Canadian Centre for Climate Modelling and Analysis A CMIP “Fast-Track” MIP Category? John Scinocca Canadian Centre for Climate Modelling and Analysis

Initial Vision for CMIP ● Continuous MIP Submission: Spread out MIP submission/execution/ analysis more evenly between successive phases of CMIP (Meehl et al. 2014) goal: eliminate/ease pressure on model execution and data analysis that typically concentrates towards the end of a CMIP cycle ● Coordinated MIP Submission: Align MIP submission with start of each new phase of CMIP to allow coordination – harmonized data request – tweak/adjust experiment designs so one set can service multiple MIPs – coordinate science across MIPs – allow modelling centres to plan and budget their limited resources – … ● In the end, the benefits of coordinating MIPs at the start of each phase of CMIP was so great that the idea of continual MIP submission has receded somewhat into the background

After Phase 6 MIP Coordination ● Once all Phase 6 MIP coordination is concluded, there will exist: – a set of well-defined phase 6 experiments including climate forcings – a well-defined phase 6 harmonized data request – phase 6 infrastructure support for the creation and dissemination of model output. – an increasing body of phase 6 model output on the ESGF and the infrastructure at each modelling centre to perform the accompanying numerical simulations ● During the intervening period that leads up to Phase 7 MIP coordination, there is opportunity to exploit the enormous effort that went into coordinating phase 6 MIPs and the resulting body of data and simulations it produced ● Consider the potential for an open, continuous, call for set of “fast-track” MIPs during the inter-phase period leading up to CMIP7

⇒ Continuous “Fast-Track” MIPs vs Coordinated “Phase n” MIPs ● “Fast-Track”, or “inter-phase”, MIPs address emerging climate-change issues that arise from either CMIP6 MIPs or from the community. ● Fast-track MIPS can take advantage of all phase 6 coordination if they: 1) use phase n models 2) use phase n experiments and forcings as the control for new experiments 3) require only the existing phase n data request Unlike phase n MIPs, fast-track MIPs would require essentially no coordination little burden on the CMIP panel and modelling centers ⇒ ● Suggestion: Once the coordination of Phase n MIPs is concluded (ie once all experiments, forcings, data request, and infrastructure), the CMIP panel could consider issuing an open call for fast-track MIPs that satisfy conditions 1) – 3) ● Consider post-CMIP5 examples

CCCma − Post CMIP5 ● At CCCma the decision was taken to build off the enormous investment of effort that went into its CMIP5 contribution (its body of simulations, infrastructure etc.) ● A new, “Applications and Analysis” meeting was instituted to identify and investigate emerging climate questions. eg 1. EMIC models were used to evaluate the role of updated volcanic aerosol forcing (Vernier et al. 2011) in the observed warming hiatus (Solomon et al. 2011) Q. Do conclusions based on EMIC models hold up in full ESMs (Fyfe et al. 2013 - CanESM2) Excellent candidate for a “fast-track” MIP using CMIP5 models ⇒ eg 2. Large ensembles (role of internal variability): − NCAR 40 member (2000-2060 A1B) - (Deser et al. 2012) − CESM-LE 30 member (1920-2100, RCP8.5) - (Kay et al. 2015) − CCCma 50 member (1950-2100, historical/RCP8.5, natural forcings, aerosol-only forcing, stratospheric ozone only forcing) - (Sigmond and Fyfe 2015)

CMIP phase 6 (~6y) phase 7(~6y) phase 6 Fast-Track MIPs coordinate phase 6 MIPs coordinate phase 7 MIPs CMIP6 model execution CMIP7 model execution phase 6 Fast-Track MIPs Advantages of fast-track MIPs – essentially no coordination as they piggyback off phase-n coordination – body of phase n MIP output provides body of control runs to perform further science – allows CMIP to entertain MIPs and support MIP science at any time, as originally intended – no reason why fast-track MIPs for one particular phase might couldn’t extend for longer than the 6y cycle – as phase n fast-track MIPs begin to overlap with the next phase n+1 of CMIP it opens the door for interesting comparisons across MIPs (ie model changes between phases, forcing changes between phases, etc.)