Spatial scale applicability of 4-Scale geometrical optics model canopy reflectance simulation

Area of Science:

• Forestry
• Remote Sensing
• Geospatial Analysis

Background:

• The 4-Scale model is crucial for simulating canopy reflectance.
• Accurate spatial scale determination enhances parameter inversion accuracy (e.g., leaf area index).
• Understanding scale effects is vital for diverse vegetation types.

Purpose of the Study:

• To determine the optimal spatial scale for the 4-Scale model in broad-leaved and mixed forests.
• To evaluate the model's performance across different spatial resolutions.
• To improve canopy reflectance simulation and parameter inversion accuracy.

Main Methods:

• Simulated forest canopy reflectance using the 4-Scale model at scales of 10, 20, 30, 40, and 50 m.
• Resampled Sentinel-2 imagery to various scales using local mean, nearest neighbor, bilinear, and cubic convolution methods.
• Compared simulated canopy reflectance with remote sensing pixel reflectance.

Main Results:

• The 4-Scale model generally underestimated canopy reflectance.
• The 20 m scale showed the worst simulation performance for both forest types, with high RMSE and MAE.
• Simulation accuracy improved at scales >20 m, with optimal performance at 40 m for mixed forest and 30 m for broad-leaved forest.

Conclusions:

• The optimal spatial scale for the 4-Scale model varies between mixed (40 m) and broad-leaved (30 m) forests.
• Scales of 30 m and 40 m provide the most accurate canopy reflectance simulations.
• The 10 m and 20 m scales yielded unstable and less accurate results, indicating limitations for high-precision inversion.