不同地形条件下植被盖度信息提取技术研究

doi:10.3724/SP.J.1258.2013.00002

摘要/Abstract

摘要：

为系统地研究特定区域的植被盖度信息提取技术, 在不同的地形条件下, 比较了目前流行的多种高光谱遥感植被盖度提取方法。结果表明: 最优高光谱归一化植被指数(NDVI₁)的建模和验证精度均高于其他两种归一化植被指数(NDVI), 直接采用NDVI建立的回归模型对研究区植被盖度的估测能力低于像元二分模型; 阴坡的最佳模型为基于一阶微分的偏最小二乘回归模型(PLSR模型), 其建模决定系数(R²)为0.810, 均方根误差(RMSE)为6.29, 验证R²为0.773, RMSE为8.85; 阳坡的最佳模型为基于二阶微分的PLSR模型, 其建模R²为0.823, RMSE为6.04, 验证R²为0.801, RMSE为7.35; 平原的最佳模型为全受限的线性光谱混合分解模型(FCLS), 其验证R²为0.852, RMSE为5.86。

关键词: 干旱半干旱, 高光谱, 模型, 植被盖度

Abstract:

Aims We compared a variety of vegetation cover extraction methods by hyperspectral remote sensing that are currently popular. Our aims were to systematically study vegetation information extraction technology and provide a reference for further study of vegetation cover.
Methods The methods of vegetation index, dimidiate pixel model, principal component regression (PCR), partial least squares regression first order differential (PLSR) and linear spectral mixture decomposition model were used to extract vegetation information.
Important findings The vegetation cover estimation ability of dimidiate pixel models established by different normalized differential vegetation index (NDVI) was higher than that of the regression models established by NDVI directly. The optimization model for shady slopes was PLSR model based on the first order differential (FD) with modeling R² = 0.810, root mean square error (RMSE) = 6.29 and validation R² = 0.773, RMSE = 8.85. The optimization model for sunny slopes was PLSR model based on the second order differential (SD), with modeling R² = 0.823, RMSE = 6.04 and validation R² = 0.801, RMSE = 7.35. The optimization model for plains was full confine linear spectral mixture model (FCLS), with validation R² = 0.852, RMSE = 5.86.

Key words: arid and semi-arid, high spectrum, model, vegetation coverage

吴见, 刘民士, 李伟涛. 不同地形条件下植被盖度信息提取技术研究. 植物生态学报, 2013, 37(1): 18-25. DOI: 10.3724/SP.J.1258.2013.00002

WU Jian, LIU Min-Shi, LI Wei-Tao. Research on vegetation cover information extraction technologies under different terrain conditions. Chinese Journal of Plant Ecology, 2013, 37(1): 18-25. DOI: 10.3724/SP.J.1258.2013.00002

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.plant-ecology.com/CN/10.3724/SP.J.1258.2013.00002

https://www.plant-ecology.com/CN/Y2013/V37/I1/18

图/表 13

表1 不同归一化植被指数(NDVI)估测阴坡植被盖度的建模和验证结果

Table 1 Modeling and verification results of shady slope vegetation coverage estimation of different normalized differential vegetation index (NDVI)

	NIR (nm)	Red (nm)	建模 Modeling		验证 Verification
	NIR (nm)	Red (nm)	R²	RMSE	R²	RMSE
NDVI₁	823.65	701.55	0.656	10.73	0.638	11.74
NDVI₂	803.3	671.02	0.508	14.65	0.497	14.40
NDVI₃	TM4	TM3	0.483	15.01	0.476	15.29

表2 不同归一化植被指数(NDVI)估测阳坡植被盖度的建模和验证结果

Table 2 Modeling and verification results of sunny slope vegetation coverage estimation of different normalized differential vegetation index (NDVI)

	NIR (nm)	Red (nm)	建模 Modeling		验证 Verification
	NIR (nm)	Red (nm)	R²	RMSE	R²	RMSE
NDVI₁	823.65	701.55	0.731	8.85	0.711	9.94
NDVI₂	803.3	671.02	0.712	9.87	0.682	10.73
NDVI₃	TM4	TM3	0.713	9.96	0.708	10.69

表3 不同归一化植被指数(NDVI)估测平原植被盖度的建模和验证结果

Table 3 Modeling and verification results of plain vegetation coverage estimation of each normalized differential vegetation index (NDVI)

	NIR (nm)	Red (nm)	建模 Modeling		验证 Verification
	NIR (nm)	Red (nm)	R²	RMSE	R²	RMSE
NDVI₁	823.65	701.55	0.746	7.09	0.732	8.95
NDVI₂	803.3	671.02	0.723	7.84	0.706	10.47
NDVI₃	TM4	TM3	0.705	9.12	0.691	10.70

图1 最优高光谱归一化植被指数(NDVI1)的阴坡植被盖度预测值验证结果。RMSE, 均方根误差。

Fig. 1 Verification results of the prediction value of shady slope vegetation coverage by the optimal hyper spectral normalized differential vegetation index (NDVI1). RMSE, root mean square error.

表4 不同地形条件下不同归一化植被指数(NDVI)的像元二分模型验证结果

Table 4 Verification results of pixel binary model of different normalized differential vegetation index (NDVI) under different terrain conditions

	NDVI_soil	NDVI_veg	阴坡 Shady slope		阳坡 Sunny slope		平原 Plain
	NDVI_soil	NDVI_veg	R²	RMSE	R²	RMSE	R²	RMSE
NDVI₁	-0.10	0.64	0.656	10.38	0.730	9.15	0.758	8.06
NDVI₂	-0.06	0.62	0.537	12.69	0.717	10.06	0.725	9.90
NDVI₃	-0.02	0.55	0.504	14.17	0.684	11.23	0.712	10.34

图2 基于最优高光谱NDVI(NDVI1)的像元二分模型阴坡植被盖度预测值验证结果。RMSE, 均方根误差。

Fig. 2 Verification results of the prediction value of shady slope vegetation coverage of pixel binary model based on the optimal high spectral NDVI (NDVI1). RMSE, root mean square error.

表5 不同光谱处理方法主成分回归(PCR)模型阴坡植被盖度建模和验证结果

Table 5 Modeling and verification results of shady slope vegetation coverage by principal component regression (PCR) models based on different spectral processing methods

光谱处理方法 Spectrum processing method	变量个数 Variable number	建模 Modeling		验证 Verification
光谱处理方法 Spectrum processing method	变量个数 Variable number	R²	RMSE	R²	RMSE
无变换数据 NO	8	0.693	11.28	0.667	11.95
一阶微分 FD	7	0.748	9.55	0.714	11.03
二阶微分 SD	9	0.732	10.17	0.708	11.15
包络线去除 CR	9	0.794	8.49	0.753	9.06

表6 不同光谱处理方法主成分回归(PCR)模型阳坡植被盖度建模和验证结果

Table 6 Modeling and verification results of sunny slope vegetation coverage by principal component regression (PCR) model based on different spectral processing methods

光谱处理方法 Spectrum processing method	变量个数 Variable number	建模 Modeling		验证 Verification
光谱处理方法 Spectrum processing method	变量个数 Variable number	R²	RMSE	R²	RMSE
无变换数据 NO	10	0.739	10.26	0.641	12.76
一阶微分 FD	12	0.766	9.14	0.685	11.57
二阶微分 SD	8	0.775	8.90	0.639	12.88
包络线去除 CR	10	0.753	9.67	0.650	12.34

表7 不同光谱处理方法主成分回归(PCR)模型平原植被盖度建模和验证结果

Table 7 Modeling and verification results of plain vegetation coverage by principal component regression (PCR) model based on different spectral processing methods

光谱处理方法 Spectrum processing method	变量个数 Variable number	建模 Modeling		验证 Verification
光谱处理方法 Spectrum processing method	变量个数 Variable number	R²	RMSE	R²	RMSE
无变换数据 NO	7	0.745	10.06	0.703	11.68
一阶微分 FD	10	0.759	9.71	0.726	10.04
二阶微分 SD	11	0.681	11.93	0.659	12.48
包络线去除 CR	9	0.730	10.51	0.717	11.45

表8 不同光谱处理方法偏最小二乘回归(PLSR)模型阴坡植被盖度建模和验证结果

Table 8 Modeling and verification results of shady slope vegetation coverage by partial least squares regression (PLSR) model based on different spectral processing methods

光谱处理方法 Spectrum processing method	变量个数 Variable number	建模 Modeling		验证 Verification
光谱处理方法 Spectrum processing method	变量个数 Variable number	R²	RMSE	R²	RMSE
无变换数据 NO	9	0.742	10.64	0.718	11.34
一阶微分 FD	13	0.810	6.29	0.773	8.85
二阶微分 SD	10	0.794	7.78	0.761	9.16
包络线去除 CR	8	0.726	10.15	0.692	11.54

表9 不同光谱处理方法偏最小二乘回归(PLSR)模型阳坡植被盖度建模和验证结果

Table 9 Modeling and verification results of sunny slope vegetation coverage by partial least squares regression (PLSR) model based on different spectral processing methods

光谱处理方法 Spectrum processing method	变量个数 Variable number	建模 Modeling		验证 Verification
光谱处理方法 Spectrum processing method	变量个数 Variable number	R²	RMSE	R²	RMSE
无变换数据 NO	11	0.783	8.18	0.756	9.84
一阶微分 FD	7	0.790	7.96	0.758	9.23
二阶微分 SD	9	0.823	6.04	0.801	7.35
包络线去除 CR	8	0.731	10.27	0.692	11.49

表10 不同光谱处理方法偏最小二乘回归(PLSR)模型平原植被盖度建模和验证结果

Table 10 Modeling and verification results of plain vegetation coverage by partial least squares regression (PLSR) model based on different spectral processing methods

光谱处理方法 Spectrum processing method	变量个数 Variable number	建模 Modeling		验证 Verification
光谱处理方法 Spectrum processing method	变量个数 Variable number	R²	RMSE	R²	RMSE
无变换数据 NO	10	0.765	8.90	0.743	10.21
一阶微分 FD	8	0.804	7.22	0.809	6.95
二阶微分 SD	7	0.836	6.13	0.760	8.57
包络线去除 CR	11	0.773	9.15	0.739	10.38

表11 不同地形条件下全受限的LSMM (FCLS)和非受限的LSMM (LS)模型分解的植被分量精度验证结果

Table 11 Accuracy verification results of vegetation component unmixed by full confining linear spectral mixture model (FCLS) and linear spectral mixture model (LS) under different terrain conditions

分解模型 Decomposition model	验证 Verification
	阴坡 Shady slope		阳坡 Sunny slope		平原 Plain
	R²	RMSE	R²	RMSE	R²	RMSE
LS	0.618	14.59	0.709	11.92	0.746	10.37
FCLS	0.653	12.87	0.761	8.45	0.852	5.86

参考文献 13

[1]	Asner GP, Lobell DB (2000). A biogeophysical approach for automated SWIR unmixing of soils and vegetation. Remote Sensing of Environment, 74, 99-112.
[2]	Blackburn GA (1998). Quantifying chlorophylls and caroteniods at leaf and canopy scales: an evaluation of some hyperspectral approaches. Remote Sensing of Environment, 66, 273-285.
[3]	Cheng HF, Zhang WB, Chen F (2008). Advances in researches on application of remote sensing method to estimating vegetation coverage. Remote Sensing for Land & Resources, 20(1), 13-18. (in Chinese with English abstract)
	[ 程红芳, 章文波, 陈锋 (2008). 植被覆盖度遥感估算方法研究进展. 国土资源遥感, 20(1), 13-18.]
[4]	Drake NA, Mackin S, Settle JJ (1999). Mapping vegetation, soils, and geology in semi-arid shrublands using spectral matching and mixture modeling of SWIR AVIRIS imagery. Remote Sensing of Environment, 68, 12-25.
[5]	Du MY, Wu WB, Guo ZD (2002). Research on multi-source geographic information based classification of desertification. Journal of Image and Graphics, 7, 740-743. (in Chinese with English abstract)
	[ 杜明义, 武文波, 郭达志 (2002). 多源地学信息在土地荒漠化遥感分类中的应用研究. 中国图形图像学报, 7, 740-743.]
[6]	Elvidge CD, Chen ZK (1995). Comparison of broad-band and narrow-band red and near-infrared vegetation indices. Remote Sensing of Environment, 54, 38-48.
[7]	Hansen MC, Defries RS, Townshend JRG, Sohlberg R, Dimiceli C, Carroll M (2002). Towards an operational MODIS continuous field of percent tree cover agorithm: examples using AVHRR and MODIS data. Remote Sensing of Environment, 83, 303-319. (in Chinese with English abstract)
[8]	Kremer RG, Running SW (1993). Community type differentiation using NOAA/AVHRR data within a sagebrush-steppe ecosystem. Remote Sensing of Environment, 46, 311-318.
[9]	Li XS, Gao ZH, Li ZY, Bai LN, Wang BY (2010). Estimation of sparse vegetation coverage in arid region based on hy- perspectral mixed pixel decomposition. Chinese Journal of Applied Ecology, 21, 152-158. (in Chinese with English abstract)
	[ 李晓松, 高志海, 李增元, 白黎娜, 王琫瑜 (2010). 基于高光谱混合像元分解的干旱地区稀疏植被覆盖度优测. 应用生态学报, 21, 152-158.]
[10]	Li XS, Li ZY, Gao ZH, Bai LN, Wang BY (2011). Estimation of vegetation cover in desertified regions from hyperion imageries using NDVI and partial least squares regression. Journal of Desert Research, 31, 162-167. (in Chinese with English abstract)
	[ 李晓松, 李增元, 高志海, 白黎娜, 王琫瑜 (2011). 基于NDVI与偏最小二乘回归的荒漠化地区植被覆盖度高光谱遥感优测. 中国沙漠, 31, 162-167.]
[11]	Thenkabail PS, Smith RB, de Pauw E (2000). Hyperspectral vegetation indices and their relationships with agricultural crop characteristics. Remote Sensing Environment, 71, 158-182.
[12]	Townshend JRG, Justice CO (1986). Analysis of dynamics of African vegetation using the normalized difference vegetation index. International Journal of Remote Sensing, 12, 1224-1242.
[13]	Yang JH, Ci LJ (2006). Progress on remote sensing-based desertification assessment. World Forestry Research, 19(6), 11-17. (in Chinese with English abstract)
	[ 杨晓晖, 慈龙骏 (2006). 基于遥感技术的荒漠化评价研究进展. 世界林业研究, 19(6), 11-17.]

[1]	苏晨飞, 田慰, 张楠, 唐龙, 赵宇玮, 王耀. 光合响应模型与理论指导的神经网络融合的多因素光合速率预测模型[J]. 植物生态学报, 2026, 50(1): 173-187.
[2]	王秀媛, 申磊, 刘婷婷, 尉雯雯, 张帅, 张伟. ‘塞外红’苹果-大豆复合系统根系时空分布与种间竞争策略[J]. 植物生态学报, 2025, 49(5): 748-759.
[3]	陈文义, 王智勇, 周梦岩, 麻文俊, 王军辉, 罗志斌, 周婧. 幼龄楸树生物量分配规律与异速生长模型[J]. 植物生态学报, 2025, 49(2): 356-366.
[4]	赵梦扬, 庄淏然, 许德浩, 马国荣, 马永成, 冯克鹏. 干旱半干旱地区灌区玉米农田土壤植物大气连续体系统氢氧稳定同位素特征及其影响因素[J]. 植物生态学报, 2025, 49(2): 256-267.
[5]	韩雨晴, 熊伟, 吴波, 卢琦, 杨文斌, 刘雅莉, 张景波, 辛智鸣, 马迎宾, 廉泓林, 王思涵. 乌兰布和沙漠梭梭茎干液流对降雨脉冲的响应[J]. 植物生态学报, 2024, 48(9): 1172-1179.
[6]	李蓓蓓, 张明军, 车存伟, 刘泽琛, 钟晓菲, 张园园, 张宇. 基于稳定同位素示踪的不同覆砂厚度下枣树水分利用策略[J]. 植物生态学报, 2024, 48(9): 1202-1212.
[7]	付照琦, 胡旭, 田沁瑞, 葛艳灵, 周红娟, 吴小云, 陈立欣. 晋西黄土区2种典型森林树种夜间液流特征及对环境因子的响应[J]. 植物生态学报, 2024, 48(9): 1128-1142.
[8]	江康威, 张青青, 王亚菲, 李宏, 丁雨, 杨永强, 吐尔逊娜依•热依木. 放牧干扰下天山北坡中段植物功能群特征及其与土壤环境因子的关系[J]. 植物生态学报, 2024, 48(6): 701-718.
[9]	张计深, 史新杰, 刘宇诺, 吴阳, 彭守璋. 气候变化下中国潜在自然植被生态系统碳储量动态[J]. 植物生态学报, 2024, 48(4): 428-444.
[10]	盘远方, 潘良浩, 邱思婷, 邱广龙, 苏治南, 史小芳, 范航清. 中国沿海红树林树高变异与环境适应机制[J]. 植物生态学报, 2024, 48(4): 483-495.
[11]	吴茹茹, 刘美珍, 谷仙, 常馨月, 郭立月, 蒋高明, 祁如意. 气候变化对巨柏适宜生境分布的潜在影响和预测[J]. 植物生态学报, 2024, 48(4): 445-458.
[12]	王复标, 叶子飘. 植物电子传递速率光响应模型的研究进展[J]. 植物生态学报, 2024, 48(3): 287-305.
[13]	兰光飞, 张强, 陈相标, 陈仕东, 熊德成, 刘小飞, 杨智杰, 杨玉盛. 中亚热带格氏栲林凋落物季节动态特征及其影响因素[J]. 植物生态学报, 2024, 48(12): 1589-1601.
[14]	杨子龙, 罗光宇, 雷富民, 余静柔, 蔡圣轩, 于晶. 浙江象山沿海岛屿苔藓植物种-面积关系及物种丰富度影响因素[J]. 植物生态学报, 2024, 48(11): 1486-1500.
[15]	张智洋, 赵颖慧, 甄贞. 基于LUCC的1986-2022年松花江流域陆地生态系统碳储量动态监测[J]. 植物生态学报, 2024, 48(10): 1274-1290.