Study for estimation of rice grain protein contents using hyperspectral data

Abstract

Ministry of Economy, Trade and Industry (METI) of Japan is developing the hyperspectral sensor that is planned to be installed in the next generation earth observation satellite. Hyperspectral sensor is able to collect information as a set of images in high spectral resolution. It has begun to spread into fields of agriculture and environmental surveillance in Japan.
The aim of this study is to study the estimation formula of rice grain protein contents using hyperspectral data. Especially, it aimed to standardize the estimation formula for the same observation area and the same rice species.
We investigated the standardized estimation formula of rice grain protein contents as paddy growing information through two approaches. One is to use the vegetation indices as biophysical indicator in previous research. Another one is to use Normalized Difference Spectral Index (NDSI). For the estimation rice grain protein contents in Osaki city, Miyagi prefecture, the standardized estimation formula in year of 2007, 2008, and 2009 was obtained. Obtained formula uses ratio of R460 and R510 as independent variable and coefficient of determination R² is 0.72.