Journal of Taiwan Agricultural Research

Author:Rong-Kuen Chen and Chwen-Ming Yang*

Abstract:

    Ground-based remotely sensed high-resolution reflectance spectra and growth traits of rice plants grown in the 1st and the 2nd cropping seasons of 2000 were used for establishing spectral characteristic models to estimate rice growth. Differences in plant growth from the estimates of spectral indices with narrow bands and simulated SPOT broad bands were compared. The 3-dimensioanl reflectance spectra of rice canopy were similar to those of other green plants. The patterns of growth traits were curvilinear, reached the climax near heading and differed between cropping seasons. By correlation intensity analysis between reflectance of narrow bands and growth traits, it showed that the correlation intensity curves were similar before and after heading, with lesser degree of correlation after heading. By the multiple linear regression (MLR) analysis, the best-2 and best-3 MLR models were established. Among the examined spectral indices, the GREEN/NIR ratio had the best correlation with growth traits before heading yet it was the RED/NIR ratio after heading. Differences in growth traits from spectral index estimations by narrow bands and the simulated SPOT broad bands were not significant.

Key words:Ground-based hyperspectral reflectance data, Spectral characteristic model, Vegetation index, Growth, SPOT satellite

Download: PDF Links