Skip to main content

PRocess-based climate sIMulation: AdVances in high resolution modelling and European climate Risk Assessment

Periodic Reporting for period 3 - PRIMAVERA (PRocess-based climate sIMulation: AdVances in high resolution modelling and European climate Risk Assessment)

Reporting period: 2018-05-01 to 2020-07-31

The overarching goal of PRIMAVERA was to develop a new generation of advanced and well-evaluated high-resolution global climate models. The experiments and deliverables in PRIMAVERA, including CMIP6 HighResMIP, were designed to address some of the most pressing questions about regional climate change, in order to provide governments, business and society with reliable information on weather and climate risk over the next few decades. Our hypothesis was that many of the long-standing errors and biases in current models could be reduced by using much higher model resolutions (finer grids), so that key climate processes are better represented and hence the resultant projections would be more reliable.
PRIMAVERA has fully achieved its model development and high-fidelity simulation goals. All seven modelling groups used our common experimental design, HighResMIP, so that our simulations would be comparable to each other, and able to be distributed to the wider climate community. Consequently, we can now confidently and robustly say that increasing resolution improves climate simulation, and via our analysis and publications this message has been clearly conveyed in the draft IPCC AR6 report. Our further developments in the “Frontiers of Climate Modelling” work, down to even finer scales, paves the way for future projects, with initial results also indicating that projections of European climate change may have been underestimated in current models.
PRIMAVERA was also designed to enhance communication between climate scientists and society (spanning individuals and companies to policy advice). We designed aspects of the project together with end-users (for example renewable energy and reinsurance companies), and provided new types of information and products relevant to them (11 companies are interested in our windstorm dataset, and the User interface Platform hosts accessible visualisations and other products from our website).
PRIMAVERA produced a new CMIP6 simulation design, HighResMIP, shared our data and code for joint analysis on a common platform (JASMIN), and published the data to the CMIP ESGF archive for community use (about 12,000 years of simulation, 1.7 PB, comparable to the whole of the CMIP5 archive). New functions (14) were developed and integrated into the ESMValTool community analysis package. We developed new model components suited to higher resolution that will become standard in next generation of climate models: aerosol-microphysics, ocean mixing, sea ice melt ponds, land surface physics, stochastic physics schemes. We developed a new generation of models, using the eddy-rich ocean, and sub-10km atmosphere (and further in DYAMOND, at 5km resolution), and an unstructured mesh ocean-sea ice model FESOM.
We have so far published 91+ peer-reviewed articles, with 54+ submitted or in preparation. We found robust changes across the multi-model ensemble for a range of important climate processes, and several chains of mechanisms were better represented at higher resolution (for example an improved Gulf Stream influencing the jet stream, storm track and blocking). For our range of resolutions, the ocean provided the bigger impact, and we showed that some changes with model resolution could lead to impacts which have implications for future change over Europe. This suggests that non-eddy resolving simulations like those uniformly used in CMIP6 are missing key processes, which may have a significant impact on climate projections and derived risk assessments.
In addition to the peer-reviewed articles produced for fellow scientists, we have contributed to and been widely cited in the draft IPCC AR6 report where the CMIP6 HighResMIP is one of the exciting new sources of information which is also being used in national climate change assessments. We have used our website, User Interface Platform and associated Data Viewer, factsheets, webinars and storymaps to engage with our end-users and others. We have actively engaged with other European projects, sharing data, analyses and insights, with several new projects predicated on use of our data, code or the models we have developed. We have made strong links with many international communities (e.g. CLIVAR), and we have ongoing collaborative analyses with many other groups (20-30 scientists given access to JASMIN) spanning aspects such as climate extremes, storm and surge modelling.
We organised sessions, town halls and side events at major conferences. Many masters, PhD and post-doc studies have been started during PRIMAVERA, and these will continue after the project. There will be a legacy in the number of young researchers who have been engaged with the project and whose research paths will be influenced by it. For example, a climate modelling summer school was run in 2019.
PRIMAVERA has advanced the state-of-the-art in global climate modelling, in the context of the CMIP6 exercise, by: reducing the grid-spacing of centennial scale global simulations from 100km to 25km; developing and applying new physical parameterisations suitable for higher resolution; coordinating the operation of seven models at 25km, run at different centres but all uploading data to a common analysis platform. Many of these developments are already being used in other new and existing projects.
PRIMAVERA has further developed next-generation models, in the sub-10km range in both the atmosphere and ocean. Four groups produced simulations with an eddy-rich ocean, initial results from which suggest the potential for significant changes to future climate risk over Europe compared to lower resolution, due to improved representation of the Gulf Stream. Two groups also created 10km and 5km versions of their atmospheric model, which can be operated with and without a convection parameterisation. These were included in the DYAMOND project which includes seven other international models at sub-5km resolution, and they show promise for the first time in the area of realistic intensity and intensification of tropical cyclones.
The ability to produce centennial scale global simulations at resolutions that were previously only possible with regional downscaling has enabled our community to involve stakeholders that have local and global exposure. Data were used for analysis of wind energy across a wide European domain, of flooding due to extreme precipitation in various European catchments, but also enabling the same studies, for companies with overseas interests, to be run in other regions. A windstorm dataset for reinsurance, and tropical cyclone risks in the Caribbean and Gulf of Mexico, were also enabled by both the exceptionally large sample provided by the PRIMAVERA ensemble, together with the skill in reproducing for example TC interannual variability and track location.
The processes emerging in PRIMAVERA models have already started to address “out of trajectory” behaviour, i.e. climate risks that are not represented in CMIP-type models due to bias. A prominent example is post-tropical cyclones, intense mid-latitude storms with a tropical origin, which can impact mid-latitude coasts with significant consequences. The promise of improved predictive skill resulting from the improved simulation of eddy-mean flow interaction in the new generation of global models, likely at sub-10km, means that the investments we have made in advancing models and analyses provides a strong basis for future advances.
Infographic illustrating the work of PRIMAVERA in end-user engagement and interaction
The added value of using high resolution in climate models for representing surface air temperature
The projected multi-model future change in Atlantic Meridional Overturning with model resolution
The multi-model tropical cyclone intensities with model resolution
The difference between low and high resolution ocean models
Illustration of “ice growth-ice thickness feedback” in the Arctic
The high resolution AWI ocean model adjusted to the baroclinic Rossby radius
The projected future change in precipitation over northern Europe from HighResMIP, CMIP6 and UKCP
Tropical Atlantic sea surface temperature bias in the Max Planck Institute Earth System Model