Monitoring Nitrogen Status And Growth Characters With Spectral Reflectance In Cotton

Remote sensing is an important tool for estimating biochemical components such as nitrogen and chlorophyll concentrations and agronomic parameters such as leaf area index and biomass in crop plants. In this study, field experiments with different cotton varieties and nitrogen fertilization rates were carried out in two years, canopy hyperspectral reflectance was measured with a portable radiometer(ASD) over the whole growth periods. The objectives of this research were to determine the relationships of reflectance characteristics of cotton to leaf nitrogen status and growth characters, to develop quantitative models for monitoring nitrogen status and growth traits based on canopy hyperspectral reflectance. This work would provide theoretical basis and key techniques for non-destructive monitoring of cotton growth in precision crop management.Based on the change patterns of leaf area index and dry matter under different nitrogen rates and growth stages, the relationships of leaf area index and dry matter to canopy hyperspectral reflectance and hyperspectral index were investigated, and sensitive spectral parameters and quantitative equations for predicting LAI and dry matter were established. The results showed that LAI and leaf dry matter increased at ealier stages, and then decreased over growth progress; LAI and dry matter increased with increasing nitrogen fertilization rates. Correlation between LAI and canopy reflectance at single waveband was significant, negative at visible and positive at near infrared. Vegetation indices were significantly correlated with LAI and dry matter. The hyperspectral index RVI[R(760～850), R550] could be used for reliably estimating the LAI and RVI[R(760～850), R550] and NDVI(750, 550) for leaf dry matter over the whole growth period. This indicates that it is feasible to predict LAI and dry matter in cotton by using remote sensing technique.The quantitative relationships of leaf nitrogen status to canopy hyperspectral reflectance and SPAD value in different cotton cultivars under varied N rates were investigated, and the sensitive parameters and monitoring equations for leaf nitrogen status determined. The results showed that the leaf nitrogen status, canopy hyperspectral reflectance and SPAD value significantly changed with levels of nitrogen fertilization. The sensitive bands of leaf nitrogen status were 600～700 nm of red valley and 750～900 nm of near infrared wavebands. The relationships of leaf nitrogen concentration to RVI [R(760～850), 700] of canopy reflectance spectra were all highly significant with average R~2 of 0.70 in four cultivars, similar correlation was seen with SPAD value over growth stages. The relationships of leaf nitrogen accumulation to NVD672 of canopy reflectance spectra were all highly significant with average R~2 of 0.61 in four cultivars. An integrated regression equation could be used for describing the dynamic change pattern of leaf nitrogen accumulation with hyperspectral parameters in different varieties, growing stages and nitrogen levels in cotton.The change patterns of C/N ratio in function leaves with nitrogen rates and quantitative relationships to canopy hyperspectral reflectance in cotton were further established. The results showed that function leaf C/N ratio decreased with increasing nitrogen rates. The function leaf C/N ratio was found to be positively correlated to hyperspectral reflectance in 550～720 nm, with highest correlation coefficients at 630nm and 710nm. But reflectance of 760～1300 nm was negatively correlated to function leaf C/N ratio. Ratio and normalized difference vegetation indices composed of the sensitive wavebands were significantly correlated with C/N ratio in function leaves. The C/N ratio was significantly correlated to the SPAD values in function leaves.To determine the best hyperspectral vegetation index for estimating pigment concentrations in cotton, the dynamic trends of pigment concentrations in relation to the characteristics of hyperspectral reflectance in function leaves of four cotton cultivars wereanalyzed. The results showed that the sensitive wavebands of function leaf pigmentconcentrations were 550～700 nm and near infrared wavebands. Vegetation indices composed of the sensitive wavebands were significantly correlated with pigment concentrations in function leaves, with the best vegetation index as Lo (Red edge minimum). The SPAD values of function leaves initially increased and then decreased; while increased with increasing nitrogen fertilization rates. The SPAD values were significantly correlated with chlorophyll concentrations in function leaves.