Predicting carob tree physiological parameters under different irrigation systems using Random Forest and Planet...

Area of Science:

• Agricultural Science
• Remote Sensing
• Machine Learning

Background:

• Climate change necessitates monitoring plant physiology for drought-resistant crops like carob trees.
• Remote sensing and machine learning are crucial for analyzing large agricultural datasets and field variability.

Purpose of the Study:

• To develop accurate models for predicting carob tree net assimilation and stomatal conductance.
• To analyze seasonal variability and the impact of irrigation systems on carob tree physiology.

Main Methods:

• Utilized Planet satellite images for spectral band reflectance values as model predictors.
• Employed the Random Forest modeling approach, comparing its performance against multiple linear regression.

Main Results:

• Random Forest models achieved high accuracy: R² = 0.81 for net assimilation and R² = 0.70 for stomatal conductance.
• Yellow and red spectral regions were identified as particularly influential predictors.
• No significant differences in intrinsic water use efficiency were found among irrigation systems or rainfed conditions.

Conclusions:

• Combines satellite remote sensing and machine learning for effective precision agriculture.
• Enables efficient monitoring of plant physiological parameters in carob trees.