Spectral characteristics analysis and nitrogen content estimation in sugarcane leaves

A field experiment included 5 levels’ fertilization was set up to measure the nitrogen content and spectral reflectance of sugarcane leaves at tillering stage and their correlations were analyzed，and the nitrogen content estimation model were built based on the sensitive reflection band. The results showed that nitrogen content of sugarcane leaf and spectral reflectance has significant negative correlation in 401-1 000 nm，and the highly significant negative correlation in 402-953 nm with two peaks in 550 nm and 741 nm，which showed that the spectral reflectance around these two bands are sensitive to nitrogen content. The determination coefficient（ R2） of the model constructed by R550 and R741 is larger than the NDVI（550，741），RVI（550，741），NDVI（730，835），and RVI（730，835） .R550、R741 are more suitable for sugarcane leaf nitrogen content estimation compared to NDVI （550，741），RVI（550，741），NDVI（730，835），RVI（730，835）. The value of the decision coefficient constructed quadratic function model by R741（ y = -119.1x2 + 61.53x + 3.851） is 0.681，root mean square error （RMSE） of the estimated nitrogen content is 1.13 g/kg-1，average of relative error is 8.92%，and the comprehensive effect is better compared to other models. The quadratic function model constructed by R741 is recommended as the leaf nitrogen content estimation model in sugarcane tillering stage.