Regional-scale forest aboveground biomass mapping using temporally consistent ICESat-2, Landsat, and field inventory...

Area of Science:

• Forestry and Remote Sensing
• Geospatial Analysis
• Ecology

Background:

• Accurate estimation of forest aboveground biomass (AGB) is vital for understanding carbon storage, ecosystem health, and biodiversity.
• Southeastern US forests are significant for carbon sequestration, but large-area AGB mapping faces challenges.
• Recent Earth-observing missions offer new opportunities for vegetation characterization.

Purpose of the Study:

• To assess the synergistic utility of Landsat and ICESat-2 data for regional AGB estimation.
• To identify the optimal modeling technique for AGB estimation using satellite-derived variables.
• To develop a high-resolution AGB baseline map for southeastern US forests.

Main Methods:

• Compared machine learning (Random Forest, Support Vector Machine) and geostatistical (Kriging) approaches.
• Utilized ICESat-2 canopy height, Landsat imagery, digital elevation models, and canopy cover data.
• Developed and validated AGB models across approximately 254,266 km².

Main Results:

• Support Vector Machine Regression Kriging (SVMRK) demonstrated superior performance (R² = 0.61, RMSE = 23.99 Mg/ha).
• Models achieved R² values from 0.34 to 0.61 and RMSEs between 22 and 31 Mg/ha.
• A 30 m resolution AGB baseline map for 2020 was generated.

Conclusions:

• The combined use of Landsat and ICESat-2 data provides a feasible approach for large-area AGB mapping.
• SVMRK is a highly effective method for integrating diverse data sources for AGB estimation.
• The generated AGB map serves as a valuable baseline for regional forest monitoring.