Journal of Taiwan Agricultural Research

Author:Chwen-Ming Yang*, Jeng-Tzung Wu, Bor-Kwei Shen, Jih-Zu Yu, Chaur-Tsuen Lo and Yuan Shen

Abstract:

    Field experiments were conducted at the experimental farm of Taiwan Agricultural Research Institute (Wufeng, Taiwan) to measure canopy reflectance spectra and plant characters of Amaranthus mangostanus L. (cultivar TNG 1) grown in the growing seasons of 2001-2002. From the simple linear correlation analysis between plant characters and canopy spectral reflectances in the range of 350-2400 nm, it showed that various correlation coefficients existed along the measured spectral wavebands. The narrow wavebands with the maximum absolute value of coefficients were located at 767 nm (r=-0.741**), 574 nm (r=-0.492*), 779 nm (r=-0.645**), 574 nm (r=-0.628**), 1975 nm (r=-0.697**) and 577 nm (r=-0.506*) for aboveground fresh yield, leaf total chlorophyll, percent dry matter, leaf area, nitrate content and nitrogen content, respectively. Regression models validation by using data from another experiment showed a significant correlation between the estimated and the measured values in leaf total chlorophyll and aboveground nitrogen. By analyzing of the relationships between plant characters and canopy spectral indices, it indicated that aboveground fresh yield was linearly correlated with GREEN/NIR ratio, RED/NIR ratio and NDVI, aboveground nitrogen was curvilinearly correlated with GREEN, RED/NIR ratio, GREEN/RED ratio and NDVI, and leaf total chlorophyll was also curvilinearly correlated with GREEN and GREEN/RED ratio. These regression models were validated effective with another set of experimental data. By the multiple linear regression analyses, fresh yield and aboveground nitrogen were fitted to 2-variable regression models with the determining factors of 0.852 and 0.761, respectively. Leaf area can be estimated with a best 3-variable regression model (R²=0.545), while leaf total chlorophyll, percent dry matter and nitrogen content can be assessed with the best 4-variable regression models, with R² of 0.697, 0.693 and 0.822, respectively. However, regression model for leaf area was found not feasible to the data from another experiment. Results suggest that plant growth and nitrogen status of vegetable amaranth may be estimated reasonably by the regression models from canopy spectral characteristics.

Key words:Canopy spectral characteristics, Amaranthus mangostanus, Plant growth, Nitrogen status, Mutiple linear regression analysis

Download: PDF Links