基于HJ-1A CCD1数据的台湾相思树叶面积指数反演
作者简介:刘玉琴(1989-),女,福建泉州人,硕士生,主要从事遥感与地理信息建模研究。E-mail:liuyq0202@sina.cn
收稿日期: 2013-07-25
要求修回日期: 2013-09-14
网络出版日期: 2014-07-10
基金资助
国家国际科技合作专项资助(2010DFA21880)
广东省省院产学研合作资金资助(2012B091100219)
欧盟第七框架项目(FP7-PEOPLE-2009-IRSES-IGIT)
高分辨率对地观测系统重大专项(09-Y030B03-9001-13/15)
Comparison of Different Methods for Retrieving Acacia Rachii Leaf Area Index Based on HJ-1A CCD1 Imagery
Received date: 2013-07-25
Request revised date: 2013-09-14
Online published: 2014-07-10
Copyright
基于HJ-1A CCD1环境卫星数据,以福建沿海地区普遍分布的台湾相思树为研究对象,利用回归分析法(NDVI、OSAVI、EVI、HJVI)和PROSAIL辐射传输模型,构建台湾相思树LAI反演模型。同时,利用同步野外地面实测数据,将模型估算LAI值与实测LAI值进行对比。结果表明:(1)相比归一化植被指数NDVI、优化土壤调节指数OSAVI和增强型植被指数EVI 3种常用植被指数,引入修正大气、土壤背景影响的蓝、绿波段的环境植被指数HJVI来反演相思树LAI具有更高的精度(R2=0.7344,RMSE=0.1421);(2)本研究所选4种植被指数构建的最优反演模型均为非线性模型,其中,环境植被指数HJVI反演LAI最优模型为幂函数模型,表明相思树LAI与植被指数之间呈非线性变化;(3)PROSAIL辐射传输模型法比回归分析法反演相思树LAI的精度有较大提高(R2=0.7903,RMSE=0.1303),可见PROSAIL模型法构建反演模型能更好地反演相思树LAI。
刘玉琴 , 孟庆岩 , 沙晋明 , 石锋 , 刘苗 , 王春梅 . 基于HJ-1A CCD1数据的台湾相思树叶面积指数反演[J]. 地球信息科学学报, 2014 , 16(4) : 645 -652 . DOI: 10.3724/SP.J.1047.2014.00645
With Acacia Rachii (Acacia confusa) grown in coastal region in Fuzhou as research object, and based on HJ-1A CCD1 imagery which was acquired from China Center for Resource Satellite Data and Applications, the Acacia Rachii LAI was monitored in field using LAI-2000 canopy analysis system, and two kinds of universal LAI inversion methods through regression analysis method and radiative transfer model PROSAIL model separately were introduced and used in this study. The simulation precisions for different models were analyzed and evaluated through comparing the simulated LAI and measured LAI. Further, we compared the research output with those of previous researchers. The results showed that: (1) Compared with the three vegetation indices (NDVI, EVI and OSAVI), HJVI vegetation index performed best in Acacia Rachii LAI inversion among all of the vegetation indices with R2=0.7344 and RMSE=0.1421, which introduced blue band and green band in order to weaken the effects of atmosphere and soil; (2) The optimal inversion models of the above four vegetation indices all were non-linear models, and the optimal regression model for Acacia Rachii LAI inversion based on vegetation indices was the power regression model of HJVI, indicating that there existed non-linear relationships between Acacia Rachii LAI and vegetation indices; (3) There had obvious improvement in the precision of LAI inversion through PROSAIL model compared with regression analysis method based on vegetation indices with R2=0.7903 and RMSE=0.1303, which indicated that PROSAIL model could better estimate Acacia Rachii LAI than regression analysis method to some extent. Therefore, radiative transfer model such as PROSAIL used to construct the inversion model is feasible. It could reflect ground condition better and possesses higher application value and broad application prospect.
Key words: Acacia Rachii; regression analysis; PROSAIL model; inversion accuracy; LAI
Fig.1 The technical route图1 技术路线 |
Tab.1 Input parameters of PROSAIL model表1 PROSAIL模型的输入参数 |
模型 | 参数 | 变量符号 | 单位 | |
---|---|---|---|---|
PROSPECT | 叶片内部结构参数 | N | -- | |
叶绿素ab含量 | Cab | μg/cm2 | ||
类胡萝卜素含量 | Car | μg/cm2 | ||
叶片等效水厚度 | Cw | cm | ||
叶片干物质含量 | Cm | g/cm2 | ||
棕色荧光成分含量 | Cbrown | -- | ||
SAIL | 叶面积指数 | LAI | -- | |
叶倾角分布函数 | LIDF | -- | ||
热点参数 | SL | -- | ||
土壤光谱反射率 | ρS | -- | ||
散射在总入射辐射中的比例 | SKYL | -- | ||
太阳天顶角 | θS | ° | ||
观测天顶角 | θV | ° | ||
太阳-观测相对方位角 | φSV | ° |
Tab.2 Regression results between vegetation indices and LAI表2 植被指数与叶面积指数间回归模型 |
植被指数 | 关系模型 | R2 | SD |
---|---|---|---|
NDVI | y = 8.252x-2.0143 | 0.3129 | 0.0465 |
y = 0.5983e2.6479x | 0.3169 | 0.0326 | |
y = 5.1686ln(x) + 5.5747 | 0.3125 | 0.0665 | |
y = 6.832x1.6585 | 0.3165 | 0.0403 | |
EVI | y = 9.5691x + 0.3531 | 0.5711 | 0.0182 |
y = 1.2827e3.0607x | 0.5747 | 0.0193 | |
y = 2.9318ln(x) + 6.7621 | 0.5911 | 0.1745 | |
y = 9.9656x0.9379 | 0.5951 | 0.0167 | |
OSAVI | y = 9.2555x-0.7694 | 0.4734 | 0.0281 |
y = 0.8936e2.9662x | 0.4783 | 0.0244 | |
y = 3.9867ln(x) + 6.5795 | 0.4775 | 0.1180 | |
y = 9.4198x1.2778 | 0.4826 | 0.0220 | |
HJVI | y = 9.1546x + 0.347 | 0.6231 | 0.0181 |
y = 1.2779e2.934x | 0.6296 | 0.0192 | |
y = 2.9303ln(x) + 6.6249 | 0.6481 | 0.1741 | |
y = 9.5567x0.9391 | 0.6548 | 0.0166 |
Fig.2 Regression models between vegetation indices and LAI图2 各种植被指数拟合的叶面积指数LAI回归模型 |
Fig.3 Sensitivity analysis of PROSAIL parameters图3 PROSAIL模型各生化参数敏感性 |
Fig.4 Correlation between LAI inversion value and the measurement图4 研究区相思树叶面积指数预测模型检验 |
Tab.3 The achievements of previous LAI inversion studies表3 叶面积指数反演前人研究成果 |
作者 | 遥感数据源 | 最优指数 | 最优模型 | 算法类别 | 地物类型 |
---|---|---|---|---|---|
陈雪洋 | 环境星CCD | RVI | 对数模型 | 回归分析法 | 冬小麦 |
陈鹏飞 | 环境星CCD | EVI | 一元一次模型 | 回归分析法 | 草地 |
尹芳 | HSI高光谱 | MSAVI | 幂函数模型 | 回归分析法 | 草地 |
张竞成 | 环境星CCD、TM | GNDVI、BNDVI | 指数模型 | 回归分析法 | 水稻 |
赵虎 | 环境星CCD | EVI、SAVI | 对数模型、一元一次模型 | 回归分析法 | 冬小麦 |
张瀛 | 环境星CCD | HJVI | 多项式模型 | 回归分析法 | 冬小麦 |
夏天 | 冬小麦冠层光谱 | NDVI | 幂函数模型 | 回归分析法 | 冬小麦 |
李淑敏 | MODIS、ASTER | PROSAIL模型 | PROSAIL模型、回归分析法 | 冬小麦 | |
Roshanak | Hyperspectral Imagery | PROSAIL模型 | PROSAIL模型、回归分析法 | 草地 |
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] |
|
[21] |
|
[22] |
|
[23] |
|
[24] |
|
[25] |
|
[26] |
|
/
〈 | 〉 |