Estimation the leaf phosphorus concentration of litchi (Litchi chinensis Sonn.) at different growth stages by canopy reflectance

This study aims to quantify litchi leaf phosphorus concentration (LPC) from canopy spectra at five different growth stages. The band iteration was used to optimize the ratio vegetation index (RVI) and ratio of band difference index (RBDI). The correlation analysis, linear regression and cross validation were used to analyze the relationship between spectral data and LPC and construct the robust estimation model for LPC. It is difficult to estimate the LPC by the general model. The correlation analysis based on the dataset at each growth stage showed that the band optimization can improve the relationship between spectral data and LPC. And the wavebands selected by the band iteration were located in blue, red, near infrared and shortwave infrared region, which were partly related to the absorption of anthocyanin, protein, starch et al. The estimation model by RBDI (905, 965, 1005) in fruit maturation stage had the best performance (R2cv>0.95). The best estimation model in flowering stage was calibrated by RVI (925, 1025) with the R2cv of 0.66. The linear model by RBDI (1525, 2105, 2225) presented the R2cv of 0.75 in autumn shoot maturation stage. And the RBDI (815, 1475, 2135) and RBDI (2005, 2095, 2375) had the best performances in flower spike growing stage and flower bud differentiation stage with the R2cvs of 0.68 and 0.63, respectively. The results indicated that it is available to estimate LPC by canopy reflectance spectra. Besides it is necessary to estimate the phosphorus content of litchi based on the growth stages.