A global 0.05° gross primary production dataset from 2001 to 2024 generated using a hybrid LSTM framework

Area of Science:

• Earth System Science
• Ecology
• Climate Science

Background:

• Gross Primary Production (GPP) is crucial for global carbon and energy cycles.
• Existing GPP models show significant discrepancies, particularly in seasonal and interannual patterns.
• Accurate GPP estimation is vital for understanding terrestrial carbon dynamics.

Purpose of the Study:

• To introduce TRAX GPP, a novel global dataset for estimating Gross Primary Production.
• To provide a high-resolution (0.05°), monthly dataset covering 2001-2024.
• To improve the accuracy of GPP estimation and monitoring of terrestrial carbon dynamics.

Main Methods:

• Developed a hybrid Long Short-Term Memory (LSTM) framework integrating Robust Seasonal-Trend decomposition (RSTL).
• Incorporated satellite-derived vegetation and meteorological data.
• Utilized atmospheric CO2 assimilation constraints to model CO2 fertilization effects.

Main Results:

• TRAX GPP dataset demonstrates strong agreement with tower flux measurements.
• The model excels at capturing seasonal dynamics and interannual GPP fluctuations.
• TRAX GPP reduces biases found in global GPP totals and trends compared to previous products.

Conclusions:

• TRAX GPP offers a significant advancement in global GPP estimation.
• The dataset provides a reliable tool for refined monitoring of terrestrial carbon dynamics.
• This work establishes a new benchmark for GPP data products.