Journal of Taiwan Agricultural Research

Author:Fang-Ming Chang, Chwen-Ming Yang*, Fu-Hsing Hsu and Kuo-Yuan Hong

Abstract:

    Ground-based remotely sensed canopy spectral data of hyperspectral resolution were acquired at the experimental field of Taiwan Livestock Research Institute, Hsinhua, to study the spectral characteristics of pangolagrass (Digitaria decumbens Stent) in associated with chemical components (crude protein, acid detergent fiber and neutral detergent fiber). From the results of correlation intensity analyses, there were varied correlation coefficients between spectral reflectance and content of chemical components along the measured wavelength domain (350-2500 nm). In general, reflectance of wavelengths greater than 740 nm in the infrared region had higher correlation with chemical components. Eight wavebands in the infrared were selected for each component to run the multiple linear regression analyses. It indicated that contents of crude protein (CP) may be expressed as a function of reflectance at 1893, 1332, 1731 and 1837 nm with coefficient of determination (R²) of 0.293 (P<0.0001). The contents of acid detergent fiber (ADF) and neutral detergent fiber (NDF) may be expressed as functions of reflectance at 1038 and 1663 nm and at 1450 and 2385 nm, with R² of 0.484 (P<0.0001) and 0.173 (P<0.0001), respectively. It is suggested that the changes of chemical components of pangolagrass may be estimated and monitored by these regression models.

Key words:Canopy reflectance spectrum, Spectral characteristics, Pangolagrass, Multiple linear regression analysis, Chemical component

Download: PDF Links