Journal of Geo-information Science >
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
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
LIU Yuqin , MENG Qingyan , SHA Jinming , SHI Feng , LIU Miao , WANG Chunmei . Comparison of Different Methods for Retrieving Acacia Rachii Leaf Area Index Based on HJ-1A CCD1 Imagery[J]. Journal of Geo-information Science, 2014 , 16(4) : 645 -652 . DOI: 10.3724/SP.J.1047.2014.00645
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] |
|
/
〈 | 〉 |