山茶
天蓬
多酚
偏最小二乘回归
化学
山茶科
食品科学
植物
园艺
数学
生物
生物化学
抗氧化剂
统计
作者
Meng Bian,Andrew K. Skidmore,Martin Schlerf,Tiejun Wang,Yanfang Liu,Rong Zeng,Teng Fei
出处
期刊:Isprs Journal of Photogrammetry and Remote Sensing
日期:2013-03-06
卷期号:78: 148-156
被引量:61
标识
DOI:10.1016/j.isprsjprs.2013.02.002
摘要
Abstract Some biochemical compounds are closely related with the quality of tea ( Camellia sinensis (L.)). In this study, the concentration of these compounds including total tea polyphenols, free amino acids and soluble sugars were estimated using reflectance spectroscopy at three different levels: powder, leaf and canopy, with partial least squares regression. The focus of this study is to systematically compare the accuracy of tea quality estimations based on spectroscopy at three different levels. At the powder level, the average r 2 between predictions and observations was 0.89 for polyphenols, 0.81 for amino acids and 0.78 for sugars, with relative root mean square errors (RMSE/mean) of 5.47%, 5.50% and 2.75%, respectively; at the leaf level, the average r 2 decreased to 0.46–0.81 and the relative RMSE increased to 4.46–7.09%. Compared to the results yielded at the leaf level, the results from canopy spectra were slightly more accurate, yielding average r 2 values of 0.83, 0.77 and 0.56 and relative RMSE of 6.79%, 5.73% and 4.03% for polyphenols, amino acids and sugars, respectively. We further identified wavelength channels that influenced the prediction model. For powder and leaves, some bands identified can be linked to the absorption features of chemicals of interest (1648 nm for phenolic, 1510 nm for amino acids, 2080 nm and 2270 nm for sugars), while more indirectly related wavelengths were found to be important at the canopy level for predictions of chemical compounds. Overall, the prediction accuracies achieved at canopy level in this study are encouraging for future study on tea quality estimated at the landscape scale using airborne and space-borne sensors.
科研通智能强力驱动
Strongly Powered by AbleSci AI