Enhancing the PCMDI Metrics Package for Comprehensive Evaluation of Earth System Models in CMIP

Earth System Models (ESMs) are essential for understanding climate dynamics and informing policy decisions. This presentation focuses on the PCMDI Metrics Package (PMP), an open-source, Python-based framework designed for objective "quick-look" comparisons and benchmarking of ESMs against the latest observational data. The PMP has been instrumental in systematically evaluating thousands of simulations from Coupled Model Intercomparison Projects (CMIPs), with a primary focus on physical climate with atmospheric mean and variability.

As we prepare for the upcoming CMIP7, our ongoing work aims to enhance the PMP's capabilities to support modeling groups throughout their development cycles. The PMP offers a diverse suite of evaluation metrics, including large- to global-scale climatology, annual cycle, and variability characteristics associated with ENSO, MJO, and numerous extra-tropical modes, and also includes key measures of simulated sea-ice and ocean states. Notably, the PMP provides a database of pre-calculated statistics for CMIP6 models, facilitating easier comparisons for modeling centers as they assess their results against established benchmarks.

Current ongoing enhancements include the evaluation of the Quasi-Biennial Oscillation (QBO) and its teleconnections to the MJO, atmospheric blocking, and atmospheric river patterns using Machine Learning algorithms. Additionally, we are implementing planetary-scale assessments through Hadley cell expansion metrics. The PMP is also evolving to accommodate higher-resolution simulations from HighResMIPs, cloud-resolving E3SM experiments, and regionally downscaled products.

This presentation will highlight the importance of routine model evaluation, introduce the latest advancements in the PMP, and discuss opportunities for community engagement and collaboration. We invite feedback and suggestions from the community to further enhance our tools and methodologies.