Nitrogen content estimation of apple trees based on simulated satellite remote sensing data.

Area of Science:

• Agricultural Science
• Remote Sensing
• Plant Nutrition

Background:

• Accurate apple tree nitrogen content diagnosis is vital for precision fertilization.
• Satellite remote sensing offers potential for regional-scale nutrient assessment.
• Variability in satellite data necessitates evaluation for apple nitrogen inversion.

Purpose of the Study:

• To systematically evaluate multi-source satellite data applicability for apple tree nitrogen content estimation.
• To identify optimal spectral bands and modeling strategies across key phenological stages.
• To provide a scientific basis for regional precision fertilization systems.

Main Methods:

• Collected canopy hyperspectral data during three phenological periods (NGS, NSS, ASS).
• Resampled data to match satellite spectral resolutions.
• Employed correlation coefficient, PLS regression, SVM, and BPNN for band selection and model construction.

Main Results:

• Visible, red edge, near-infrared, and yellow edge bands were identified as sensitive to nitrogen content.
• The Support Vector Machine model using Sentinel-2 simulated data demonstrated optimal performance.
• The optimal model achieved R² values of 0.81 (training) and 0.61 (validation).

Conclusions:

• Satellite remote sensing, particularly Sentinel-2, is effective for regional apple tree nitrogen estimation.
• Phenological stages significantly influence nitrogen-sensitive spectral bands.
• This study provides a valuable technical paradigm for remote sensing-based nutrient management in orchards.