[Research on Modeling Method for Chlorophyll Content Fine Measurement Based on Neural Network]

Area of Science:

• Plant Physiology
• Spectroscopy
• Machine Learning

Background:

• SPAD values, a common measure of chlorophyll, are easily influenced by leaf thickness and water content.
• Accurate chlorophyll content estimation is crucial for plant health monitoring and agricultural applications.

Purpose of the Study:

• To develop a more accurate method for retrieving chlorophyll content in living plant leaves.
• To evaluate the effectiveness of a multi-parameter BP neural network model for chlorophyll estimation.

Main Methods:

• Leaf transmittance was measured at specific wavelengths (650 nm, 940 nm, 1450 nm) to obtain SPAD values and water index (WI).
• Leaf thickness was measured using a micrometer.
• Chlorophyll content was determined spectrophotometrically.
• A BP neural network model incorporated SPAD, WI, and thickness to predict chlorophyll content.

Main Results:

• The multi-parameter BP neural network model significantly improved prediction accuracy compared to a single-parameter SPAD model.
• Average absolute relative error for chlorophyll content decreased from 7.55% to 5.22%.
• The fitting determination coefficient increased from 0.83 to 0.93.

Conclusions:

• A multi-parameter BP neural network model effectively enhances the prediction accuracy of living plant chlorophyll content.
• This method offers a more robust approach to chlorophyll estimation, overcoming limitations of single-parameter methods.