As we will recall, to become climate neutral between now and 2050, the European Union programmed two very specific and quite ambitious programs: “Green Deal” and “Digital Strategy”. To be carried out successfully, climate scientists and computer scientists have launched the “Destination Earth” initiative, which will begin in the middle of this year, 2021, and is expected to last up to ten years. During all this time, scientists will create a high-precision digital model of the Earth; a digital twin of the planet, in order to map climate development and extreme events as accurately as possible in space and time.
This ‘digital twin’ of Earth will therefore be used to predict future events caused by climate change to help world leaders better prepare than they are doing now.
The terrestrial digital twin is being developed by European scientists and ETH Zurich University (Switzerland). According to experts, the machine learning algorithm will develop and test simulations until 2050. The virtual model will also predict all processes “in the most realistic way possible”, including the influence of humans on water, food and water management. energy and processes in the Earth’s physical system.
“If someone is planning a two-meter-high dike in the Netherlands, for example, I can analyze the data from my digital twin and see if the dike is very likely to still protect against extreme events predicted in 2050,” explains Peter. Bauer, deputy director of Research at the European Center for Medium-Range Weather Forecasts (ECMWF) and co-initiator of Destination Earth, to the journal Nature Computational Science that collects the study.
How exactly will it work?
Although still under development, this virtual Earth will be designed with real-time observations such as satellite images and weather forecasts. The digital twin will allow high-resolution simulations representing the complex processes of the entire Earth system. But to achieve this, the codes of the simulation programs must be adapted to the new technologies. Thus, thanks to artificial intelligence, which makes it possible to accelerate simulations and filter important information in large amounts of data, this large-scale virtual twin will be possible that, according to experts, would require a system with around 20,000 GPUs, which would consume approximately 20MW of power.
“The simulation-observation fusion is performed in space and in a time window in which the model ensures that the optimal and physically consistent evolution occurs, accepting that observations do not measure all state variables everywhere all the time. This optimization has a huge computer footprint, since the state of the Earth system comprises billions of degrees of freedom and deals with a non-linear and chaotic system “, clarify the authors.
This project is promoted by the ECMWF, the European Space Agency (ESA) and the European Organization for the Exploitation of Meteorological Satellites (EUMETSAT). Along with other scientists, Bauer is pushing forward the climate science and meteorological aspects of Earth’s digital twin, but they are also relying on the technical know-how of computer scientists at ETH Zurich and the Swiss National Supercomputing Center (CSCS). A true partnership between earth science and computer science.