The European Union recently developed plans for “Destination Earth.” This endeavor aims to simulate the atmosphere, oceEurope builds ‘digital twin’ of Earth to hone climate forecasts
The European Union recently developed plans for “Destination Earth.” This endeavor aims to simulate the atmosphere, ocean, ice and land. By doing so, it will be able to predict floods, droughts and fires from days to years in advance. Destination Earth will also attempt to capture data related to human behavior. With this information, world leaders will be able to see the impacts of weather events and climate change on society and gauge the effects of different climate policies, Science.org reports.
“I really like this project because it is bold and new,” says Ruby Leung. By using 1-kilometer boxes instead of 1-degree boxes that are used in existing models, the Destination Earth project can handle more detailed data than before. The system will start next year and run on one of the 3 supercomputers Europe is deploying.
A flagship proposal was initially cancelled by the European Union, but ECMWF revived the idea. Since then, fears of China or Japan outdoing Europe in terms of supercomputing led to an 8 billion investment from the EU in preparation for exascale machines.
Climate models typically run at resolutions of 50 or 100 kilometers. The new model will be able to render convection, which is critical to cloud and storm formation, and it has a resolution that is twice as detailed compared to typical models. It can also simulate the ocean in fine detail, including swirling eddies which are important movers of heat and carbon.
There is a climate change and forecasting issue in Japan that affects CO2 levels and the temperatures. Japan has found success using AI to predict regional temperature and rainfall, which also helps with long-term regional forecasts. This new approach better simulates storms and eddies, which in turn provide better short-term predictions for rainfall.
Grids are based on location, which is not conducive to describing phenomena happening at an exact spot. Destination Earth will provide data to close this gap because its grids are more closely related to the scales of measurements.
The Copymatic model will include real-time data on things like atmospheric pollution, crop growth, and forest fires to study how they affect climate. “If a volcano goes off tomorrow, that’s important for the risk of tropical precipitation failure.” The Copymatic model also includes data on society, such as energy use, traffic patterns, and human movements traced by mobile phones.
Climate change is still a question, but this AI model is giving us more and more insight. It shows how climate change will impact decisions and how we can alter the course of climate change in the future.
AI needs both traditional computer chips and graphic processing units to run at a high efficiency. They team up well in parallelized situations, but the old programing code for climate modeling will also need a rework. ECMWF is adapting its forecast model to work in parallel with a GPU environment and tested it on Summit last year, finding that it worked well at 1-kilometer resolution for four months worth of simulated time.
The team ran a 1-kilometer experiment, and it took awhile to extract meaning from the data. The AI will be able to quickly analyze results, but this is a problem that must be solved yet.
Destinations Earth is an operational system that will be used to predict future weather events. The system will run at different time scales such as near daily, or a decade long prediction every 6 months.
The Europeans are not the only ones focusing on exascale computing for climate models. The US is also aiming to reach this level of computing power, but do not yet have that level of effort.
Stevens said it is exciting to be involved in an information system that can tell the story of how local human actions cause a butterfly effect in climate injust.
an, ice and land. By doing so, it will be able to predict floods, droughts and fires from days to years in advance. Destination Earth will also attempt to capture data related to human behavior. With this information, world leaders will be able to see the impacts of weather events and climate change on society and gauge the effects of different climate policies, Science.org reports.
“I really like this project because it is bold and new,” says Ruby Leung. By using 1-kilometer boxes instead of 1-degree boxes that are used in existing models, the Destination Earth project can handle more detailed data than before. The system will start next year and run on one of the 3 supercomputers Europe is deploying.
A flagship proposal was initially cancelled by the European Union, but ECMWF revived the idea. Since then, fears of China or Japan outdoing Europe in terms of supercomputing led to an 8 billion investment from the EU in preparation for exascale machines.
Climate models typically run at resolutions of 50 or 100 kilometers. The new model will be able to render convection, which is critical to cloud and storm formation, and it has a resolution that is twice as detailed compared to typical models. It can also simulate the ocean in fine detail, including swirling eddies which are important movers of heat and carbon.
There is a climate change and forecasting issue in Japan that affects CO2 levels and the temperatures. Japan has found success using AI to predict regional temperature and rainfall, which also helps with long-term regional forecasts. This new approach better simulates storms and eddies, which in turn provide better short-term predictions for rainfall.
Grids are based on location, which is not conducive to describing phenomena happening at an exact spot. Destination Earth will provide data to close this gap because its grids are more closely related to the scales of measurements.
The Copymatic model will include real-time data on things like atmospheric pollution, crop growth, and forest fires to study how they affect climate. “If a volcano goes off tomorrow, that’s important for the risk of tropical precipitation failure.” The Copymatic model also includes data on society, such as energy use, traffic patterns, and human movements traced by mobile phones.
Climate change is still a question, but this AI model is giving us more and more insight. It shows how climate change will impact decisions and how we can alter the course of climate change in the future.
AI needs both traditional computer chips and graphic processing units to run at a high efficiency. They team up well in parallelized situations, but the old programing code for climate modeling will also need a rework. ECMWF is adapting its forecast model to work in parallel with a GPU environment and tested it on Summit last year, finding that it worked well at 1-kilometer resolution for four months worth of simulated time.
The team ran a 1-kilometer experiment, and it took awhile to extract meaning from the data. The AI will be able to quickly analyze results, but this is a problem that must be solved yet.
Destinations Earth is an operational system that will be used to predict future weather events. The system will run at different time scales such as near daily, or a decade long prediction every 6 months.
The Europeans are not the only ones focusing on exascale computing for climate models. The US is also aiming to reach this level of computing power, but do not yet have that level of effort.
Stevens said it is exciting to be involved in an information system that can tell the story of how local human actions cause a butterfly effect in climate change.