CMIP Community Workshop 2026

The 2026 Coupled Model Intercomparison Project (CMIP) Community Workshop took place at the Kyoto International Conference Centre in early March. More than 600 researchers from around the globe (Figure 1) came together at this historic venue (Figure 2), the birthplace of the 1997 Kyoto protocol, to discuss findings from CMIP6 and plans for CMIP7. Helene Hewitt (Met Office, UK) and John Dunne (NOAA-GFDL, USA) co-chaired the meeting, supported by an international scientific organising committee and a local organising committee led by Masa Watanabe and Tomoki Miyakawa (University of Tokyo, Japan). The week-long conference, which featured 99 talks and 354 poster presentations (Figure 1), spanned the broad field of Earth system modelling and brought together modelling centres, data users and other stakeholders, such as policymakers. Alongside these presentations, several side sessions provided an opportunity for community-wide discussions ranging from technical demonstrations of new software to thought-provoking conversations about inequity in the CMIP community. Founded in 1995 by the World Climate Research Programme (WCRP), the Coupled Model Intercomparison Project (CMIP) is designed to coordinate global climate modelling efforts under a common experimental protocol (Durack et al., 2025). The goal of this project is to allow for comparison between climate models over a common set of experiments, and to focus the modelling community's efforts on improving models and best supporting the Intergovernmental Panel on Climate Change (IPCC), and other similar climate assessments (Meehl et al., 1997). These experiments cover a wide range of scientific questions: from core simulations of the Earth's recent past and future to more targeted simulations investigating the carbon cycle, tipping points, ocean dynamics, cloud physics and much more. CMIP6 is the most recent (sixth) phase of the project (Eyring et al., 2016), bringing together simulations from 48 institutions and 26 countries. The advances in knowledge made possible by these simulations fed into the sixth IPCC Assessment Report (AR6) and will continue to inform AR7, which is due to be published in late 2029. CMIP7 has been launched in 2026 (Dunne et al., 2025), which will continue the legacy of CMIP6. Whilst the premise of the project remains the same, updates have been made to experimental designs, forcing datasets, infrastructure (data storage and handling, software, documentation, etc.) and the models themselves, all reflecting the advancement of our understanding in recent years. One key development in experimental design across the CMIP is a general shift towards emission-driven simulations in which CO2 emissions, rather than concentrations, are prescribed. Models have also evolved individually by incorporating more complex processes and parametrisations, and also tackling persistent biases. The full extent and effectiveness of these model updates will become clear once model data becomes available. Each of these themes featured two keynotes, several scientific talks and a poster session (Figure 3). The first keynote talk for Theme 1 was given by Isla Simpson (NCAR, USA) who discussed a recent review of the discrepancies between trends in observations and CMIP6 climate models (Simpson et al., 2025). The talk focussed on areas in which models and observations disagree – such as sea surface warming patterns in the tropical Pacific – and areas in which model and observational trends agree – such as the weakening of the Eurasian jet. Brian Soden (University of Miami, USA) then followed with the second Theme 1 keynote, showing a new method for calculating Effective Radiative Forcing from aerosol-cloud interactions (ERFaci). This refers to the net change in the radiative energy balance at the top of the atmosphere due to interactions between anthropogenic aerosols and cloud properties. Explicitly including aerosol activation rate in estimating ERFaci leads to a reduced estimate, but with considerably smaller uncertainty (Park et al., 2025). In Theme 2, Jennifer Mecking (National Oceanography Centre, UK) talked about how Atlantic meridional heat and freshwater transports are projected to change in CMIP6 models, and how these transports relate to the expected weakening of the Atlantic Meridional Overturning Circulation (AMOC) in the future. Whilst meridional heat transport is likely to decrease due to AMOC weakening, the freshwater transport will likely see little change as salinity-driven azonal changes cancel the zonal overturning trends (Mecking and Drijfhout, 2023). We then heard from Irina Melnikova (National Institute for Environmental Studies, Japan) who analysed the mechanisms of projected Amazon dieback in CMIP5 and CMIP6 models, showing that this process may start within the twenty-first century under 5.0 ± 1.9°C of warming (Melnikova et al., 2025). Not every model captures this dieback, however, highlighting the differences in model responses to anthropogenic climate change. Daisuke Takasuka (Tohoku University, Japan) began the Theme 3 keynotes by showing the benefits of using a global km-scale model to analyse large-scale variability (Takasuka et al., 2024). By improving the representation of submonthly-scale phenomena, the Nonhydrostatic Icosahedral Atmospheric Model (NICAM) can better resolve non-linear feedbacks controlling larger-scale phenomena. Finally, Jesus Fernandez (CSIC-Universidad de Cantabria, Spain) introduced developments in the Coordinated Regional Climate Downscaling Experiment (CORDEX) protocol in preparation for CMIP7. CORDEX will bring together regional and global climate modelling to downscale CMIP7 models for use in more localised contexts. Whilst these keynote talks give an insight into the wide breadth of the workshop, we were also treated to a smorgasbord of scientific analysis throughout the parallel oral and poster sessions. Though there are too many highlights to mention here, the full program (including downloadable presentations) can be viewed online (https://cmip2026.org). Whilst the CMIP Community Workshop provided an opportunity to discuss the analysis of CMIP6 output, it also served to facilitate planning for CMIP7. Helene Hewitt began the workshop with an overview of CMIP7, including the motivating vision for the next phase of CMIP and the Assessment Fast Track (Dunne et al., 2025). This vision is structured around four broad scientific questions: how patterns of sea-surface warming will evolve; how dangerous weather patterns and associated extremes will change; how the Earth system will respond to human efforts to manage the carbon cycle; and what risks of tipping points or other irreversible changes may arise under different climate trajectories. Together, these questions show how CMIP7 is intended to link fundamental climate-process understanding with the needs of future assessments and policy-relevant applications. This summary was then expanded by Isla Simpson, who outlined the development of the Assessment Fast Track – a subset of CMIP7 experiments which have been earmarked as highest priority for the upcoming IPCC AR7 (see Dunne et al. (2025) for the experiment list), and aim to achieve data delivery by December 2026. Tomoko Hasegawa (Ritsumeikan University, Japan) then provided an overview of the new CMIP7 future scenarios, highlighting how they relate to previous generations of the scenarios and to different policy options. A detailed description of the new scenario framework can be found in the recent publication by Van Vuuren et al. (2026). Dedicated side sessions discussed the new Rapid Evaluation Framework, access to CMIP data and the importance of Global South inclusion within CMIP, among a range of other topics. These side sessions allowed for more in-depth discussions and encouraged engagement as a community. By bringing together modelling centres and data users, the workshop provided a unique opportunity for two-way dialogue about goals, needs and challenges ahead of CMIP7. Early career researchers (ECRs) made up more than one third of the attendees at the workshop (Figure 1), highlighting the strong presence of emerging talent across the climate-modelling community. A particular highlight of the event was the ECR workshop. Alongside discussion of the wide range of career paths within the CMIP community, the session featured a structured networking exercise designed to help participants build connections across the community. Delegates were first divided into groups and then repeatedly reorganised according to different shared characteristics, including research subject, research area, mobility and career stage. In each round, participants were given labels, spent several minutes in discussion and then moved into a new grouping. This rotating format allowed attendees to meet many different people from a range of backgrounds while still sharing at least one point of common ground in every conversation. The workshop concluded with breakout discussions on how ECRs can contribute to the CMIP community, the barriers that currently limit that contribution and possible ways to overcome them. These barriers included the fact that ECRs are not typically funded to work on CMIP directly, as well as the limited availability of platforms through which they can connect with the wider community, such as webinars and training courses. It was also noted that some of these challenges can be particularly acute for ECRs in the Global South due to a lack of access to funding, visibility, opportunities and the large-scale computing infrastructure required for climate modelling. Overall, the workshop provided ECRs with valuable opportunities to connect with one another, engage with the development and analysis of CMIP7, and demonstrate the breadth of upcoming talent that will help shape the future of the CMIP community. We would like to thank the many people who came together to make the CMIP Community Workshop 2026 possible: the CMIP panel, the WCRP ESMO Infrastructure Panel (WIP), the scientific and local organising committees, the CMIP International Project Office (IPO), the Kyoto International Conference Centre staff, and the CMIP 2026 sponsors. We would also like to thank Beth Dingley (CMIP-IPO) for providing images and attendee infographic. Daniel T. Skinner: Conceptualization; writing – original draft; project administration. Yousheng Li: Writing – review and editing. Hanna Heidemann: Writing – review and editing. Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.