基于TAVI的山区毛竹林LAI反演分析
作者简介:江 洪(1975-),男,福建永安人,博士,副研究员。研究方向为遥感技术与应用及电子政务等。E-mail:jh910@fzu.edu.cn
收稿日期: 2014-02-20
要求修回日期: 2014-04-13
网络出版日期: 2015-04-10
基金资助
国家科技支撑计划项目“南方红壤水土流失综合监测”(2013BAC08B01)
福建省自然科学基金项目“基于地形调节植被指数的毛竹林叶面积指数遥感反演研究”(2011J01267)
Bamboo Forest LAI Retrieval and Analysis in Mountainous Area Based on TAVI
Received date: 2014-02-20
Request revised date: 2014-04-13
Online published: 2015-04-10
Copyright
本文采用地形调节植被指数(TAVI),以RapidEye高分辨率多光谱遥感影像为数据源,对福建省永安市毛竹林山区进行了叶面积指数(LAI)地面实测、遥感建模及反演分析。通过TAVI与归一化植被指数(NDVI)、比值植被指数(RVI)的对比研究,结果表明:(1)毛竹林实测LAI与TAVI、NDVI和RVI线性回归的决定系数(R2)分别为0.6085、0.3156和0.4092,最佳非线性回归的R2分别提高到0.6624、0.5280和0.6497。LAI与NDVI或RVI非线性(U曲线)模型可以很好地解释LAI-VI的散点分布规律,但难以解决LAI-VI间因地形影响导致的“同物异谱”和“异物同谱”问题,因此,在山区大面积推广应用需慎重。(2)通过实测LAI的验证表明,LAI-TAVI回归模型可有效避免因地形影响导致的“同物异谱”和“异物同谱”问题。TAVI具有良好的削减地形影响作用,可用于山区植被LAI的遥感反演。
江洪 , 张兆明 , 汪小钦 , 何国金 . 基于TAVI的山区毛竹林LAI反演分析[J]. 地球信息科学学报, 2015 , 17(4) : 500 -504 . DOI: 10.3724/SP.J.1047.2015.00500
As one of the key biophysical parameters in the bamboo forest evaluation, the leaf area index (LAI) retrieval from remote sensing data has always been challenged by the topographic effect in mountainous area. In this paper, the topographic-adjusted vegetation index (TAVI) was proposed to eliminate the topographic influence for the bamboo forest LAI derivation in Yongan city, Fujian, based on the Rapideye high spatial resolution satellite imagery. Normalized difference vegetation index (NDVI) and ratio vegetation index (RVI) were also utilized in the statistical analysis with respect to LAI for comparison with TAVI. The regression results indicate: (1) LAI is more linearly correlated with TAVI than NDVI or RVI. R2 (coefficient of determination) of the linear regression between LAI and TAVI, NDVI, RVI are 0.6085, 0.3156, and 0.4092 respectively. For the optimal non-linear fitting model, the corresponding R2 had increased to 0.6624, 0.5280 and 0.6497 respectively. Although the quadratic polynomial regression model can well explain the relationship between LAI and NDVI or RVI, it can hardly illustrate the typical phenomenon of "same object with different spectra" and "different objects with same spectrum" that resulted from topographic effect. (2) Both the LAI-TAVI regression models and the in-situ measurement demonstrate that the proposed method can effectively avoid the above problems with a correlation coefficient (r) of 0.7674 between in-situ and the simulated LAI, and a RMSE of 0.3403. In conclusion, TAVI shows good capability to alleviate the topographic effect and can be effectively applied to the LAI retrieval of the bamboo forest in mountainous area.
Fig. 1 Multi-spectral image, NDVI, RVI and TAVI of sample area图1 样区影像、NDVI、RVI、TAVI |
Tab. 1 Vegetation index表1 植被指数 |
植被指数 | 公式 | 备注 |
---|---|---|
归一化差值植被指数(NDVI) | Rouse等(1974)[19] | |
比值植被指数(RVI) | Jordan等(1969)[20] | |
地形调节植被指数(TAVI) | 江洪等(2010, 2011)[17-18] | |
注:Bnir表示近红外波段数据;Br表示红光波段数据;Mr表示研究区红光波段数据的最大值;f(Δ)表示地形调节因子;SVI表示阴影植被指数 |
Tab. 2 Regression analyses between NDVI、RVI、TAVI and cosi表2 NDVI、RVI、TAVI与cosi回归分析结果 |
植被指数 | 斜率(k) | 截距(d) | 相关系数(r) |
---|---|---|---|
NDVI | 0.6060 | 0.3323 | 0.5774 |
RVI | 0.6677 | 0.1584 | 0.6196 |
TAVI | -0.0442 | 0.6617 | 0.0044 |
Fig. 2 Regression analyses between in-situ LAI and VI (green stands for linear regression, red stands for non-linear regression)图2 实测LAI-VI回归分析(绿色为线性回归,红色为非线性回归) |
Tab. 3 Regression models between in-situ LAI and VI表3 实测LAI-VI拟合回归模型 |
回归类型 | 植被指数 | 模型 | R2 | F(F0.05=3.9) |
---|---|---|---|---|
线性 | NDVI | y=3.5554x+0.5478 | 0.3156 | 5.53 |
RVI | y=2.6755x+2.1746 | 0.4092 | 8.31 | |
TAVI | y=20.252x-13.058 | 0.6085 | 18.65 | |
非线性 | NDVI | y=39.598x2-58.451x+24.088 | 0.5280 | 13.43 |
RVI | y=20.591x2-15.73x+5.6826 | 0.6497 | 22.26 | |
TAVI | y=9.1868x4.8818 | 0.6624 | 18.75 |
注:y指实测LAI,x指相应的植被指数 |
Fig. 3 Retrieved LAI images of sample area图3 研究区LAI反演结果 |
Fig. 4 Correlation analysis of in-situ LAI and the simulated LAI图4 实测LAI-模拟LAI相关分析 |
The authors have declared that no competing interests exist.
[1] |
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
/
〈 | 〉 |