Tuning Earth System Models Without Integrating to Statistical Equilibrium.

Area of Science:

• Climate Science
• Computational Modeling
• Statistical Analysis

Background:

• Earth System Models (ESMs) are crucial for climate prediction but parameter estimation is challenging.
• Simulations often exhibit climate drift before reaching statistical equilibrium, complicating parameter estimation.
• Existing methods can yield unsatisfactory parameter estimates, particularly for short or drifting simulations.

Purpose of the Study:

• To develop and evaluate a novel algorithm for robust parameter estimation in Earth System Models (ESMs).
• To address limitations in current methods when dealing with non-equilibrium simulations and climate drift.
• To improve the accuracy and reliability of parameter estimates for ESMs, specifically the Community Earth System Model version 2 (CESM2).

Main Methods:

• The study proposes treating ESM time series as outputs of an autoregressive process.
• A new strategy divides the parameter estimation into two linear phases to overcome nonlinear system limitations.
• The algorithm is applied to estimate parameters within the convection scheme of the Community Earth System Model version 2 (CESM2).

Main Results:

• The modified algorithm successfully produces accurate parameter estimates from perturbation runs as short as 2 years.
• The method effectively handles simulations exhibiting climate drift, yielding comparable accuracy to methods that do not explicitly account for it.
• Initial results demonstrate the autoregressive approach's potential for rigorous statistical framework-based model tuning.

Conclusions:

• The proposed two-phase linear algorithm offers a promising strategy for tuning Earth System Models (ESMs) by explicitly accounting for climate drift.
• Accurate parameter estimation is achievable even with short simulations and inherent climate drift.
• Current performance limitations are likely technical and amenable to future investigation and improvement.