A hybrid model for refining gross primary productivity estimation by integrating multiple environmental factors

Area of Science:

• Ecology
• Environmental Science
• Remote Sensing

Background:

• Estimating gross primary productivity (GPP) is challenged by environmental factors' complexity in light use efficiency (LUE) models.
• Existing physical formulas struggle to capture diverse environmental constraints on maximum LUE (εmax).

Purpose of the Study:

• To develop a hybrid model integrating ecological stress factors and seasonal canopy structure for improved GPP estimation.
• To enhance the accuracy of light use efficiency models by incorporating machine learning techniques.

Main Methods:

• A hybrid model (TL-CRF) was developed, combining a two-leaf LUE (TL-LUE) model with the random forest (RF) technique.
• Seasonal clumping index (CI) variations were estimated globally for different vegetation types and leaf life cycle stages.
• Ecological stress factors were integrated into the TL-LUE model using an RF submodule.

Main Results:

• The TL-CRF model effectively incorporates complex environmental variables to scale theoretical εmax to actual ε.
• The hybrid approach leverages the strengths of both process-based and data-driven models.
• Significant improvements in global GPP estimation were achieved.

Conclusions:

• The proposed TL-CRF model offers a more robust approach to global GPP estimation.
• Integrating machine learning with process-based models enhances ecological modeling capabilities.
• Accurate GPP estimation is crucial for understanding carbon cycling and climate change impacts.