Study of the optimal wavelength band using the radiative transfer simulation for rice protein contents estimation

Abstract

Hyper-spectral sensor that has high spectral resolution is expected to the application of agriculture field. In the rice cultivation in Japan, attention has focused on making high quality rice.
In the previous others, several vegetation indices which use various wavelength bands were effective for rice protein contents estimation. For example, Asaka et al. (2003) proposed the vegetation index computed from SPOT/HRV band 2 (610-680nm) and band 3 (790-890nm) data as the predictor variable of the estimate equation, and built the estimate equation by regression analysis. And Inoue et al. (2008) proposed the other vegetation index which uses two hyper-spectrum bands (570nm, 970nm). In the same manner, Suhama et al. (2010) proposed the vegetation index which uses two hyper-spectrum bands (460nm, 510nm). In order to evaluate the validity of these vegetation indices, it is necessary that the theoretical examination which uses radiative transfer simulation in addition to regression analysis.
Then, this research shows theoretically the validity of spectral reflectance at these wavelength bands using the PROSPECT model as radiative transfer simulation. The results obtained by this research are shown below.
1)610-680nm and 570nm are the sensitive region of chlorophyll-a, -b contents of rice leaf.
2)790-890nm is the sensitive region of equivalent water thickness of rice leaf.
3)970nm is the sensitive region of equivalent water thickness and dry matter of rice leaf.
4)460nm and 510nm are sensitive region of chlorophyll-a, -b contents and carotenoid contents of rice leaf.