Estimation of plant leaf water content based on spectroscopy

Area of Science:

• Plant Physiology
• Remote Sensing
• Spectroscopy

Background:

• Leaf water content is a critical indicator of plant health and growth.
• Spectroscopy offers an effective method for monitoring plant physiological conditions.
• Developing accurate, broadly applicable models for leaf water content estimation is essential.

Purpose of the Study:

• To improve the accuracy of leaf water content estimation models.
• To develop models applicable to diverse plant species.
• To visualize leaf water distribution using hyperspectral imaging.

Main Methods:

• Collected 1,680 hyperspectral and water content data points from peach tree leaves.
• Developed estimation models using vegetation indices (NISDI) and characteristic wavelengths (CARS-RF).
• Validated model generalization using apple tree and lettuce leaf data.

Main Results:

• The optimal model, a linear regression based on NISDI (3 wavelengths), achieved R_P²=0.9636.
• The CARS-RF model (12 wavelengths) showed higher accuracy (R_P²=0.9861) but used more wavelengths.
• The chosen NISDI model accurately estimated water content in apple (R_P²=0.9504) and lettuce (R_P²=0.8211) leaves.

Conclusions:

• The developed model demonstrates generalization ability for estimating leaf water content in related species.
• The model shows some performance degradation across different plant growth forms.
• This study provides a reference for creating generalized plant leaf water content estimation models.