Earth & Sky

For more information:

Ron Miller’s Bibliography (NASA’s GISS)

NASA GISS’s Global Climate Modeling Homepage

Pew Center for Global Climate Change link to additional resources

Notes from an interview with physical scientist Dr. Ron Miller:

At our lab, the major goal is to create a computer simulation of climate to understand the changes we’ve observed such as the dust bowl, and ice ages-and to use that understanding to make predictions of the future. Specifically, what affect will changes in atmospheric concentrations of greenhouse gases have on future climate?

Climate models are mathematical equations. We divide the surface of the Earth into cells or grid blocks. Within these cells we track things like energy, water, mass, and winds. By doing this, we keep track of momentum, or Energy. This is a lot like balancing a checkbook each month. We know how much gets deposited, and then need to check to see if our predictions match reality.

For example: Consider Temperature in one grid block. We can measure how much Energy is coming in. If the grid block is over the ocean, then the ocean evaporates water into atmosphere. If warm air comes in, there is an Energy gain. If cold air enters the grid block, then Energy is lost.

We have simultaneous budgets to keep track of conceptually.

Now consider resolution. How much detail do we need to describe the planet? Think of a digital camera, and the # of pixels you use to create an image. With more pixels, you get ever more detail, or resolution. To describe climate, our lab uses 3,312 horizontal blocks (or pixels) over the surface of Earth. They are about 3’ of longitude wide, and 4’ of latitude tall. We also have 20 vertical block overtop of each of the 3,312 horizontal blocks. This is actually a minimum number of blocks for getting an accurate model for Earth’s climate.

For each of these blocks, we begin with actual observations of the variables to know where we are starting off.

Historical climate models were a big leap of faith. Historically these models were used to forecast the weather. By the 1940s and 50s they began to describe how the atmosphere moved around Earth. These equations didn’t include changes in the Earth’s surface-e.g. ocean versus land. Rather they were based upon universally agreed upon equations in elementary thermodynamics to forecast the future. The advantage of this is that the physics are well known (they date back to Isaac Newton). But the disadvantage is that the models were a great oversimplification of reality. Seventy percent of Earth’s surface is water, and 30% is land. These early climate models did not account for this, or anything else such as differences in topography, farmlands, forests, deserts, etc.

We try to bring in an understanding of all these variables to our current climate models-it is a bit of an art form. It is difficult to describe these complications since we don’t understand how al these variables work.

Consider this: How does the surface of the planet respond to sunlight coming in? The surface gets rid of Energy because it tends toward remaining in equilibrium. If the ocean doesn’t get rid of Energy, it will get warmer and warmer and warmer. To remain in equilibrium to the ocean dissipates heat through evaporation-it cools off. Thus, the surface of Earth responds to the heat of the Sun. Evaporation, by the way, is a very efficient way of dispelling heat.

Thus land, on the other hand, depends very heavily on soil moisture to dissipate incoming heat Energy. This relationship has been studied for a few decades, and no simple elegant theory has yet emerged. Scientists use a semi-empirical approach to understand this. For instance, if it rains heavily does the moisture stay in the root zone or seep down much deeper? How soil moisture affects the land can be different in all the quadrants (grid blocks) across the planet.

So, as a result of this-as well as other issues-the physics of climate simulations runs the gamut from well-known equations to empirical observations (like soil moisture). This is both a drawback and a benefit of climate modeling. There is not overarching physical theory to account for everything in a given climate model. But what we have now is far more detailed and predictive than what we had 50 years ago.

We know for example that the empirical observations that we “understand” could change over time as climate itself shifts. For example, in 50 years from now, how will the change in atmospheric concentration of greenhouse gases change the ability of plants to absorb soil moisture?

Or consider that the positioning of rainwater can vary all over the planet, as well as through time. We must therefore, always be aware that empirical relationships could change in the future.

One way to test for the importance of this is by using different empirical/physical schemes within the model for given variables. By observing how much the answer depends on those data, we can get a range of climate estimates.

There are many, many other parameters (besides rainwater, and soil moisture). These include, for example, clouds: their forms, brightness, and reflectivity may all influence the climate. Another is sea ice.

And with future changes in climate, our empirical understanding of all these parameters will change. We will get different answers.

As a result of this, it can be difficult to isolate the source of uncertainty in climate models. It is, therefore, very good that there are plenty of different climate models worldwide. Further, that there is consensus among these indicates we have a quite a good idea of how climate will respond to changes in atmospheric concentrations of greenhouse gases. The proof is in the pudding.

What’s more, we have measured the changes of these gases in the atmosphere over the last few decades. By entering these data into our models, we can “hindcast.” These are historical predictions of climate based on atmospheric concentrations of greenhouse gases. This is a valuable test of climate models. This hindcasting strongly suggests that changes in greenhouse gases over the last 30 years have lead to the observed ?’ C change in temperature (warming) also observed over that time period.

One of the big questions researchers are grappling with is how will the predicted 1-2’ C warming affect various regions? What will be the local impacts of climate warming? This is a research frontier for future climate modelers. Indeed, the IPCC (Intergovernmental Panel on Climate Change) has focused on doing a better job of local climate forecasting. It would help if we had more resolution. With more detailed resolution (pixels or grid blocks, it would be easier to predict what will happen on a local scale). For example, if there is a 3’ C increase globally, what does that mean for the Everglades and Bangladesh?

Another big issue is physical interactions on a local level. We are getting better and better at understanding physical interactions on a local level. For example, dealing with the Rockies. We can now describe the topography and associated Energy fluctuations in more detail. We are searching for a theory to describe how things (e.g. vegetation) respond as climate changes. Consider than there will be vegetation range shifts as the climate warms, pine forests absorb more energy than wheat fields. How will those range shifts in turn affect climate?

A big benefit to global climate models is their ability to reproduce global scale changes. We can say with some certainty what will happen to global climate with a reduction in carbon dioxide, for example.

Of course, it is useful to have some skepticism. In fact, partly because of earlier political scrutiny, our current models are far more reliable. Scientists went out, got observations and data over the last 10-15 years, and now those models are much better. These inspections helped us define our earlier ignorance more precisely. How exactly do ice sheets, clouds, and soil absorption, aerosols (and acid rain) affect climate? We know more about these than we did, but they remain a big challenge-we still have a lot of fine-tuning to do. For instance, we know that climate change will affect ice sheets. This is a really important question-we don’t really understand yet exactly how the rate of ice sheet melting will affect sea level rise. We have learned over the years that the more you know, the more you realize you don’t know.

Still, we have had great success in using the models to reproduce actual observed changes in global temperature-and the global agreement of various climate models underscores their predictive value for historical, current and future climate change. The biggest challenge ahead, I think, is bringing this success with the global models to our regional and local models. How will various regions on Earth be affected by climate change? We know, for instance, that temperature is likely to increase by 3’ C on Earth in general, yet we still don’t know what is going to happen in the middle of the U.S.

It’s hard to know if we will ever “win” over the climate change skeptics. There is currently a lot of debate about how precise our models are for forecasts of the future.

Another benefit to these models is their collaborative nature. Hundreds of people are typically involved in constructing each model. There is a lot of organization and collaboration.

In general, if we didn’t have these models, we would be blind. By changing the concentration of atmospheric (greenhouse) gases we are running an experiment on Earth. Without these mathematical equations (models) to assess these inputs, we would be totally blind to the outcome. The models are great because they give us a chance to see what affect our actions will have before we have to live with the consequences of our actions. The only other way to find out what will happen is to wait around and see. That would be risky, risky business.

A follow-up conversation with Ron Miller

The problem we have is that we can’t really – just because of computational resources, just because of the speed of our computers, we can’t really afford to use an arbitrarily large amount of resolution, we can’t really describe the Earth with as much detail as we’d like. And, in particular, we can’t really resolve for a thunderstorm that you might get on a summer day. Because these storms are really too small to really describe in a global model. So what we have to figure out is, given the climate processes, given the ways that we can resolve, given the large changes in scale to circulation, we have to try and figure out in terms of these variables what the small scales are going to do, what the thunderstorms are going to do. The big problem we have is trying to represent the climate on scales that are smaller than our grid. Because whenever there are processes that are very important that we’re not explicitly resolving, that we’re not really explicitly describing, we have to take these into account somehow the processes that we can’t represent. I mentioned thunderstorms as being one thing that we can’t really resolve because we don’t really have the spatial resolution to do the spatial detail. Clouds in general are something where we try to make educated guesses whether or not a cloud will form given the large-scale properties. To make things more complete, It might be helpful for me to say, these cells that I talked about, these pixels, to use that analogy, of these global model tend to be on the order of say 50 to 100 miles on the side, either are squares or they’re 100 miles on the side, they’re relatively large areas. And those are in the best models. In terms of resolution, we represent the troposphere, the layer where all the weather happens in the inner atmosphere, we divide that typically into between 10 and 20 layers. So, there are processes that go on in a smaller scale, then we can’t really describe those explicitly, and so we have to represent those in terms of the things that we can represent on a larger scale. And I mentioned thunderstorms and clouds as an example.

By changing the composition of the Earth’s atmosphere, we know that it’s going to change the climate. And the big question that everyone would like to answer is, by how much are we going to change the climate. And beyond that, how is that going to affect life on this planet. As an alternative to doing these computer simulations, we really have no way to predict the future without looking back at the past. And one way that we can try to guess how the climate is going to change is by looking back at the changes with the ice ages for example, and trying to figure out how changes in the amount of energy coming into the planet from the sun were reflected in changes in the temperature. And that gives us some idea of how sensitive the climate is to a change in radiant forcing. The problem is that this method is very imprecise. And also, we really don’t know if the ice age is really a good analog for what’s going to happen in the future. Now on the one hand we know for a fact that it was very cold during the ice ages, and we were getting less energy from the sun. And we know that it’s not in a way a good direct analogy for global warming. In fact, it’s going to get hotter, and we’re not going to have as large a change in the composition in the atmosphere. We know that we’re not going to have less radiation coming out of the top of the atmosphere. But I think in the absence of these computer models we really have very little idea of what is going to be the effect of our actions in terms of changing the composition of the atmosphere.

We really are changing the climate by changing the composition of the atmosphere. And so one question that we would like to answer is, if we stop changing the composition of the atmosphere or we don’t change it to the extant that we’ve been changing it so far, if we burn less fossil fuels, and emit less carbon dioxide into the atmosphere, how much warming is that going to spare us? Can we really avoid a lot of the more extreme outcomes of changing the climate. Can we avoid some of the worst outcomes like sea level rise or increased frequency of droughts. for example? And in order to really answer that question you have to have some way of knowing how sensitive the climate is to changes in the composition. And, using these models, we can say, well, if we increase the amount of carbon dioxide into the atmosphere at a rate that’s half as much as we have been over the 20 years, that’ll spare us a certain amount of warming. In that sense the models are very useful. And just to give you an idea of what are the changes in our energy consumption that we’re going to have to make in order to avoid some of these outcomes of climate change.

I think that because this is such a contentious issue, one of the advantages of that is that it’s really placed a lot of scrutiny on these climate simulations. I think that because of the scrutiny, we actually have a lot more confidence in these computer models, in these simulations, than we had ten years ago. And so, to a certain extent, because this is such a controversial issue, there hasn’t really been a lot of action taken in terms of energy consumption. But maybe the silver lining in all that is the fact that I think we know a lot more about climate now just because we’ve been forced to look at it very carefully in order to try to predict the future.

The following person was interviewed for today’s program. Or thanks to:

Dr. Ron Miller
Physical Scientist
NASA Goddard Institute for Space Studies
New York, NY