摘要
以ASDFieldSpec光谱仪实测了不同生长季大豆的冠层高光谱,同期采集了对应大豆LAI、地上鲜生物量。逐波段分析了冠层光谱反射率、导数光谱与大豆LAI、地上鲜生物量的相关关系;采用单变量线性回归逐波段分析了冠层光谱反射率、导数光谱与大豆LAI、地上鲜生物量确定性系数随波长的变化趋势;并建立了以近红外与可见光波段的冠层光谱反射率的比值植被指数RVI与大豆LAI、地上鲜生物量的高光谱遥感估算模型。结果表明,冠层光谱反射率在350~680nm、760~1050nm波谱区与大豆LAI、地上鲜生物量相关性较大,而在红边区680~760nm的波段相关性较大;导数光谱则在红边区与大豆LAI、地上鲜生物量相关程度高。而通RVI方式建立的遥感估算模型能较为准确估算大豆LAI、地上鲜生物量。
Canopy reflectances of soybean were measured with ASD FieldSpec during different growth stages; and simultaneously, LAI and biomass were acquired. The relationships between canopy reflectance, derivative and LAI, biomass were analyzed with every single band. Determination coefficient (R2) was obtained with linear regression of soybean reflectance and derivative against LAI and biomass. Different regression models were applied to establish the relationship between RVI and soybean LAI and biomass. It is found that soybean canopy reflectance has an intimate relationship with soybean LAI, biomass in visible spectral region 350 - 680 nm and near infrared spectral region 760 - 1050 nm while the relationship between reflectance and soybean LAI, biomass has a loose relation in red edge spectral region at different growth stages; for derivative, the relationship shows an inverse trend, it has an close relationship between derivative and soybean LAI, biomass in red edge spectral region, but in visible spectral region 350 - 680 nm and near infrared spectral region 760 - 1050 nm, this is not the case for that with soybean canopy reflectance. The study shows that RVI model constructed by hyperspectral remote sensing reflectance from near infrared spectral region and visible spectral region regressing against soybean LAI, biomass with power functions and exponential functions can greatly improve remote sensing ability for estimating soybean LAI, biomass.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2005年第1期36-40,共5页
Transactions of the Chinese Society of Agricultural Engineering
基金
中科院知识创新工程重大项目(KZCX1-SW-19)
关键词
高光谱
反射率
冠层
叶面积指数
地上鲜生物量
回归分析
hyperspectrum
reflectance
canopy
leaf area index
aboveground biomass
regression analysis