Nondestructive detection of rape leaf chlorophyll level based on Vis-NIR spectroscopy

Area of Science:

• Agricultural Science
• Plant Physiology
• Spectroscopy

Background:

• Chlorophyll content is crucial for assessing plant growth and optimizing agricultural water and fertilizer management.
• Hyperspectral reflectance analysis offers a non-destructive method for monitoring plant physiological status.
• Effective spectral data preprocessing is essential for accurate chlorophyll estimation.

Purpose of the Study:

• To quantitatively investigate the impact of spectral data preprocessing methods on hyperspectral feature extraction for rapeseed chlorophyll content.
• To develop and optimize a predictive model for estimating chlorophyll content in rapeseed leaves.
• To identify the most effective preprocessing and transformation techniques for enhancing chlorophyll estimation accuracy.

Main Methods:

• Collected hyperspectral reflectance data (350-2500 nm) from rapeseed leaves using an ASD FieldSpec Pro spectrometer.
• Applied spectral preprocessing techniques including Savitzky-Golay smoothing (SG), Multi-Scale Correction (MSC), and Standard Normal Variate (SNV) transformation.
• Utilized first derivative (FD) and reciprocal logarithm (LOG) transformations, combined with Partial Least Squares Regression (PLSR) for model development.

Main Results:

• The optimal preprocessing combination was identified as R+SG+SNV+LOG+FD, yielding high calibration (Rc²=0.97) and validation (Rv²=0.98) accuracy.
• The established PLSR model demonstrated excellent predictive performance with a validation RPD of 7.52.
• The optimized model proved stable and precise for rapid, regional chlorophyll content monitoring in rapeseed.

Conclusions:

• Spectral data preprocessing significantly influences the accuracy of chlorophyll content estimation in rapeseed.
• The combination of specific preprocessing steps (R+SG+SNV+LOG+FD) and PLSR provides a robust method for monitoring chlorophyll.
• This approach enables efficient and precise assessment of rapeseed physiological status, aiding agricultural management decisions.