The CMIP workshop was held in Kyoto, Japan during 9-13 March 2026. The workshop brought together researchers from the CMIP community around the globe to present and discuss their research. This post focuses on one of the poster sessions at the workshop, which explored how paleoclimate archives, observational datasets, and model simulations can be used to improve understanding of key Earth system variables and support model evaluation for CMIP7.
Observational datasets
A big focus of the session was on observational datasets.Projects such as the Observations for Model Intercomparison project (obs4MIPs), Quality Assurance framework for Earth Observation (QA4EO), as well as Rapid Evaluation Framework (REF) showcase how observational datasets can be used for climate model evaluation and beyond.
Two key variables emerged as the central themes across this session: precipitation and radiative forcing. Precipitation remains one of the most difficult climate variables to simulate, with large uncertainties across models. The posters cover a broad geographic range, including evaluations of precipitation variability over northern Australia, Europe, and the Southern Ocean, as well as historical changes in the Amazon, East Asia, and multi-model intercomparisons.
Secondly, radiative forcing, as the fundamental driver of anthropogenic climate changes, was also widely discussed. Contributions included comparisons of prescribed and interactive historical stratospheric aerosol forcings in the UK ESM model, analyses using Scaling Cloud Adjustment Observations, and investigations into the role of CO₂ forcing in paleoclimate simulations.
Paleoclimate
The other major topic was on using paleoclimate to understand the climate systems and evaluate model response to different perturbations in the boundary conditions. Several posters highlighted how changes in the climate background mean state (due to changes in greenhouse gas concentrations, orbital parameters, ice sheets, etc.) in the past can have different processes and responses to external forcings.
A notable focus from the Paleoclimate Modelling Intercomparison Project (PMIP) community was their contribution to CMIP7 through the ‘abrupt127k’ experiment for the AR7 FastTrack Assessment. This experiment targets the Last Interglacial (~127 ka), a period when the Arctic may have been seasonally ice-free. It is especially useful for investigating Arctic sea-ice processes, such as the impacts of Arctic melt ponds on the climates, and provides insights relevant to future projections.
Beyond the PMIP experiment, the session also showcased diverse approaches for exploring past climates. These included a new coupled paleo-reanalysis, an eddy-resolving ensemble model, paleo-tuning of a CMIP model, and the Water Isotope Model Intercomparison Project (WisoMIP), alongside protocols for high-resolution PMIP and HighResMIP simulations. There are also simulations of historical periods, multi-model comparison studies, and so forth. I found it very fascinating to see this diversity of methods for evaluating past climates and model performance.
Takeaways from my perspective
From a personal perspective, I found the paleoclimate theme especially engaging, as I also presented a poster myself in this session on the variability of the South Pacific Convergence Zone in LGM and Holocene simulations. I had lots of meaningful discussions regarding this work. Beyond my own research, I was also interested in seeing how results from various paleoclimate reconstruction approaches are used to evaluate model behaviour, examine key variables, and provide insights for future climate projections.
