Hyperspectral vegetation indexes to monitor wheat plant height under different sowing conditions

Abstract

Accurate estimation of plant height is essential for precision crop production. Remote sensing provides a beneficial tool for monitoring plant height, but few attempts had been made to monitor wheat plant height using ground-based hyperspectral vegetation indexes. This study systematically analyzed the quantitative relationship between various hyperspectral vegetation indexes and wheat plant height based on a field experiment under different sowing conditions (i.e., normal sowing and late sowing). The optimal vegetation indexes that were highly sensitive to wheat plant height were derived by comparing the newly constructed hyperspectral vegetation indexes with published vegetation indexes. The experiment was conducted at the Agrometeorological Station, Nanjing University of Information Science and Technology, Jiangsu Province, China. Wheat canopy reflectance spectra were synchronously determined with plant height in tillering-anthanis stage. Results showed that the newly developed vegetation index (R785 – R810)/(R785 + R810 – 2×R802) in normal sowing treatment and R1029 – R1004 in late sowing treatment, generated the optimal testing performance for estimating wheat plant height with the coefficient of determination (R2) of 0.78 and 0.89, root mean square error (RMSE) of 7.37 and 5.92 cm, and residual prediction deviation (RPD) of 2.04 and 3.09, respectively. This study suggests that the ground-based hyperspectral remote sensing technique can provide spectral indexes for accurately monitoring wheat plant height.