Data Analysis

There is a weak peak at 761 nm for the apparent reflectance ( Figure 4 ) due to the solar-induced ChlF emission and the in-filling effects in the O₂–A absorption band. Therefore, this unique feature can give rise to a significant peak for the first derivative spectra ranging from 755 nm to 763 nm ( Figure 5 ). The ChlF in-filling in the O₂–A absorption band at 760 nm was sensitive to variations of fluorescence (Pérez-Priego et al., 2005), and the Chl a+b content and LAI were highly correlated with F₇₆₀ (Ni et al., 2015; Van der Tol et al., 2016). Accordingly, this study was intended to examine the feasibility of estimating crop CCC by developing two new indices, i.e., REArea₇₆₀ (sum of the first derivative reflectance between 755nm and 763nm) and REA₇₆₀ (maximum value of first derivative reflectance between 755 and 763 nm).

Three canopy reflectance spectra measured at the V6, V12 and R1 growth stages for the Jingyu 7 cultivar from Exp. 1 (A). These apparent reflectance spectra ranging from 755 nm to 763 nm were demonstrated by local enlarged (B) and characterized by a weak peak at 761 nm because of the in-filling effects in the O2–A absorption band.

The curves of two proposed first derivative spectral indices REArea₇₆₀ and REA₇₆₀ for the Jingyu 7 cultivar measured at V6 and V10 growth stages in Exp.1. The high CCC and low CCC were 3.58 and 0.91g/m², respectively. REArea₇₆₀ is defined as the sum of the first derivative reflectance ranging from 755 nm to 763 nm, and REA₇₆₀ is the maximal among the first derivative value from 755 nm to 763 nm.

REArea₇₆₀ is defined as:

REA₇₆₀ is denoted as:

Correlation analyses were performed between the spectral indices related to Chl content and corn CCC using SPSS 17.0 (SPSS, Chicago, IL, USA). In total, 30 existing spectral parameters were calculated ( Table 1 ), and linear inversion models for CCC were established based on Exp.1. The coefficient of determination (R²) was used to evaluate these models. Additionally, to investigate the robustness of the spectral indices, we employed the data from Exp.2 to validate the fitted linear inversion models based on the data from Exp.1. The predictive performance of the spectral indices was assessed by ranking the RMSE values in ascending order. The overall performance of the spectral indices was then evaluated by finding the sum of the RMSE ranks and the R² ranks of fitted linear inversion models. Finally the spectral indices were ordered according to their summed ranks, such that the best performing spectral indices had the lowest summed rank. Root mean square error (RMSE) in the equation were utilized to measure the fitness between predicted and observed values. RMSE was calculated with the following formula:

Summary of selected chlorophyll-related spectral indices reported in the literature.

R is the reflectance at the given wavelength. E.g., R₇₂₀, R₇₄₀ and R₇₅₀ are the spectral reflectance at 720, 740 and 750nm, respectively; R_λ is the spectral reflectance at wavelength λ; D_λ denotes the first derivative value at wavelength λ.

Where P_i and O_i are predicted and observed CCC values and n is the number of samples.