Estimation models for maize leaf water content at various stages using near-infrared spectroscopy

The leaf water content (LWC) of maize is crucial for its photosynthesis and overall growth. Accurate and nondestructive estimation of LWC can enhance the monitoring of growth status and optimize field management practices. In this study, spectral data in the range of 900-1700 nm from maize leaves at various positions and growth stages were used to identify the optimal measurement method and develop models for quantitative estimation of LWC. Principal component analysis, competitive adaptive reweighted sampling, and successive projections algorithms were employed to extract effective wavelengths. These wavelengths were then combined with partial least squares regression and support vector regression algorithms to build quantitative models for LWC estimation and to determine the optimal measurement position on the leaves. In addition, a deep learning model based on a multilayer perceptron (MLP) and permutation importance (PI) method was developed. The results indicated that the middle of maize leaves is the optimal position for LWC estimation. The MLP-PI model using the extracted 12 wavelengths, achieved a coefficient of determination of 0.91 and a root mean square error of 1.23 %, respectively. This study demonstrates that the middle position of maize leaves, when combined with the near-infrared spectra and deep learning techniques, provides a robust approach for non-destructive LWC estimation.