Arizona's next top model

Thursday, April 12, 2012
by Diane Boudreau

Before you read the next paragraph, try to think of all the different factors that affect outdoor temperature in a desert city like Phoenix, Arizona.

You probably thought of at least a few of these: building materials and building height, pavement, plants and trees, bodies of water, humidity, cloud cover, wind, and even global warming. Maybe you thought of other things, as well.

Now imagine that you are trying to understand how all of these things work together and affect each other. Perhaps you want to predict what the temperature will be in a future with more buildings, less water and more grass.

How could you calculate all those different factors in order to make a prediction? Actually, you can’t. Even the world’s brainiest scientists can’t—not without help.

“It’s so complex that a brain cannot hold it all together and do the calculations. You have to have a computer to solve these equations,” says Susanne Grossman-Clarke.

Grossman-Clarke is a physicist at the Potsdam Institute for Climate Impact Research in Potsdam, Germany. She works with ASU researchers studying the heat island in Phoenix. She is trying to understand what influences heat in urban areas. She also wants to know what is likely to happen in the future.

The computer models that she uses “speak” in the language of mathematics. Scientists enter numbers into the model to represent a set of conditions, like building heights or wind speed. The computer runs these numbers through a set of equations to simulate, or imitate, what would happen in the real world under these conditions.

Raw data from a computer
These numbers describe temperatures (in kelvin) around Phoenix on a summer night. Do they make sense to you?

Map of temperature data
The data from above has been put onto a map and color-coded by temperature (in Fahrenheit). Does it make more sense now?

The computer spits out even more numbers in return. The scientists use software to take those numbers and turn them into graphs, maps or other images that help them understand the results.

“We take everything we know about atmospheric processes and how the air interacts with the land surfaces. Then we formulate equations and put them in these models,” she explains. “You have to use the best supercomputers in the world. Even 10 years ago the computing technology available couldn’t do these simulations on a city level.”

Grossman-Clarke doesn’t just use the computer models. She helped develop them, as well. She spent seven months at the National Center for Atmospheric Research in Boulder, Colorado. There, she worked on improving the physics for city surfaces so that they are represented in the models.

To show the effects of cities on temperature, Grossman-Clarke needs to work at a very high resolution. What does that mean?

Resolution describes how many units are contained in a certain amount of space. In photos, for example, we measure resolution in dots per inch (dpi). Each dot contains a single color. The more dots you have (high dpi), the more detail you can see in the photo.

There is a tradeoff, however. A high-dpi photo is sharper, but it takes up more computer memory and requires more ink to print.

For computer models, each “dot” is a piece of information. So in a climate model, a “dot” might represent the average temperature for that area.

Global climate models usually have a 100-kilometer resolution. This means they can find the average temperature for an area of 100-square-kilometers. There may be lots of different temperatures inside of that area, but the model will only “see” the average of all of them.

This level of resolution cannot show the heat island effect for an individual city. But it lets scientists study global phenomena like the jet stream.

Grossman-Clarke runs models at a 1-kilometer resolution. This allows her to see differences between much smaller areas, like neighborhoods. She can figure out how city structures—like buildings and roads—affect temperature.

Just like with photos, the resolution of a computer model has its tradeoffs. Modeling at a high resolution, like 1 km, takes much more time and computing power than modeling at a lower resolution, like 100 km.

“To model a whole summer at 1 kilometer resolution takes the computer about a week,” says Grossman-Clarke. And these are not average desktop computers. The supercomputer that she uses has 256 processors. Computers on this scale are housed at universities and research centers where lots of scientists can use them.

Grossman-Clarke has been simulating past and future heat waves for Phoenix. But why simulate the past if we already know what happened? She says that our observations can tell us what happened in the past, but they can’t explain why. Models help her to understand the processes that caused the heat waves we all sweated through.

Another reason to simulate the past is to find out if our models of the future will be reliable. For instance, a scientist might enter information about the weather, buildings, pavement, plant life and wind flow for a particular day in the past. Then he or she will run the model to simulate the rest of the summer. If the simulation matches what really happened that year, the scientist can be confident that the model is accurate.

“You never know if the model predicts the future right, but you can trust it more if it simulated the past well,” says Grossman-Clarke.

Grossman-Clarke knew that heat waves happened in Phoenix in 2003, 2005, 2006 and 2007. She simulated each of those summers in her model.

She found that the model simulated the heat waves quite well. She also found that changes in land use played a big part in causing temperatures to rise. These changes mostly involve converting farmland to urban spaces.

“The development hugely increased the nighttime temperatures,” she says. “If you develop agricultural land into urban, it’s 10 degrees Celsius warmer at night.”

These results help Grossman-Clarke answer her main research question: how much is temperature change affected by land use, and how much by global climate change?

“For the future, we can separate land use and global warming effects, because we understand now what land use effects do,” she says.

So what will happen in the future? Grossman-Clarke says it’s hard to make predictions, because we don’t know how much the population will grow, how land use will change, or how much carbon dioxide people will emit. However, she simulated the summer of 2058 using land use information from today.

She found that the future is likely to bring more frequent, and longer, heat waves (by today’s definition) than we experience today. In fact, they could last as long as 50 days!

“In the future the risk that it will get hotter, and for longer periods of time, is very high,” she says, adding, “But I would be very happy to be proven wrong!”


Teachers: find resources and lesson plans about the urban heat island at Ecology Explorers.