Estimation of Winter Wheat Canopy Chlorophyll Content Based on Canopy Spectral Transformation and Machine Learning Method

Canopy chlorophyll content (CCC) is closely related to crop nitrogen status, growth and productivity, detection of diseases pests, final yield. Thus, accurate monitoring in crops great significance for decision support precision agriculture. In this study, winter wheat the Guanzhong Plain area Shaanxi Province, China, was selected as research subject explore feasibility canopy spectral transformation (CST) combined with a machine learning method estimate CCC. A hyperspectral ground dataset situ measured construct CCC prediction models over three seasons from 2014 2017. Sensitive-band reflectance (SR) narrow-band index (NSI) were established based on original spectrum (OS) CSTs, including first derivative (FDS) continuum removal (CRS). Winter estimation constructed using univariate regression, partial least squares (PLS) random forest (RF) regression SR NSI. The results demonstrated reliability CST First, compared OS-SR (683 nm), FDS-SR (630 nm) CRS-SR (699 had larger correlation coefficient between CCC; secondly, among parametric methods, CRS-NDSI independent variable achieved satisfactory estimating wheat; thirdly, method, RF multiple variables best accuracy (the determination validation set (Rv2) 0.88, RMSE (RMSEv) 3.35 relative deviation (RPD) 2.88). modeling could be used basic predict area.