Climate Modeling Expert Shares Insights on Progress, Coming Advances
Eric Barron, Director, National Center for Atmospheric Research
Credit: The National Center for Atmospheric Research
To gain insight on the strengths and weaknesses of modern Global Climate Models, Southwest Climate Change Network reporter Chelsey Killebrew interviewed Eric Barron, director of the National Center for Atmospheric Research and a long-time expert on climate models. As a graduate student at NCAR, Barron worked with the research world’s first supercomputer, developed by Cray. As a postdoctoral researcher there, he pioneered the use of climate models to reproduce climatic conditions of the past. He continues to work with GCMs and monitor their strengths and weaknesses.
Chelsey Killebrew: What are some of the strengths of global climate models?
Eric Barron: I would say one of the great strengths is the ability to represent the modern climate. Secondly, we’re capable of simulating the last 150 years. The reason why that’s so important is that many different things have been happening over the last 150 years that are influencing climate. The sun has changed, greenhouse gases have changed, the amount of particles put up in the atmosphere by humans and volcanoes have changed.
So if we can simulate the last 150 years, that suggests that the models are able to capture how the ocean-atmosphere system works, and can simulate the responses to agents that are working to change it —what we would call “forcing factors.”
So that’s what I would say is a strength—the ability to capture the modern climate and the historical climate.
Chelsey Killebrew: What are some of the weaknesses in global climate models?
Eric Barron: One is capturing the variability in the climate system. Now, this is the variability that’s on the time scale of years to decades, and so you should think about how the whole system operates—if soil moisture is changing, if ocean temperatures in a particular region are changing - all of these contribute to the differences from one year to the next.
So in order to make predictions from one year to a decade, we have to know nearly everything. We have to know everything about how the system is changing and how all its components interact. We don’t have the observations to do that, and we don’t have the knowledge of all those interactions to do it ... so that’s viewed as a weakness.
The second weakness is the fact that the scale of most things—the scale of vegetation, the scale of farms, the scale of water systems—tend to be, spatially, very fine. Yet, we don’t have the computer power to get that level of detail.
Twenty years ago our models had two points that would resolve (or represent) the state of Pennsylvania, and now we’re pushing the resolution of these models to get down to 30 kilometers (i.e., 30 km by 30 km, roughly 350 square miles). But in a resolution of 30 kilometers, you still don’t capture all of the atmosphere’s features or all of the characteristics of the landscape. So this means that our ability to say what’s happening regionally is also a challenge.
In my view, those are the two big ones—getting the variability right and capturing the fine scale.
Chelsey Killebrew: has modeling improved in recent years?
Eric Barron: With each year and with each increase in computer power, we’re increasing that resolution—that ability to resolve features. So that’s one significant improvement.
Figure 1. This image captures the distribution of water vapor in the atmosphere at one moment in time during a climate simulation by the NCAR-based Community Climate System Model.
| Enlarge This Figure |
Credit: University Corporation for Atmospheric Research
The other thing that you can see happening is that we’re increasingly adding different components … of the entire Earth system. We recognize that we want to reach to the top of the atmosphere in the models. We realize that we want to include the land-surface in the models. We realize that some of the deeper ocean physics has to be included in the models. So what you’re seeing is more of the system being added to the models and a higher resolution of those models.
Chelsey Killebrew: has increasing computer power helped the model projections?
Eric Barron: It’s absolutely hand-and-hand. This is a dependency that has existed for 30 years, in which, with each increase in computing power, we push the climate models a little harder. And so we’re absolutely following in sync with the increase in computer power.
Chelsey Killebrew: In what ways could the global models be applied to consider impacts at the regional scale, keeping the Southwest in mind?
Eric Barron: This is what I call a burgeoning area of science and a burgeoning area of growing connection to society. For a long period of time, climate modelers have been working to make better and better predictions.
Figure 2. NCAR's bluefire supercomputer has a peak speed of more than 76 teraflops (76 trillion floating-point operations per second).
| Enlarge This Figure |
Credit: University Corporation for Atmospheric Research
And all of a sudden, in the last decade, but not really funded yet, are groups of people looking the connection between climate and society—what are the implications for agriculture, what are the implications for water resources, or for human health. And they’re starting to work to do things like downscale models … to get a finer resolution data set out of the models so that they can apply it to a particular region like the Southwest.
More and more people are starting to develop what I would call a “second generation” of models or analyses to try to look at everything from human health to water to agriculture to how ecosystems will change. And so my bet is this will be a tremendous area of growth in science over the next decade, but it’s still in its infancy.
Chelsey Killebrew: What are some of the challenges modelers face, especially with climate change, at the regional level?
Eric Barron: The task is to project climate change well into the future and then to try and see how it will impact ecosystems.
One of the things that we’re seeing is, for example, that warm air and drought are combining to make lodgepole pine susceptible to a beetle infestation. And so you see an almost instantaneous change in the characteristics of the forest across the West as so many of these trees are damaged. So this is a case where, if you thought about time (in geological terms, since the beginning of living organisms), almost in the snap of a finger you’ve had this substantial impact on ecosystems. … yet so far the level of climate change is fairly modest. It wasn’t an ecosystem change that we predicted.
This is the reason why the assessment of climate change impact and the connection between climate to all these different factors requires a lot more science to it. Because we’re used to saying, “Oh, the maple tree likes to live in this particular climate,” but we’re not used to saying how the dynamics of the forest will change when you start including everything, including things like how insects will respond.
Chelsey Killebrew:In what ways are models improving, and likely to improve in the future?
Eric Barron: I think what you’re going to see is a continual push to get to higher resolution, and that means that your physical understanding of what’s happening—whether it’s clouds or the land surface—has to improve right along with it.
It doesn’t do you any good to get finer and finer spatial resolution if you don’t understand the physics at that resolution. And so this is a continual march between improving resolution and understanding the physics.
Then, the next thing you see is that people will work to incorporate more and more of the Earth’s systems (e.g., vegetation, deep-sea and upper atmospheric conditions).
And the third thing that you see is a very conscious effort to try to promote regional simulations, because we realize that’s what the decision makers need.
Chelsey Killebrew:Do you have some ideas on how stakeholders can apply these projections to their areas of interest, like water managers for example?
Eric Barron:I think it’s really healthy to apply a lot of different scenarios, and a range of different model simulations, and try to get a sense of the range of possibility. If a whole group of models predict the same tendency for water availability then it doesn’t matter which model you’re using out there … you have a problem.
One (model) may suggest a big change and the other one suggests a smaller change, but you realize that either case suggests a significant impact on water availability in Arizona, for example.
So this is a matter of looking at these climate models and recognizing that there’s a certain level of uncertainty in them, but you’re looking at the risks and the vulnerabilities and then deciding whether it’s worth it. You may also already recognize that you’re up against a problem and any change that goes in the direction the climate models suggest is going to make the problem worse..
Then, in my mind, those start to become areas where you take action, or it makes sense to take action. There are a lot of cities that are living at the edge of their water resources and even smaller changes have impact.
So if you started to look at climate model after climate model and you looked at what they were simulating and you realized they all were giving you—or 85 percent of them were giving you—the same tendency, then I think I would pay attention.
Editor’s Note: Following the interview, a few changes to improve clarity were made, in consultation with Eric Barron.