Science on Saturday lectures break down the CO<sub>2</sub> problem
Throughout the month of February, scientists from the Physical and Life Sciences (PLS) directorate virtually participated in three of the four 2022 Science on Saturday (SOS) lectures, presenting on the theme “Energy and the Environment.” The SOS lecture series is an annual collaboration between scientists at Lawrence Livermore National Laboratory (LLNL) and staff members from local school districts, aimed at educating middle and high school students about the science performed at the Laboratory and encouraging them to explore careers in STEM.
The 2022 SOS lectures brought awareness to the rapid climate changes Earth currently faces and the effects of global warming—such as rising sea levels, melting ice in places like Antarctica and Greenland, and increasing wildfires. Each presentation explored LLNL’s research efforts in mitigating climate change through atmospheric carbon dioxide (CO2) removal and improved climate model predictions to help researchers anticipate local impacts from climate change.
Direct CO2 capture
Atmospheric, Earth and Energy Division scientist Roger Aines and Materials Science Division scientist Sarah Baker kicked off the SOS series on February 5, and presented on the topic “Too much trash in the air, how can we clean it up?” This lecture set the tone for the series, comparing atmospheric CO2 to a heat trapping scarf or blanket that is warming up Earth and causing the climate to change. Sarah says, “Every year we emit 50 gigatons of CO2 into the atmosphere. For perspective, one gigaton is equal to two hundred million elephants—that’s a lot of CO2.”
LLNL scientists are actively working to develop machines capable of directly capturing and removing CO2 from the atmosphere so that it can be compressed to a liquid state and permanently stored deep underground. To do this efficiently, Roger and Sarah explained how at LLNL scientists are using 3D printing techniques to develop a unique packing structure, which will be used inside these CO2 capture machines. High-performance computer simulations help scientists come up with the best 3D designs and solutions to maximize CO2 capture. Compared to conventional structures, these printed packing structures work like the body’s lungs and have a very high internal surface area for more effective CO2 absorption.
Algae to the rescue
On February 12, scientists from the Nuclear and Chemical Sciences Division Xavier Mayali and Ty Samo presented the second installment of the 2022 SOS series, “Small algae, big world: the impact of microalgae on global carbon cycling and sustainable biofuels.” Xavier says, “Algae are more than just green stuff floating in the ocean. They are awe-inspiring creatures and key players in the global carbon cycle with the potential to fix climate change. In fact, aquatic microalgae are responsible for producing 50 percent of the oxygen on Earth and absorbing atmospheric CO2 through photosynthesis.” Recognizing the sustainable bioenergy capabilities of algae, LLNL researchers are growing their own small-scale algae in algal cultures under optimal laboratory conditions, with the goal of fine-tuning the algae’s carbon cycle to clean up the air and produce biofuel that can be recycled. In the lab, scientists feed algae CO2. The algae then absorb carbon through photosynthesis, and produce algal oil, which is harvested and burned as fuel. CO2 is then released back into the atmosphere, starting the cycle over again.
Livermore scientists are taking a unique approach to studying algae. Ty says, “Algae have their own microbiome, which is a bacterial community that attaches to and interacts with the algae. LLNL scientists are specifically looking at algae with microbiomes to understand algal growth patterns and compare it to algae that is grown alone and without a microbiome.” This research will help scientists learn how to manipulate the microbiome, enabling the algae to grow better and cycle carbon more efficiently.
Predicting the future
In the fourth and final SOS lecture, presented on February 26, Atmospheric, Earth, and Energy Division scientists Gemma Anderson, Mark Zelinka, and Aaron Donahue discussed the “Future in focus: predicting climate change through observations, modeling, and artificial intelligence.” The team started off by explaining the two uncertainties scientists face when it comes to predicting future global climate changes—the first being human activities that generate greenhouse gases and the second being how the climate responds to those greenhouse gases. To clear up these uncertainties, LLNL scientists are using Livermore super computers to run climate simulations—pinning down just how sensitive the climate is and demonstrating what will happen to Earth’s climate when greenhouse gases like CO2 are increased.
Global climate models
Scientists rely on global climate models (GCMs) to simulate and explore the possible climate states that may occur in the future. GCMs break the Earth down into horizontal and vertical grids, consisting of hundreds and thousands of little pieces that make running computer simulations easier and more manageable. For GCMs to accurately simulate Earth’s climate, they must take into consideration all the physics within the atmosphere, such as convection, fluid dynamics, microphysics, temperature fluxes, turbulence, aerosols, radiative transfer, etc.—in addition to all the physics in the ocean, land, and ice. Aaron says, “Each process affecting the climate is distilled into the computational code, enabling the GCMs to compute the state of the world—everywhere on the planet and at each point in time—like a synthetic Earth, that allows us to project into the future.”
These climate projections allow scientists to understand the implications of society’s future energy choices. Mark says, “In just a single lifetime, CO2 from human activities has caused the Earth to warm an average of 2 degrees Fahrenheit (oF). If we continue on a high-emissions path, then we are looking at close to 10oF of global temperature increases by 2100. 10oF may not seem like a big change, but when discussing average global temperatures, a change of that size puts us in a totally different world. For example, during the Last Ice Age—when an ice sheet 2 miles thick sat over much of North America—the global average was only about 10oF colder than today. In contrast, if we drastically reduce our emissions to net-zero, we can largely prevent further warming—and keep the Earth from becoming a different world.”
Clouds and how they change as the planet warms is another crucial factor that must be taken into consideration when predicting climate change, one which LLNL scientists are actively working to observe and model in the Cloud Feedbacks project. As the planet warms, cloud coverage can increase or decrease. More clouds means that more sunlight is reflected away, providing a sunscreen-like shield, but less clouds means that more light is absorbed, resulting in more warming. While increasing evidence indicates that cloud changes amplify warming, GCMs still require improved cloud representation to better capture their response to warming and produce more confident climate predictions. The high-resolution models currently underway at LLNL show great promise in their ability to represent the microphysical processes of clouds with increased accuracy in GCMs.
Quality over quantity
LLNL scientists are currently working to refine the resolution of GCMs in the Energy Exascale Earth System Model (E3SM), an ongoing collaboration in its second iteration. Conventional GCMs have a resolution of 100 kilometers (km) along the Earth’s surface, which would encapsulate the entire Bay Area. While these low-resolution models can simulate hundreds of years into the future, they cannot provide scientists with the fine, granular details needed to understand what is happening at the local level.
New GCMs like E3SM can zoom in with a resolution of 3 km, however, this level of detail proposes some challenges as it takes a great deal of computational power to simulate. At this increased resolution and with current state-of-the-science computer systems, scientists can only produce simulations one year into the future. Aaron gives an example, saying, “Imagine you’re hosting a dinner party. Now, if you wanted to serve PB&J sandwiches, you could probably make five sandwiches per minute, or 300 an hour—allowing you to serve more people. But, if you wanted to serve sushi rolls, you could probably only make three rolls per hour because they are more complicated to make, and now you’re serving 100 times less food per hour.” This concept, of serving a more complex dish, is similar to producing a higher resolution climate model—where the 100 km resolution is a peanut butter and jelly sandwich, and the 3 km is sushi. To mitigate some of these costs, LLNL scientists are collaborating with other Department of Energy laboratories and leveraging next-generation computer architectures. Until multi-decadal 3 km resolution simulations are available, Livermore scientists are also applying cutting edge artificial intelligence (AI) techniques to extract local trends from conventional resolution simulations.
Artificial intelligence
“AI surrounds us in our everyday lives—for example, Apple’s Siri, self-driving cars, and even the spam filters in our email—and it also has many useful applications in climate science,” says Gemma. She goes on to discuss how her research team uses AI super-resolution techniques—predicting a high-resolution image given a low-resolution image—to quickly and cheaply downscale conventional GCM predictions to very high resolutions. However, before the generated GCM predictions can be used for decision making, they often have biases that need to be corrected. To correct these biases, Gemma and her team use novel AI methods, resulting in more accurate and reliable predictions. Before concluding the lecture, Gemma explains how climate scientists can use also AI methods to improve seasonal forecasts. She says, “Predictions on seasonal timescales are important to a wide range of societal sectors, including agriculture, but they are hindered by observational, model, and computational limitations. To this end, we have developed an AI probabilistic forecast model that learns from GCM simulations to make predictions a season ahead.”
The 2022 SOS lecture series explored some of the many facets of climate science and demonstrated that climate research and CO2 mitigation strategies require a range of scientific disciplines, talents, and techniques. Each lecture told a story about how the climate is changing, what is causing those changes, and how advanced climate model simulations and cutting-edge artificial intelligence techniques can be used to inform scientists about the climate’s future.
View past lectures on LLNL's Science on Saturday YouTube channel.
– Shelby Conn
