Winter Wheat Canopy Water Content Monitoring Based on Spectral Transforms and “Three-edge” Parameters

•At different growth stages of winter wheat, the most closely related spectral parameters such as trilateral parameters and spectral transformation were obtained.•The combinational models of canopy water content models specific to individual growth stages are built on spectral transforms or “three-edge” parameters.•Principal component regression spectral monitoring models are developed based on the combination of growth stages. Suitable spectral monitoring models of canopy water content provide a scientific basis for real-time dynamic, accurate, non-destructive diagnosis over large acreage. This work investigates winter wheat under different water treatments to examine the relationship between canopy water content and spectral reflectance. Principal component regression spectral monitoring models are developed based on the combination of growth stages. The growth stage constraints are divided, and the influence of other background noises is removed to achieve accurate and stable spectral monitoring results of canopy water content at all growth stages. The following main conclusions are derived. (1) At the stem elongation–booting, booting–milking, and milking–ripening stages and during the entire growth period, the spectral transforms with the highest correlation with winter wheat canopy water content are the first-order derivative, division by R930, division by R450-750, and division by R930, respectively; the corresponding sensitivity bands are 758, 759, 690, and 759 nm, respectively. At the stem elongation–booting, booting–milking, and milking–ripening stages and during the entire growth period, the “three-edge” parameters with the highest correlation with winter wheat canopy water content are Rg/Rr, SDr/Sdy, (Rg − Rr)/(Rg + Rr), and (SDr-SDb), respectively. (2) In accordance with the rationale that the spectral parameters should have the highest correlation coefficients with canopy water content at each growth stage, combinational models of canopy water content that are specific to individual growth stages are developed based on spectral transforms or “three-edge” parameters. Compared with the optimal single-parameter regression model, the combinational models significantly improve the estimation accuracy of canopy water content at each growth stage. (3) Monitoring models based on principal component analysis are constructed with comprehensive spectral information. These models can improve the monitoring accuracy at other growth stages, especially at th