI					1	0.939^**
简化比率植被指数RSR						1

表3 TM影像的植被指数与植被地上生物量的回归模型

Table 3 Regression model from the vegetation biomass and the vegetation index (VI) of TM image

年 Year	植被指数 VI	模型类型 Model types	模型 Model	复相关系数 Multiple correlation coefficient
2005 (n=22)	RVI	线性 Linear model	y=-7.565x+1 829.100	R²=0.617
		对数 Logarithm model	y=-1 565.600ln(x)+8 610.100	R²=0.624
		二次多项式Second-degree polynomial model	y=0.090x²-44.561x+5 639.400	R²=0.636
		三次多项式Cubic polynomial model	y=0.007x³-4.120x²+820.270x-53 430.000	R²=0.649
	NDVI	线性 Linear model	y=10.022x-1 308.000	R²=0.636
		对数 Logarithm model	y=1 575.100ln(x)-7 696.800	R²=0.631
		二次多项式Second-degree polynomial model	y=0.081x²-15.622x+716.520	R²=0.641
		三次多项式Cubic polynomial model	y=-0.016x³+7.450x²-1 182.000x+62 109.000	R²=0.654
	SAVI	线性 Linear model	y=6.708x-885.270	R²=0.621
		对数 Logarithm model	y=1 147.500ln(x)-5 635.900	R²=0.612
		二次多项式Second-degree polynomial model	y=0.043x²-8.245x+404.440	R²=0.629
		三次多项式Cubic polynomial model	y=-0.005x³ +2.681x²-464.400x+26 579.000	R²=0.644
	MSAVI	线性 Linear model	y=7.584x-1 055.2	R²=0.619
		对数 Logarithm model	y=1 305.800ln(x)-6 470.100	R²=0.609
		二次多项式Second-degree polynomial model	y =0.089x²-23.343x+1 632.400	R²=0.637
		三次多项式Cubic polynomial model	y=-0.007x³+3.691x²-651.320x+37 995.000	R²=0.650
	RSR	线性 Linear model	y=3.039x-733.520	R²=0.611
		对数 Logarithm model	y=1 027.500ln(x)-5 685.700	R²=0.615
		二次多项式Second-degree polynomial model	y=-0.010x²+9.835x-1 876.900	R²=0.617
		三次多项式Cubic polynomial model	y=0.001x³-0.512x²+179.380x-20 874.000	R²=0.623
1991 (n=18)	RVI	线性 Linear model	y=-5.593x+631.780	R²=0.763
		对数 Logarithm model	y=-447.920ln(x)+2 143.500	R²=0.757
		二次多项式Second-degree polynomial model	y=-0.010x²-4.033 1x+568.820	R²=0.764
		三次多项式Cubic polynomial model	y=-0.009x³+2.156x²-181.050x+5 315.200	R²=0.814
	NDVI	线性 Linear model	y=4.033x-515.400	R²=0.771
		对数 Logarithm model	y=687.180ln(x)-3 356.500	R²=0.773
		二次多项式Second-degree polynomial model	y=-0.007x²+6.362x-714.010	R²=0.772
		三次多项式Cubic polynomial model	y=0.003x³-1.748x²+301.240x-17 238.000	R²=0.832
	SAVI	线性 Linear model	y=4.115x-520.190	R²=0.724
		对数 Logarithm model	y=684.560ln(x)-3 334.000	R²=0.722
		二次多项式Second-degree polynomial model	y=0.002x²+3.582x-475.530	R²=0.724
		三次多项式Cubic polynomial model	y=0.003x³-1.607x²+269.170x-14 959.000	R²=0.768
	MSAVI	线性 Linear model	y=5.563x-745.670	R²=0.721
		对数 Logarithm model	y=919.220ln(x)-4 519.500	R²=0.724
		二次多项式Second-degree polynomial model	y=-0.023x²+13.199x-1 375.500	R²=0.723
		三次多项式Cubic polynomial model	y=0.012x³-5.791x²+960.330x-53 029.000	R²=0.799
	RSR	线性 Linear model	y=2.508x-190.730	R²=0.687
		对数 Logarithm model	y=376.490ln(x)-1 697.000	R²=0.699
		二次多项式Second-degree polynomial model	y=-0.009x²+5.271x-397.490	R²=0.695
		三次多项式Cubic polynomial model	y=0.001x³-0.550x²+83.298x-4 067.300	R²=0.757

表3 TM影像的植被指数与植被地上生物量的回归模型

Table 3 Regression model from the vegetation biomass and the vegetation index (VI) of TM image

年 Year	植被指数 VI	模型类型 Model types	模型 Model	复相关系数 Multiple correlation coefficient
2005 (n=22)	RVI	线性 Linear model	y=-7.565x+1 829.100	R²=0.617
		对数 Logarithm model	y=-1 565.600ln(x)+8 610.100	R²=0.624
		二次多项式Second-degree polynomial model	y=0.090x²-44.561x+5 639.400	R²=0.636
		三次多项式Cubic polynomial model	y=0.007x³-4.120x²+820.270x-53 430.000	R²=0.649
	NDVI	线性 Linear model	y=10.022x-1 308.000	R²=0.636
		对数 Logarithm model	y=1 575.100ln(x)-7 696.800	R²=0.631
		二次多项式Second-degree polynomial model	y=0.081x²-15.622x+716.520	R²=0.641
		三次多项式Cubic polynomial model	y=-0.016x³+7.450x²-1 182.000x+62 109.000	R²=0.654
	SAVI	线性 Linear model	y=6.708x-885.270	R²=0.621
		对数 Logarithm model	y=1 147.500ln(x)-5 635.900	R²=0.612
		二次多项式Second-degree polynomial model	y=0.043x²-8.245x+404.440	R²=0.629
		三次多项式Cubic polynomial model	y=-0.005x³ +2.681x²-464.400x+26 579.000	R²=0.644
	MSAVI	线性 Linear model	y=7.584x-1 055.2	R²=0.619
		对数 Logarithm model	y=1 305.800ln(x)-6 470.100	R²=0.609
		二次多项式Second-degree polynomial model	y =0.089x²-23.343x+1 632.400	R²=0.637
		三次多项式Cubic polynomial model	y=-0.007x³+3.691x²-651.320x+37 995.000	R²=0.650
	RSR	线性 Linear model	y=3.039x-733.520	R²=0.611
		对数 Logarithm model	y=1 027.500ln(x)-5 685.700	R²=0.615
		二次多项式Second-degree polynomial model	y=-0.010x²+9.835x-1 876.900	R²=0.617
		三次多项式Cubic polynomial model	y=0.001x³-0.512x²+179.380x-20 874.000	R²=0.623
1991 (n=18)	RVI	线性 Linear model	y=-5.593x+631.780	R²=0.763
		对数 Logarithm model	y=-447.920ln(x)+2 143.500	R²=0.757
		二次多项式Second-degree polynomial model	y=-0.010x²-4.033 1x+568.820	R²=0.764
		三次多项式Cubic polynomial model	y=-0.009x³+2.156x²-181.050x+5 315.200	R²=0.814
	NDVI	线性 Linear model	y=4.033x-515.400	R²=0.771
		对数 Logarithm model	y=687.180ln(x)-3 356.500	R²=0.773
		二次多项式Second-degree polynomial model	y=-0.007x²+6.362x-714.010	R²=0.772
		三次多项式Cubic polynomial model	y=0.003x³-1.748x²+301.240x-17 238.000	R²=0.832
	SAVI	线性 Linear model	y=4.115x-520.190	R²=0.724
		对数 Logarithm model	y=684.560ln(x)-3 334.000	R²=0.722
		二次多项式Second-degree polynomial model	y=0.002x²+3.582x-475.530	R²=0.724
		三次多项式Cubic polynomial model	y=0.003x³-1.607x²+269.170x-14 959.000	R²=0.768
	MSAVI	线性 Linear model	y=5.563x-745.670	R²=0.721
		对数 Logarithm model	y=919.220ln(x)-4 519.500	R²=0.724
		二次多项式Second-degree polynomial model	y=-0.023x²+13.199x-1 375.500	R²=0.723
		三次多项式Cubic polynomial model	y=0.012x³-5.791x²+960.330x-53 029.000	R²=0.799
	RSR	线性 Linear model	y=2.508x-190.730	R²=0.687
		对数 Logarithm model	y=376.490ln(x)-1 697.000	R²=0.699
		二次多项式Second-degree polynomial model	y=-0.009x²+5.271x-397.490	R²=0.695
		三次多项式Cubic polynomial model	y=0.001x³-0.550x²+83.298x-4 067.300	R²=0.757

图3 植被指数-生物量回归模型的复相关系数图 RVI、MDVI、SAVI、MAVI、RSR:见表1 See Table 1

Fig.3 Contrast of correlation parameters of regression model on different vegetation index (VI)

表4 生物量实测值与NDVI三次多项式模型模拟值的误差表

Table 4 Error contrast of the observation value and the simulation value based on the experiential model

	实测值 Observation value	模拟值^* Simulation value	差值(实测值-模拟值) Subtraction (observation value-simulation value)	百分比(%)(差值/实测值) Percentage (subtraction/observation value)
1	231.73	198.10	33.63	15.14
2	154.43	156.30	-1.87	-1.21
3	178.80	177.30	1.50	0.84
4	179.60	214.20	-34.59	-19.26
5	163.65	255.00	-91.35	-55.82
6	260.00	255.00	5.00	1.92
7	283.00	244.30	38.80	13.71
8	144.27	153.40	-9.13	-6.33
9	189.07	164.10	24.97	13.21
10	191.23	174.30	16.93	8.85
11	201.47	226.20	-24.73	-12.28
12	164.67	174.60	-9.93	-6.03
13	269.90	250.10	19.80	7.34
14	61.33	54.60	6.73	10.98

图4 基于NDVI-生物量三次多项式模型的锡林浩特地上生物量分布图

Fig.4 The map of biomass based on NDVI model in Xilinhot

参考文献 27

[1]	Brown L, Chen J, Leblanc S, Cihlar J (2000). A shortwave infrared modification to the simple ratio for LAI retrieval in boreal forests: an image and model analysis . Remote Sensing of Environment, 71,16-25.
[2]	Cheng Q (程乾), Huang JF (黄敬峰), Wang RC (王人潮) (2005). Preliminary analysis of difference between MODIS and NOAA/AVHRR vegetation indices. Bulletin of Science and Technology (科技通报), 21,205-209. (in Chinese with English abstract)
[3]	Fabio M, Stefano R, Lorenzo B, Gaetano Z (2003). Use of NOAA-AVHRR NDVI images for the estimation of dynamic fire risk in Mediterranean areas. Remote Sensing of Environment, 86,187-197.
[4]	Fernando S, Gong P, Manuel F (2005). Land cover assessment with MODIS imagery in southern African Miombo ecosystems. Remote Sensing of Environment, 98,429-441.
[5]	Hame T, Salli A, Andersson K (1997). A new methodology for the estimation of biomass of conifer dominated boreal forest using NOAA/AVHRR data. International Journal of Remote Sensing, 18,3211-3243.
[6]	Huang JF (黄敬峰), Wang XZ (王秀珍), Hu XB (胡新博) (1999). Studies on grass yield monitoring for different natural grassland types using remote sensing data in northern Xinjiang. Grassland of China (中国草地), (1),7-11. (in Chinese with English abstract)
[7]	Huete AR (1988). A soil adjusted vegetation index (SAVI). Remote Sensing of Environment, 25,295-309.
[8]	Lefsky MA, Harding D, Cohen WB (1999). Surface lidar remote sensing of basal area and biomass in deciduous forests of eastern Maryland, USA. Remote Sensing of Environment, 67,83-98.
[9]	Li J (李净), Wang J (王建) (2004). Estimating plant biomass in Junma Chang of Shandan County using Landsat TM data. Remote Sensing Technology and Application (遥感技术与应用), 19,343-347. (in Chinese with English abstract)
[10]	Li JL (李建龙), Jiang P (蒋平), Dai RL (戴若兰) (1998). Advances in study on the remote sensing technology and GPS and GIS integration systems in estimating grassland yield applications in the north of Xinjiang, China. Acta Ecologica Sinica (生态学报), 18,504-510. (in Chinese with English abstract)
[11]	Ma XZ (马秀枝), Wang YF (王艳芬), Wang SP (汪诗平), Wang JZ (王金枝), Li CS (李长生) (2005). Impacts of grazing on soil carbon fraction in the grasslands of Xilin River Basin, Inner Mongolia. Acta Phytoecologica Sinina (植物生态学报), 29,569-576. (in Chinese with English abstract)
[12]	Michael JH, Peter JV, Peter E, Graham ED (1999). Pasture land cover in eastern Australia from NOAA-AVHRR NDVI and classified Landsat TM. Remote Sensing of Environment, 67,32-50.
[13]	Miina R (2005). Retrieval of leaf area index for a coniferous forest by inverting a forest reflectance model. Remote Sensing of Environment, 99,295-303.
[14]	Niu ZC (牛志春), Ni SX (倪绍祥) (2003). Study on models for monitoring of grassland biomass around Qinghai Lake assisted by remote sensing. Acta Geographica Sinica (地理学报), 58,695-702. (in Chinese with English abstract)
[15]	Pieter SA, Clement A, Kjell AH, Bernt J, Andrew KS (2006). Improved monitoring of vegetation dynamics at very high latitudes: a new method using MODIS NDVI. Remote Sensing of Environment, 100,321-334.
[16]	Qi J, Chehbouni A, Huete AR, Kerr Y (1994). A modified soil adjusted vegetation index. Remote Sensing of Environment, 48,119-126.
[17]	Rouse JW, Haas RH, Schell JA, Deering DW (1974). Monitoring vegetation systems in the Great Plains with ERTS. In: Freden SC, Becker M eds. Proceedings of Third Earth Resources Technology Satellite—1Symposium. NASA SP-351, Greenbelt,310-317.
[18]	Steven MD, Biscoe PV, Jaggard KW (1983). Estimation of sugar beet productivity from reflection in the red and infrared spectral bands. International Journal of Remote Sensing, 4,325-334.
[19]	Sun R (孙睿), Zhu QJ (朱启疆) (2000). Distribution and seasonal change of net primary productivity in China from April, 1992 to March, 1993. Acta Geographica Sinica (地理学报), 55,36-45. (in Chinese with English abstract)
[20]	Todd SW, Hoffer RM, Milchunas DG (1998). Biomass estimation on grazed and ungrazed rangelands using spectral indices. International Journal Remote Sensing, 19,427-438.
[21]	Tong C (仝川), Yong WY (雍伟义), Wu YN (乌云娜), Zhao LQ (赵利清), Jiang C (姜超), Yong SP (雍世鹏) (2001). Change in the spatial structure of grassland vegetation in the Xilin River Basin from 1985 to 1999. Acta Scientiarum Naturalium Universitatis Neimongol (内蒙古大学学报), 32,562-566. (in Chinese with English abstract)
[22]	Wang H (王宏), Li XB (李晓兵), Yu HJ (余弘婧) (2006). Monitoring growing season of typical steppe in northern China based on NOAA/AVHRR NDVI data . Journal of Plant Ecology (植物生态学报), 30,365-374. (in Chinese with English abstract)
[23]	Wang XZ (王兮之), Du GZ (杜国桢), Liang TG (梁天刚), Dai RL (戴若兰), Wang G (王刚) (2001). The establishment of grassland productive assessment model and the determination of precipitation resource distribution pattern of Gannan Tibetan Autonomous prefecture based on RS and GIS. Acta Prataculturae Sinica (草业学报), 10,95-102. (in Chinese with English abstract)
[24]	Wang ZX (王正兴), Liu C (刘闯), Zhao BR (赵冰茹), Liu AJ (刘爱军) (2005). ANPP estimate from MODIS-EVI for the grassland region of Xilingol, China. Journal of Lanzhou University (Natural Sciences) (兰州大学学报(自然科学版)), 41(2),10-16. (in Chinese with English abstract)
[25]	Xiao QG (肖乾广), Zhou SS (周嗣松), Chen WY (陈维英), Zhang LX (张丽霞), Lin CG (赁常恭), Xiao SZ (肖淑招), Meng XY (盂宪钺), Zhao ZK (赵忠凯), Zhang GZ (张桂宗) (1986). Productivity estimate using meteorological satellite. Journal of Remote Sensing (遥感学报), 1,37-43. (in Chinese with English abstract)
[26]	Yu F (喻锋), Li XB (李晓兵), Wang H (王宏), Yu HJ (余弘婧), Chen JH (陈云浩) (2005). Land cover classification in China based on the NDVI-Ts feature space. Acta Phytoecologica Sinica (植物生态学报), 29,934-944. (in Chinese with English abstract)
[27]	Zhao MS, Faith AH, Ramakrishna RN, Steven W (2005). Improvements of the MODIS terrestrial gross and net primary production global data set. Remote Sensing of Environment, 95,164-176.