How supercomputing contributes to the fight against climate change

We talked to Albert Soret, coordinator of the Earth System Services Group at the Barcelona Supercomputing Center (BSC-CNS), about the opportunities of supercomputing for climate science.

Manel Domingo

The computing revolution initiated in the 20th century enabled science to take a giant step forward. The processing of large amounts of data in a matter of seconds, the simulation of phenomena and systems that are impossible to study in a laboratory, and the new opportunities for collaboration between scientists all around the world have driven enormous advances in just a few decades.  

Well into the 21st century, supercomputing promises to give new impetus to this scientific development. And climate science is one of the fields that is already reaping the benefits. But what is supercomputing?  

Supercomputing makes it possible to generate digital twins of the Earth that mimic the processes of its climate system

Certain simulations — examples being the folding of a protein or the Earth's climate system — require extremely complex calculations, and therefore greater computing capacity. There reaches a point where creating bigger computers is no longer efficient, so it becomes necessary to connect many computers that work in parallel. In this way, each node calculates part of the model to be developed in coordination with the others. 

So explains Albert Soret, coordinator of the Earth System Services Group at the Barcelona Supercomputing Center (BSC-CNS), which now houses one of the most powerful supercomputers in the world, the MareNostrum 5.  

Recreating the Earth

This new addition will play a key role in climate research, which boasts one of the most organized scientific communities. The BSC-CNS — an institution with which Esade collaborates since 2023 — takes part in the Coupled Model Intercomparison Project (CMIP) organized by the World Climate Research Program, a global alliance which coordinates a multi-model perspective of climate change

The CMIP recreates the many facets of climate change in order to make predictions on different time scales. For this purpose, a digital twin of the Earth is generated, which imitates the processes and interactions between the major components of our climate system, including the atmosphere, the oceans, the land surface, the biosphere, and the cryosphere (the planet's large ice masses). 

We cannot predict the future, but we can define a range of potential scenarios

“The atmosphere is chaotic more than 5 or 10 days out, but other systems do not evolve so rapidly,” Soret notes. “If you simulate them correctly, you can understand the inertia of the climate system on different time scales.” Soret recognizes that “no one can predict the future, but we can define a range of potential scenarios, and on the strength of these, we can evaluate whether or not we are on the right track.” 

The results compiled by the CMIP are used to produce the reports made by the Intergovernmental Panel on Climate Change (IPCC), and based on these, governments  

Closer than the long term

The Earth System Services Group led by Soret is formed by an interdisciplinary team of 46 people, divided into 3 sub-groups.These three groups work together, one of them conducting research into air quality and another engaging in climate research, while a third team focuses on knowledge transfer. “The main added value is that we are all at the same table, and someone with a highly technical profile has to reach an understanding with a philosopher or an economist,” Soret explains. 

In addition to potential long-term scenarios, Soret's team also studies the present impacts of climate change to improve the way we adapt to these. “An important strand of our work is to understand what the climate will be like next summer, rather than the end of the century,” Soret observes. This is particularly relevant for sectors that are more sensitive to climatic variables, such as agriculture, water management, or renewable energies

Sectors such as agriculture, water management and renewables are very exposed to climate variables

For example, a cold winter leads to higher energy demand, but if it is less windy than usual, renewables won’t be able to cover demand. In the case of the wine sector, it is advisable to leave more leaves on the vines during a hot summer, but if it turns out to be wetter than expected, the leaves will facilitate the proliferation of fungi. “This situation can be extrapolated to critical products like rice or corn, where shortages can cause serious problems when they coincide with migrations and wars,” Soret warns. 

Information at one’s disposal

For all these reasons, an essential part of the team’s work centers around so-called climate services. Some users such as the aforementioned — in addition to public administration and citizens in general — have an increasing need for climate information that goes beyond the weather forecast for the next few days, without extending to long-term climate projections that are decades away. 

This information exists, but it is not easily accessible, since it consists of large volumes of data that have to be correctly interpreted. The Earth System Services Group audits, evaluates, and synthesizes the useful part of the information and makes it available to the end user.  

Another project, coordinated by Soret together with Marta Terrado, is a European initiative (ASPECT) involving 11 research centers. The project is focused on coordinating communities that investigate climate at different time scales (weeks, months or years ahead), whose processes and methods differ significantly. "These differences make it difficult to provide an integrated view over different time horizons," explains Soret. This is what is called seamless prediction, or the grouping of models that allows short-, medium- and long-term predictions to be made with the same tools. 

His team is also studying how the information compiled at the BSC-CNS can be integrated into the decision-making processes around clean energies. This involves projecting potential scenarios in order to predict energy production and demand, so that a robust system based on renewable sources can be generated. 

All written content is licensed under a Creative Commons Attribution 4.0 International license.