The Terrestrial Biosphere Model Farm.

Area of Science:

• Earth System Science
• Climate Modeling
• Terrestrial Ecosystems

Background:

• Model Intercomparison Projects (MIPs) are crucial for understanding land surface responses to climate change.
• Conducting MIPs is complex due to the decentralized nature of global modeling teams and common protocol execution.

Purpose of the Study:

• To explore the feasibility and implications of centralizing multiple terrestrial biosphere models on a single supercomputing system for MIPs.
• To identify the benefits and challenges associated with this centralized modeling approach.

Main Methods:

• Nine offline terrestrial biosphere models (CABLE, CENTURY, HyLand, ISAM, JULES, LPJ-GUESS, ORCHIDEE, SiB-3, SiB-CASA) were run on a supercomputing system.
• Models were integrated into a software framework using common forcing data and protocols from the Multi-scale Synthesis and Terrestrial Model Intercomparison Project (MsTMIP) for 1901-2100.

Main Results:

• Centralization simplifies processing multiple models for MIPs and ensures consistent protocol execution, potentially reducing output variability.
• Unique multimodel experiments were enabled, generating novel data for analysis.
• Significant challenges were identified, including high technological demands for data storage/transfer, model version lags, and the continued need for core development team expertise.

Conclusions:

• Centralizing models on a supercomputing system offers benefits for MIPs but presents substantial technological and logistical hurdles.
• Integrating with open-source, cloud-based systems like the Predictive Ecosystem Analyzer (PEcAn) may offer a viable solution to overcome identified challenges.