Beyond Static Benchmarking: Using Experimental Manipulations to Evaluate Land Model Assumptions.

Area of Science:

• Earth System Science
• Ecology
• Climate Modeling

Background:

• Land models are crucial for simulating future environmental changes but lack evaluation against experimental manipulation data.
• Experimental data can refine model assumptions and improve predictions of ecosystem responses.

Purpose of the Study:

• To evaluate and compare three versions of the Community Land Model (CLM 4, 4.5, and 5) using experimental data.
• To assess model performance in simulating terrestrial responses to nitrogen (N) and carbon dioxide (CO2) enrichment.

Main Methods:

• Simulated N and CO2 enrichment using CLM versions 4, 4.5, and 5.
• Compared simulation outputs with meta-analyses of observations from experimental manipulations.
• Assessed model performance against FLUXNET-MTE observations for control simulations.

Main Results:

• CLM versions showed a poleward increase in gross primary productivity and reduced bias over time.
• CLM transitioned from nitrogen-limited (CLM4) to CO2-responsive (CLM5).
• CLM5 demonstrated improved agreement with observed ecosystem responses to N and CO2 enrichment.

Conclusions:

• CLM5 offers better agreement with experimental data than previous versions, advancing terrestrial carbon cycle modeling.
• Model simulations revealed shortcomings in plant physiology, allocation, and nutrient uptake parameterizations.
• Integrating experimental manipulation data into model benchmarking is essential for reliable future projections.