In a First, Global Climate Models Will Be Made Available Via the Cloud

At present, climate modeling is available to just a privileged few. While computing has become more sophisticated in recent decades, financial and practical barriers have meant that capturing the complexity of how climate change could impact the Earth's systems is out of reach for most researchers.

However, one of the world’s biggest companies is making a bid to democratize the process. Amazon.com Inc.’s Sustainability Data Initiative is launching a first-of-its-kind partnership with the nonprofit SilverLining and the National Center for Atmospheric Research on Tuesday that will make climate model simulations available on the cloud, potentially expanding the club of researchers with access to this data to include those in small coastal or island nations already bearing the brunt of climate change.

“Even in the United States and in Western countries, computing capacity is one of the biggest barriers to why we're not understanding the climate system faster,” SilverLining Executive Director Kelly Wanser said in an interview. “And it’s also a power disparity problem, because only certain countries have access to those kinds of capabilities.”

Dr. Ana Pinheiro Privette, lead of the Amazon Sustainability Data Initiative, added that putting the models on the cloud “eliminates the cost of acquisition by making the data openly available” and “minimizes the need for storage of multiple copies of the data, which decreases the carbon footprint of climate research.”

The e-commerce giant’s initiative will allow NCAR to use Amazon Web Services’ technologies and technical support to run a collection of 30 model simulations from 2035 to 2070. The process will use two NCAR models -- the Community Earth System Model Version 2 and the Whole Atmosphere Community Climate Model -- operating under a median scenario for warming. 

The United Kingdom’s Meteorological Office will then replicate the model simulations, and related simulations will utilize NASA's Goddard Institute for Space Studies model. The datasets that result from these analyses will be available via open access to researchers globally.   

“Cloud hosting of climate-model simulations can accelerate the pace of science and enable far wider access for, and more meaningful collaboration with, researchers around the world,” Jean-François Lamarque, director of NCAR’s Climate and Global Dynamics Laboratory, said in a statement.

And indeed, that collaboration is built into this initial project. Research teams from four U.S. universities have worked on the project alongside those from the Bangladesh Centre for Advanced Studies, the University of Cape Town, the University of Nairobi, the University of the Philippines and others.

The resulting free, open access dataset will allow research teams internationally to skirt one of the major barriers to specialized climate modeling, even for those who have the computing capacity to make it happen: cost. Wanser said running the 30 simultaneous simulations would normally cost roughly $700,000, and take two months to run. 

The AWS program will cover all costs associated with hosting and sharing data from the cloud, and accessing and downloading it will be free. Grants will be available to users who choose to analyze or run additional models on AWS.

The planned simulations will be run at a resolution of 100 kilometers (i.e. each pixel of the resulting Earth model will be 100 square kilometers), Wanser said; a simulation of higher resolution (e.g. 10 km) would take up to 100 times as much computing capacity.  

Wanser said AWS, SilverLining and NCAR are angling to finalize the data within six months. Going forward, the three hope to make more of the world’s global climate models and datasets accessible via the cloud, according to the news release. 

“Traditionally researchers and climate scientists have not been able to equally bring their expertise, perspectives and concerns to the solution space,” Privette said. “Climate change is a global problem and a complex one, so having more diversity of thought and knowledge will help us build more robust solutions.”

This is reflected by the experience of Dr. Chris Lennard, senior researcher at the University of Cape Town’s Climate System Analysis Group, who said in a statement that his ability to study severe drought and other climate impacts happening locally is “limited by local computing resources and networks." 

"Open access to models and datasets, and cloud-based high-performance computing environments to work with them, allows us to engage equally as researchers,” he added, “and to design and execute studies to help people in our part of the world that were not conceivable in the past.”