[Band depth analysis and partial least square regression based winter wheat biomass estimation using hyperspectral...

Area of Science:

• Agricultural Remote Sensing
• Spectroscopy
• Biomass Estimation

Context:

• Traditional vegetation indices often saturate at high biomass levels, limiting accurate crop biomass estimation.
• Band depth analysis offers a novel approach to spectral data interpretation in the visible domain (550-750 nm).
• Partial Least Square Regression (PLSR) is a robust statistical method for analyzing spectral data and predicting variables.

Purpose:

• To develop and evaluate a winter wheat biomass estimation model using band depth analysis and PLSR.
• To compare the accuracy of the proposed model against models based on conventional vegetation indices.
• To address the spectral saturation problem inherent in existing biomass estimation techniques.

Summary:

• Band depth analysis metrics, including band depth ratio (BDR), were used alongside PLSR to model winter wheat biomass.
• Models integrating band depth analysis and PLSR demonstrated superior accuracy compared to traditional vegetation index models.
• The combination of BDR and PLSR achieved the highest accuracy (R² = 0.792, RMSE = 0.164 kg m⁻²).

Impact:

• The combined band depth analysis and PLSR approach effectively overcomes the saturation issue in biomass estimation.
• This method significantly improves the accuracy of winter wheat biomass estimation, particularly for large biomass values.
• Provides a more reliable tool for agricultural monitoring and yield prediction using spectral data.