New high-resolution models merge weather and climate

August 6, 2021 – Torrential rains, hailstorms and flooding in the Alpine region and northwestern Europe: The last few weeks have highlighted the impacts of severe thunderstorms. But how exactly are extreme weather events related to global warming? This is one of the central questions for researchers studying and modeling the interaction between weather and climate.

By representing the underlying fundamental physical processes, models are a very powerful tool for understanding these interactions. But current models and the required IT infrastructure have hit a wall, limiting the extent to which researchers can draw conclusions about how, for example, climate change affects extreme weather conditions. To overcome this problem, ETH Zurich has teamed up with partners to launch the EXCLAIM research initiative. This project aims to dramatically increase the spatial resolution of the models, thereby improving their accuracy in simulating weather on a global scale in a hotter future world.

Transparent weather simulations in climate models

“Thanks to their high resolution, the new global models will simulate key processes such as storms and weather systems in much more detail than before, allowing us to study the interaction of climate change and weather events from much more precisely ”, explains Nicolas Gruber, EXCLAIM principal IP and professor of environmental physics.

EXCLAIM is interdisciplinary: with climatologists from the Center for Modeling Climate Systems at ETH (C2SM), computer scientists from ETH, the Swiss National Center for Intensive Computing (CSCS), the Swiss Data Science Center (SDSC), Swiss Federal Laboratories for Materials Science and Technology (Empa) and MeteoSwiss, the Federal Office of Meteorology and Climatology, are all involved in the project. This collaboration will not only improve climate modeling, but also make the meteorological forecasts provided by MeteoSwiss more reliable. International partners in the project include the German National Meteorological Service, the Deutscher Wetterdienst (DWD) and the Max Planck Institute for Meteorology (MPI-M), who jointly developed the ICON (Icosahedral Nonhydrostatic) model – the basis of EXCLAIM – as well as the European Center for Medium-Range Weather Forecasts (ECMWF), of which Switzerland is a full member.

With EXCLAIM, researchers aim to radically increase the spatial resolution of weather and climate models. To simulate the global weather and climate with all of its regional details, such models place a virtual three-dimensional grid on the Earth. The researchers then use the laws of physics to calculate the respective climatic conditions of each cell in their models. Current global climate models typically have grid cells 50 to 100 kilometers wide. In the long term, the EXCLAIM researchers aim to increase the resolution to just one kilometer.

In the past, given the limited computing power of modern supercomputers, only regional weather could be simulated with such a fine grid – and for relatively short periods of time at most. With the new models, researchers now hope to achieve this fine resolution around the world, allowing them to simulate weather conditions from a global climate perspective and with much more precise focus. It’s like giving global climate models an extra zoom feature for small-scale events. “Additionally, the new models will pave the way for ‘forecasting’ the weather in the future climate, providing answers on how the extreme weather events like torrential downpours that we experienced this summer might look like. future, ”says Christof Appenzeller, MeteoSwiss Head of Analysis and Forecasting.

A powerful infrastructure for climate simulations

A customized IT infrastructure is essential to get the most out of new models. Weather and climate models are among the most complex and data-intensive computational problems, which is why EXCLAIM models are developed in parallel with supercomputer hardware and software. “The computing and data infrastructure is tailored to the exact requirements of weather and climate models,” explains Thomas Schulthess, director of the Swiss National Center for High Performance Computing (CSCS) in Lugano. For example, the new “Alps” high-performance computing system is configured to enable high-resolution climate models to correctly resolve convective systems, such as thunderstorms.

To effectively simulate global weather and climate over decades with a grid width of just a few kilometers, the model will need to run about 100 times faster than is currently possible. The first option to achieve this goal is to deploy faster and more powerful computers. The transition from the current CSCS supercomputer to the “Alps” system will be decisive in this regard.

One of the challenges is the end of “Moore’s Law,” which states that processor performance doubles about every 20 months. “Since processors haven’t increased their serial performance for about 15 years, the only way to improve supercomputer performance is to improve their parallel processing architecture,” says Schulthess. “In addition, it is worthwhile to set up the architecture of the supercomputer specifically to enable it to solve classes of research problems in an optimal way. The key to delivering the required computing power here lies in a hybrid computing architecture in which conventional CPUs (central processing units), responsible for performing calculations and sharing data between memory and components, are deployed in conjunction. with GPUs (graphics processing units).

The second option concerns the software, namely the optimization of the model code so that it takes full advantage of the hybrid IT architecture. EXCLAIM adopts a revolutionary approach by dividing the source code into two parts: a first part which represents the interface with the developers and users of the model; and a part of the underlying software infrastructure in which the core algorithms of the model are implemented with a high degree of efficiency for the respective hardware. The CSCS, MeteoSwiss and C2SM have already used this approach in the current MeteoSwiss weather model with great success. This approach is now applied to the ICON meteorological and climate model. “We were able to speed up the MeteoSwiss weather model by a factor of ten, thereby improving the reliability of MeteoSwiss forecasts,” explains Schulthess.

Manage the flow of data

Computing speed alone is not the deciding factor. Increasing the resolution of models also leads to a data explosion. In addition, meteorological and climate research requires and produces a wide variety of data. To ensure efficient throughput, it is also crucial that computers are able to both access the data and write the results to storage media as quickly as possible. IT processes must be organized accordingly, while memory bandwidth is maximized and costly data transfers avoided. “For new weather and climate models to produce meaningful results, we need to optimize the entire infrastructure. To this end, we are making use of the expertise gained over many years of collaboration with MeteoSwiss and the ETH Domain, ”says Schulthess.

New efficient meteorological model leads to more precise estimates of greenhouse gas emissions

As part of ETH’s EXCLAIM project, in which Empa is involved as an external partner, a highly efficient weather and climate model is being developed which makes optimal use of the capabilities of the latest generation of computers. high performance and innovates in programming to achieve this goal. The starting point for this development is the ICON model, which was mainly developed by the Deutscher Wetterdienst (German Meteorological Service) and the Max Planck Institute for Meteorology, and which will be used in the future by MeteoSwiss for its weather forecasts.

Atmospheric models, however, can be used not only for weather and climate forecasting, but also to simulate air quality or the dispersion of plumes of polluting emissions, for example from volcanic eruptions or incidents. nuclear.

Empa uses such models to estimate greenhouse gas emissions from individual sources or entire countries by comparing the simulated concentrations with measurements, for example Empa measurements at the Jungfraujoch. Their estimates of Swiss emissions of methane and nitrous oxide are published in the National Greenhouse Gas Inventory, which is submitted annually by Switzerland to the UNFCCC under the Paris Climate Agreement. Empa thus offers an independent and valuable review of the inventory published each year.

In order to perform simulations at a previously inaccessible resolution of around a few kilometers, Empa will in future rely on the powerful model developed in EXCLAIM. This will require simulating up to several hundred different realizations of greenhouse gas concentrations – a complex process that in the past was only possible with coarse spatial resolution. It will also make it possible to use measurements from future satellites, which measure the global distribution of CO2 and methane, to estimate emissions. (Pof. Dominik Brunner, Amanda Caracas, Empa)

Source: Florian Meyer, ETH Zurich