Assessing and quantifying the saturation effect of forest aboveground biomass mapping using Landsat 8, Sentinel 1/2

Area of Science:

• Forestry
• Remote Sensing
• Geospatial Analysis

Background:

• Remote sensing is crucial for forest aboveground biomass (AGB) mapping, but signal saturation limits accuracy in dense forests.
• Existing methods struggle to quantify saturation effects, hindering reliable AGB estimation.
• A standardized metric is needed to evaluate saturation and mitigation strategies.

Purpose of the Study:

• To propose and validate a quantitative metric (SAT) for assessing saturation in forest AGB mapping.
• To evaluate saturation levels of Landsat 8 and Sentinel-1/2 data across diverse tree species.
• To analyze the impact of data integration and machine learning models on saturation.

Main Methods:

• Developed the SAT metric to separate saturation effects from mapping errors.
• Assessed Landsat 8 and Sentinel-1/2 data saturation for ten tree species in southern China.
• Quantitatively analyzed multi-source data integration and nine machine learning models.

Main Results:

• Tree species significantly influenced AGB estimation performance and saturation levels (SAT: 48.15–147.93 Mg/ha).
• Integrated decision tree-based ensemble models showed improved stability and higher saturation levels.
• Sentinel-1 data and optical texture features enhanced AGB estimation accuracy and saturation levels.

Conclusions:

• The proposed SAT metric provides explicit quantification of saturation-induced underestimation in high-biomass forests.
• SAT supports informed selection of sensor combinations and modeling strategies for AGB mapping.
• This research enhances the reliability and interpretability of remote-sensing-based forest AGB products, especially in dense forest areas.