Updated nextGEMS models for the next simulation cycle

22. June 2022

by Thomas Rackow*, Daniel Klocke**, and the MPI-M and ECMWF-AWI modelling teams***

The nextGEMS model development is structured into simulation cycles. Each simulation cycle is followed by a hackathon, where simulation results are evaluated extensively by the nextGEMS community. The first nextGEMS hackathon in Berlin in October 2021 analysed the very first simulations. Based on the results, the two models participating in the project, IFS and ICON, were updated significantly for the cycle 2 simulations. These simulations where just completed and the unique datasets are waiting for the community to analyse them at the upcoming hackathon at the end of June. 
In case of IFS, besides updates to the atmospheric model component (read more here) and a more realistic treatment of snow, another update has been the use of a higher-resolution ocean that resolves eddies over large parts of the globe. Eddies in the ocean impact exchanges of energy and matter across the ocean-atmosphere interface, they transport heat both horizontally and vertically, and they were shown to alter projections of global climate in a warming world.

Figure 1: Ocean resolution used in the IFS-FESOM simulations for Cycle 2. The grid points of the NG5 configuration are concentrated in higher latitudes in order to resolve ocean eddies over larger areas of the globe compared to a more homogeneous distribution of grid points.

The operational high-resolution 9km forecasts at ECMWF include an ocean that applies a ¼ degree resolution (about 25km at the equator). While many coupled effects such as the atmospheric and oceanographic interaction during tropical cyclone conditions (Mogensen et al. 2017) can be realistically simulated at this resolution, ocean eddies are still only ‘permitted’ in mid-latitudes compared to the even coarser 1 degree standard-resolution climate models. This is far from the goal to explicitly resolve mesoscale ocean eddies all around the globe and is a potential source of many long-standing biases in climate models. Importantly, mesoscale features can also affect the predictability of European weather downstream of the Gulf Stream area.

A recent development for the IFS Cycle 2 simulations is a nextGEMS grid (NG5) for FESOM2, which was designed to be of equivalent size to the ICON-5km ocean grid with about 7.5 million surface nodes (Figure 1). Making use of the multi-resolution capability of FESOM2, relatively more surface nodes were concentrated in higher latitudes in order to extend the area where eddies are resolved – from the mid-latitudes into higher latitudes. Similar to the initialization strategy in ICON, the ocean grid has been spun up for several years with ERA5 forcing until 20 January 2020, the common starting point of the nextGEMS simulations, before being coupled to IFS. 

Early preliminary analyses of Cycle 2 compare rather well to the ERA5 reanalysis, OSI-SAF, and observational data, such as a long 40-day forecast of sea ice concentration evolution (Figure 2). We are looking very much forward to the next Hackathon in Vienna later this month where in-depth analyses from all project partners might reveal new surprises – both in terms of weaknesses but also in terms of novel strengths that only this new generation of climate models can provide.

Figure 2: 40-day forecast of sea ice concentration in the coupled IFS-FESOM Cycle 2 simulation with 2.5km IFS and FESOM2-NG5 (left), compared to data from OSI SAF (right). (figure kindly provided by Lorenzo Zampieri, NCAR, https://ncar.ucar.edu)

—————————-
* Scientist at ECMWF

** Model development at MPI-M
*** Special thanks to all participants of the 1st nextGEMS hackathon


Further Articles

calendar date image

26. November 2024

Join us for our final hackathon!

Hazard Hackathon opening session

15. October 2024

nextGEMS‘ 5th Hazard Hackathon gathers over 80 participants in Wageningen for an exciting challenge

easy gems logo

30. September 2024

Introducing Easy Gems: your gateway to Next-Generation climate modelling

Privacy

Here you can make settings regarding data protection.