51����

This is Fix the Planet, the weekly climate change newsletter that reminds you there are reasons for hope in science and technology around the world. To receive it in your inbox, sign up here.

This week saw the launch of a €140 million European Union project to build “digital twins” of Earth: detailed, interactive simulacra of the planet that will allow everyone from scientists to energy companies to explore the impacts of climate change and how we adapt to them.

Destination Earth is a hugely ambitious European Commission endeavour that has been years in the making. Researchers hope it could usher in a new level of granularity and accuracy to our projections of future climate change at a local level, so governments can better prepare for extreme weather events like the floods that hit Germany and neighbouring countries last year��.

But will it really shift the dial on our ability to build more effective flood defences and develop more efficient renewable energy schemes? I caught up with one of the key architects of the project,  at the European Centre for Medium-Range Weather Forecasts (ECMWF), to find out.

What is it?

The digital twins are being built by what is effectively a start-up operation based in Bonn, Germany, with an eventual staff of about 50. The team will bring together applied and computational science to create the interactive simulations, which will initially be focused on replicating how extreme weather changes as global average temperatures rise to 1.5°C above pre-industrial levels in the coming decades.

The twins are also intended to help countries create dedicated zones for renewable energy installations, says Bauer: they might give an energy firm new insight into how wind speeds are expected to change across the North Sea in coming years, for example. Bauer likes to call the digital twins an “interactive space” to explore “this metaverse of climate data”. By metaverse, he doesn’t mean Meta’s vision of avatars and virtual reality, but “a good illustration of what we mean when we say interactivity: I can play with data in the four dimensions of space and time; I can bolt on applications.”

Haven’t we been here before?

The scheme has emerged from the ruins of an even more ambitious climate change supercomputing project called , which the EU eventually cancelled. But this time it’s happening. The budget was approved last December, meaning funding is guaranteed to the end of 2024, yet the whole enterprise should run for up to 10 years.

How is this different from typical climate models?

The level of detail it will offer at a local level – the resolution – should be greater than most climate change modelling today. “Developing a high-resolution digital twin of planet Earth is vital if we are to meet the challenges of climate change,” says Tim Palmer at the University of Oxford, who had been a proponent of ExtremeEarth. “We know that the Earth is warming, but our knowledge of regional impacts is poor. The core of Destination Earth will be a new ultrahigh-resolution description of the climate system.”

The twins won’t be able to predict the next flash flood or storm months ahead, because of the chaotic nature of weather, says Erich Fischer at ETH Zurich in Switzerland. But they should help us plan better. “They can inform us of what potentially unseen extreme weather events we need to prepare for today and in the next decades, and thereby inform our decisions and planning, for instance, of future cities or energy systems,” he says. Destination Earth should also make greater use of machine learning than most existing models, says Bauer.

What’s happening now?

Bauer and his colleagues are very much in start-up mode. While ECMWF is building many elements of the digital twins itself, it is also outsourcing a lot. Last week, it tendered out a contract for the climate change adaptation aspect, so it can simulate changes in coming decades. This week, it has tendered for work on the extreme weather part of the twins, dealing with fires, pollution, flooding and so on. The team is also hard at work obtaining supercomputing resources – one system has been switched on in Finland, one’s being built in Italy, one is being negotiated in S pain – to run the simulation on. They are all “pre-exascale” in processing power, meaning they are ranked at below a million teraflops. The Finnish one is the world’s most powerful publicly-known supercomputer, capable of 552,000 teraflops. For comparison, the UK’s most powerful supercomputer was switched on last November, and is rated at 19,539 teraflops.

How much of an impact will these digital twins really make?

Fischer says it’s an “an extremely ambitious and highly innovative project”. Palmer hopes it will “kick-start a revolution in Earth-system modelling”. He thinks the arrival of exascale high-performance computing will make that shift possible. “For the sake of society around the world, we need this sooner than later,” he says.

Ted Shepherd at the University of Reading, UK, says the new system should offer more realistic levels of detail for extreme weather events than current climate models. But he’s concerned that expectations for how accurately the digital twins can project climate change’s future impact might be unrealistically high. Some of the most persistent uncertainties in climate models, such as low clouds, act at scales too small to be picked up by Destination Earth, but have worldwide ramifications for global warming and how it plays out locally. “Thus, while the weather systems themselves will be more realistically simulated than at present, it is quite likely that the effect of climate change on those systems will remain highly uncertain,” he says.

The EU isn’t the only entity trying to build a climate change digital twin of Earth, either. US tech company NVIDIA  that it’s building an “Earth-2” digital twin, to simulate climate change down to a resolution of a metre, rather than the 10 to 100-kilometre resolution usually used in models. Bauer says his team is talking to NVIDIA about how they can work together. One key thing about Destination Earth is that it’s publicly funded, so the data should be free and open to use.

What are the next steps?

The digital twins will be active by June 2024 at the very latest, so don’t be surprised if you start seeing 51���� stories a few months later about new scientific papers using the system to project future storm surges in Europe, or how future temperatures might affect solar panels’ output. A second phase of the project post-2024 should see new layers added to the digital twins, around oceans, biodiversity and geophysics (earthquakes, tsunamis and more). And what’s the ultimate goal? “Bigger and better,” says Bauer. “Bigger in the sense that more and more applications and people can use it. Better in the sense that we should be able to run more accurate models with the same computing power. We will learn a lot more.”

More fixes

• Wind and solar power generated a tenth of the world’s electricity last year, a record level. In total, more than 50 countries have now passed the 10 per cent milestone, according to a report by the think tank, Ember.

• Talking of solar, the energy source is having a moment in the UK despite an absence of subsidies in recent years. More than 30 large solar projects with a combined capacity of more than 3 gigawatts, enough to power hundreds of thousands of homes, have planning permission and are due to be built soon, . One to watch is Cleve Hill solar farm in Kent, where construction is due to start in the next few months.

• A date for your diary: Next Monday sees the publication of the Intergovernmental Panel on Climate Change’s report on how we mitigate global warming. Things to watch out for: a close look at carbon dioxide removals, the social aspect of how we cut emissions and a focus on near-term action rather than what we do in the second half of the century.

If you enjoy Fix the Planet, please do recommend it to your friends, family and colleagues –  they can sign up here. And if you haven’t already, to access all our science journalism and help keep this newsletter free.