Vis/NIR reflectance spectroscopy for hybrid rice variety identification and chlorophyll content evaluation for different nitrogen fertilizer levels.

Abstract

Nitrogen is one of the most important nutrient indicators for the growth of crops, and is closely related to the chlorophyll content of leaves and thus influences the photosynthetic ability of the crops. In this study, five hybrid rice varieties were cultivated during one entire growing period in one experimental field supplied with six nitrogen fertilizer levels. Visible and near infrared (vis/NIR) reflectance spectroscopy combined with multivariate analysis was used to identify hybrid rice varieties and nitrogen fertilizer levels, as well as to detect chlorophyll content associated with nitrogen levels. The support vector machine (SVM) algorithm was applied to identify five varieties of hybrid rice and six levels of nitrogen fertilizer. The results demonstrated that different varieties of hybrid rice for each nitrogen level can be well distinguished except for the highest nitrogen level, and no nitrogen level for each rice variety can be completely identified from the other five nitrogen levels. Further, 12 spectral indices combined with partial least square (PLS) analysis were applied for estimating chlorophyll content of rice leaves from plants subjected to different nitrogen levels, and a root mean square error of cross-validation (RMSECV) of 0.506, a coefficient of determination ( R 2 ) of 97.8% and a ratio of performance to deviation (RPD) of 4.6 for all rice varieties indicated this as a preferable procedure. This study demonstrates that Vis/NIR spectroscopy can have a great potential for identification of rice varieties and evaluation of nitrogen fertilizer levels.

Plant phenotyping relevance

Vis/NIR分光とPLS解析により、イネ葉のクロロフィル含量を定量推定し、交差検証指標で性能評価しており、表現型取得・推定法が中心である。

Code and data availability

The authors deposit the paper's vis/NIR reflectance spectral data (used for rice variety identification and chlorophyll/SPAD analysis) in the Dryad Digital Repository, with an explicit public URL.