Identifying the dominant climate-driven uncertainties in modeling gross primary productivity

Area of Science:

• Ecology
• Climate Science
• Earth System Science

Background:

• Accurate simulation of gross primary productivity (GPP) is vital for global carbon budget estimations.
• Uncertainties in GPP modeling often stem from biases in climate forcing data, which remain poorly quantified.
• FLUXNET sites provide crucial ground-truth data for validating vegetation models.

Purpose of the Study:

• To quantify climate-driven uncertainties in GPP simulations at 91 FLUXNET sites.
• To compare GPP simulations using ground-based meteorology versus assimilated reanalyses.
• To identify key meteorological factors contributing to GPP uncertainties.

Main Methods:

• Utilized a well-validated vegetation model for site-level GPP simulations.
• Compared simulations using site-level meteorology and CO2 with those using Modern-Era Retrospective Analysis (MERRA) reanalyses.
• Conducted sensitivity tests by varying meteorological and CO2 inputs.

Main Results:

• MERRA reanalyses increased GPP root mean square error (RMSE) by 30% compared to site-level data.
• GPP uncertainties correlated linearly with biases in meteorological forcing data.
• Diffuse photosynthetically active radiation (PAR) was identified as a dominant factor modulating GPP uncertainties.

Conclusions:

• Biases in meteorological forcings, particularly diurnal and seasonal variability, introduce significant GPP uncertainties.
• Simulations using climate reanalyses for dynamic global vegetation models require caution.
• Urgent improvements in the climatic variability, especially diffuse radiation, within reanalysis datasets are needed.