The PCMDI Metrics Package (PMP)
The PCMDI Metrics Package (PMP) provides a diverse suite of relatively robust high-level summary statistics that gauge differences between models and observations. The emphasis is on the physical system in Earth System Models (ESMs) and their component sub-models. The PMP provides holistic summaries spanning space and time scales for the simulated atmosphere, ocean, ice and land with the CMEC examples available on this site highlighting atmospheric characteristics.
The goals of the PMP are to:
- Provide objective performance summaries of all CMIP DECK + Historical simulations
- Incorporate results spanning six generations of CMIP/AMIP, and to track performance changes through eras of climate model evolution
- Enable modeling groups to use the PMP in their own analysis/development workflow, so to generate near instant feedback at the time of the model simulation completion, rather than await for feedback through the multi-year process that occurs through peer- publication
- Collaborate with US and international expert teams to further diversify the suite of objective tests and expand these in new and novel ways
- Further leverage and advance the conventions and protocols that were developed by PCMDI for CMIP, and use these to facilitate open source software collaboration across research groups and institutions
The PMP summaries include metrics from PCMDI’s research, which rely on several summary diagrams designed by PCMDI (Taylor, 2001; Gleckler et al., 2008) that are now widely adopted by the research community. Well-established metrics generated by various expert teams (e.g., CLIVAR Pacific Region Panel for ENSO) are being implemented into the PMP. The software leverages advanced Python-based model analysis tools that have been developed with over two decades of support from the U.S. Department of Energy, along with the Earth System Grid Federation (ESGF), CMIP6, CMIP5, CMIP3 and earlier data holdings based on-site at LLNL.
As PMP development continues, we welcome collaborations from modeling groups and expert teams to further expand and improve the comprehensiveness of the calculated metrics.
Quick links: Repository,
AMIP and Historical Results, CMIP5/6 interannual variability
Contact: Peter J. Gleckler (firstname.lastname@example.org)