Humans are now living in the warmest climate in modern history. Wildfires, droughts, hurricanes, flooding, extreme heat, and coastal erosion are more intense and more frequent with each passing year.
What do these climate changes mean for national security?
In January 2021, the Biden administration issued the Executive Order on Tackling the Climate Crisis at Home and Abroad, which stated that the climate crisis is “at the forefront of this nation’s foreign policy and national security planning” and declared that “the United States will … move quickly to build resilience, both at home and abroad, against the impacts of climate change that are already manifest and will continue to intensify according to current trajectories.”
Soon after, the National Intelligence Council issued a National Intelligence Estimate on climate change, which concluded that “climate change will increasingly exacerbate risks to U.S. national security interests as the physical impacts increase and geopolitical tensions mount about how to respond to the challenge.” Among the report’s main concerns were global competition over clean-energy materials, societal strife in regions with chronic drought, strains on military readiness, and potential conflict over fossil fuel resources in the Arctic.
However, identifying concerns is different than addressing them. The overwhelming scope and complexity of climate change make it difficult to tackle or even know where to start. But climate modeling—using computer software to create virtual representations of past, present, or future climate scenarios anywhere on Earth—can help.
According to the National Oceanic and Atmospheric Administration’s Climate.gov website, climate models “use mathematical equations to characterize how energy and matter interact in different parts of the ocean, atmosphere, and land.” When developing climate models, researchers must identify and quantify Earth system processes and represent them with mathematical equations. Then, variables are set to represent conditions and changes. Supercomputers are used to solve the equations and simulate various scenarios.
“The evolution of climate models has been one of increasing complexity run on faster and larger computers,” explains the National Intelligence Estimate. “The first climate models examined how the Earth’s energy balance and atmosphere might vary over time, and only considered atmospheric physics and rudimentary representations of the oceans and land. In time, scientists added more detail, such as ocean and land chemistry and biology.”
Today, the report continues, “climate models operate by solving a very large set of sophisticated equations for three-dimensional grids in the atmosphere and oceans.”
Climate modeling is all about making choices, explains Luke Van Roekel, of the Fluid Dynamics and Solid Mechanics group at Los Alamos National Laboratory. “You have to say there are some things I’m going to explicitly resolve and some things I’m not going to pay attention to because they may make the model unnecessarily complex without adding value,” he says. “Imagine the Gulf Stream, a big process that is hundreds of kilometers long. If you zoom in, there are little waves and eddies that break off from the waves, and if you zoom in more, there are even smaller eddies. So we have to understand what scale of these features is most important to model and what might have a smaller impact on our understanding of these systems.”
The Energy Exascale Earth System Model
As of 2021, more than 30 major climate-modeling centers around the world are able to make projections 10, 50, 100, or even 1,000 years into the future using sophisticated models that run on supercomputers.
Since 2013, Los Alamos National Laboratory, along with about a dozen universities and seven other national labs, has been working on the Energy Exascale Earth System Model (E3SM). E3SM, which is funded by the Department of Energy, aims to be the fastest, most accurate, and highest resolution Earth system model in the world.
“Imagine being able to explore and quantify changes in the physical climate system at unprecedented resolution and for any imaginable future climate conditions,” explains Stephen Price, a member of the Lab’s Fluid Dynamics and Solid Mechanics group and the institutional lead for the Lab’s E3SM efforts. “This is our goal for E3SM.”
E3SM is essentially a suite of models that can operate independently or together. Individual models are used to simulate various natural and manufactured aspects of the Earth’s land, oceans, and atmosphere, such as ice sheets, rivers, storm surge, soil saturation, vegetation, and even crop yields.
Some individual models account for the interactions of several Earth systems, such as E3SM’s sea ice model, which can simulate sea ice melt in the north and south poles by taking into consideration the physics of how sea ice freezes and melts, how it moves across the ocean’s surface, and how it is influenced by ocean currents, wind, and other systems.
When coupled together by appropriate software, these individual models can simulate fluxes of heat, moisture, energy, and features all the way down to scales as small as the absorption of greenhouse gases by algae. “The coupling of the individual models in E3SM is becoming increasingly sophisticated,” says Van Roekel, who co-leads the Lab’s E3SM Water Cycle efforts, “such that we could look at hurricanes as they blow water onto the coast and quantify regions susceptible to sea water inundation and how this will change in different climate conditions. That would be pretty sophisticated for a climate model.”
Early results for one model used in E3SM, the MPAS-Albany Land Ice (MALI) model, have already proven extremely useful. In August 2021, Los Alamos, along with 38 other international collaborators, studied nearly 900 simulations to understand how melting ice sheets (large masses of glacial ice) will impact global-mean sea-level rise by 2100. They found that under unabated emissions, sea-level rise could be as high as 30 inches. Alternatively, by limiting global warming to 1.5 degrees Celsius, sea-level rise by 2100 could be limited to five inches—thus confirming that human choices will indeed play a part in determining just how detrimental climate change will be.
A new era
In addition to global sea-level rise caused by melting ice sheets, scientists are using E3SM to focus on the more localized effects of melting ice sheets.
Price explains that recently, Lab scientists noticed an alarming result in their E3SM simulations of the Southern (Antarctic) Ocean and its possible interactions with the Antarctic ice sheet. In one simulation, cold freshwater circulating along the Antarctic coast in a counter-clockwise direction was arriving at the Filchner-Ronne ice shelf—the Earth’s largest floating ice shelf by volume. This was of particular interest to researchers because the simulation showed that this influx of water contributed to a rapid increase, about 10 times the normal rate, of this ice shelf melting over only a few years.
The implications for ice loss and sea-level rise are significant as Antarctica’s ice shelves are critical for limiting the movement of ice off of the continent and into the oceans.
This was just a simulation. But it had researchers worried and wondering if such an event could occur in the real world. If so, how?
Luckily, complex models such as E3SM are the perfect tool for exploring questions like “what if?” and “how?” Through a series of E3SM simulations, the team discovered that, due to ocean currents, changes in ice shelf melting in one region of Antarctica can have catastrophic implications (including ice shelf melting, which results in sea-level rise) in other regions of Antarctica.
This revelation would have been impossible to discern from research expeditions because of the size and complexity of the physical system involved. Researchers are now aware of this interconnectedness—and its implications—because of E3SM.
Computing power
The supercomputers currently used to run E3SM models operate at between 10 and 100 petaflops. A single petaflop can make one quadrillion computational operations every second. The Department of Energy owns two of the top three petaflop computers, but to unlock the full potential of E3SM will require something faster.
Resolution is also limited by current supercomputers. The resolution of most E3SM simulations can be drilled down to 60-square-mile cubes. So, scientists may be able to answer a question like, “How high will the water rise along the Eastern Coast of the United States in 50 years?” But that resolution doesn’t allow for more specific answers to questions such as, “In half a century, will a specific military base be under water?”
E3SM is being designed to run on exascale computers, which are in development. One exaflop (as processing power is called for exascale computers) is 1,000 times faster than a petaflop, which means that the resolution of most simulations will be improved to 6-square-mile cubes, and select regions may be even finer resolution. For climate modelers, that difference in resolution could mean differentiating between an entire city being flooded versus specific sections of a city being destroyed by floodwaters.
However, the transition to exascale computing will not be seamless. “Exascale computing presents an acute challenge in that the new computer architectures are quite different from those before,” explains scientist Elizabeth Hunke of the Lab’s Fluid Dynamics and Solid Mechanics group. “Using exascale computers will require recoding many of our older models, but these new computing systems will open grand new possibilities for climate science.”
Collecting data
How can a model—a vast assemblage of mathematical calculations—simulate often incomprehensible acts of nature?
All modeling begins with data, and that data is provided by scientists such as Joel Rowland, of the Lab’s Earth Systems Observations group. “I try to understand how we go about getting enough real-life observations to incorporate into the climate models, which then make up the mathematical architecture of the model,” he explains.
In many cases, these real-life observations come in the form of historical data. In the United States, especially in the lower 48 states, extensive climate records are available through government organizations such as the National Oceanic and Atmospheric Administration, the United States Geological Survey, and the National Aeronautics and Space Administration. Then, “we do what’s called a spinup,” Rowland says. “We’ll run the model, but we set it for conditions that existed 160 years ago. If the simulation reflects known historical observations—for example the precipitation over a specific region matches what we know to have occurred in the past—then we can make judgments on the accuracy of the model.”
Unfortunately, other countries do not have such meticulous climate records, which is problematic for building a global model such as E3SM.
The climate of the Arctic, for example, is not well documented—especially when it comes to the region’s floodplains and rivers, which are Rowland’s focus area. Without historical data to plug into models, Rowland must collect current climate data himself. Recently, he’s traveled to the Yukon and the Koyukuk rivers in western Alaska. There, he uses river gauges, takes water samples, and measures soil erosion where land interfaces with running water. He takes the data back to Los Alamos where he analyzes it and eventually enters the information into a model that feeds E3SM.
Rowland hopes to learn more about how water impacts permafrost—permanently frozen layers below the Earth’s surface. As the ice in the ground melts, mounds form on the Earth’s surface. These mounds alter the flow of water on the surface, which further accelerates ice melt in the ground and permafrost thawing. This thawing releases gases, including carbon dioxide and methane—both important drivers for global warming.
“We had to ask ourselves if we needed to enter this mounding interaction into the model,” Rowland says. “Is it important? Well, in this case, that seemingly small dynamic becomes very important.” Globally, permafrost holds nearly twice as much carbon as already exists in the atmosphere, so any interaction that speeds up the release of this carbon is vital to understanding the climate future.
National security implications
Before coming to Los Alamos in 2019 to work as part of the Information Systems and Modeling group, Travis Pitts worked in the U.S. Intelligence Community. “I was focused on the Middle East for a long time, then I turned my attention to security issues in the Arctic,” Pitts says. “I started to realize really quickly that, in both of those areas, if I wanted to answer questions about the future of national security I had to involve scientists.”
That’s when Pitts decided to come to Los Alamos and join the Lab’s Climate Impacts on National Security (CINS) team, originally led by Cathy Wilson of the Earth Systems Observations group and now co-led by Pitts and Price. In this role, Pitts has become heavily involved in figuring out how to apply E3SM to national security problems.
“We’ve had a vision for a long time about what we want from E3SM,” Pitts says. “We want it to focus in on future climate impacts in strategically important areas of the world and to be able to do that frequently and with ease.” Doing that could mean simulating the impacts of climate change on important national security infrastructure domestically and internationally.
“It really comes down to looking more closely at strategically important locations and how climate impacts will affect everything from infrastructure to food yields,” Pitts says.
Rising sea levels, for example, will likely wreak havoc on U.S. military bases across the world. To understand exactly how bad the danger could be, E3SM, using a model called New Science for Multisectors Adaptation, can simulate rising sea levels in a specific area combined with any number of other factors such as hurricane storm surge and coastal erosion. The simulations can help researchers and policymakers understand if military facilities will become more vulnerable in the future.
Climate change might also lead to conflicts that affect national and global security. In 2010, for example, food shortages caused by drought contributed to uprisings in the Middle East that killed more than 60,000 people.
Looking forward, in the Arctic, melting ice will alter how commercial and military vessels travel near the North Pole. Receding ice may also make currently inaccessible resources, such as oil, accessible. Who does that oil belong to? The answer is complicated because the Arctic falls under the jurisdiction of eight coastal countries. Will competition ensue? What will that competition do to the geopolitical situation?
“Internal instability in one area of the world can lead to instability elsewhere,” Pitts says. “That’s obviously important when we think of national security.”
To address the interface between humans and the climate, Lab researchers on the CINS team are working with Earth system models to look at food security. For example, by taking historical crop yield data from staple foods (corn, soybeans, rice, and wheat) and cross-referencing those with historical climate data, they can understand how past shifts in precipitation or temperature affect the quantity of crops produced. Once they know that, E3SM can be used to predict future yields based on variations in the climate.
“In the United States, we’ll be able to look at changes in crop yields over agricultural hotbeds like those in the Midwest and California, which we know will both experience a high degree of climate change,” says Kurt Solander, a researcher with the Computational Earth Sciences group. “Globally, we know that areas like Asia and Africa will also be very susceptible to higher temperatures, and both of those regions grow a lot of food for the world. In a global supply chain, it doesn’t matter if as a country you consume a lot of soy or not, because the ripples from the losses in one region of the Earth will touch far beyond.”
Going classified
As an “open source” model, E3SM is available for use by any interested climate researcher. But E3SM and its component models have also been transferred to the Lab’s classified computing networks. It might seem odd, at first, to have models that simulate climate change running alongside models that are most often used to simulate nuclear explosions. But, just like the Lab’s nuclear weapons work, the Lab’s climate work has become essential for national security, and moving E3SM models to classified supercomputers allows researchers to conduct simulations and analysis in specific regions, under specific scenarios, that have national security implications.
“With the model being on the classified system now, we’ll be able to answer a lot more questions for a wider variety of clients,” Pitts says. “We’ve already had a lot of interest in E3SM recently.”
Hunke agrees. “E3SM allows more detailed calculations to be focused in regions of interest, computed concurrently with the entire, global Earth system,” she says. “Now that climate change implications for national security are becoming obvious, E3SM has a clear role in providing the Department of Energy and other agencies the climate prediction capabilities needed for decisive action.” ★