More than half of the people on the planet live in cities, yet cities have been largely ignored by climate scientists. But city climate is important, because it can tell us a great deal about how climate change will affect billions of people.
The reason for this lack of research, says Dan Li, a Boston University College of Arts & Sciences (CAS) assistant professor of Earth and environment, is that until quite recently computer models lacked the power to properly study the climate of cities, which are minuscule compared to the areas usually covered by climate researchers.
Working through BU’s Undergraduate Research Opportunities Program (UROP), Li and Yaofeng Gu (CAS’17) are testing a new climate model that incorporates these urban areas. UROP matches students with faculty mentors across the University and provides funding for projects.
On most temperature maps of the United States, it’s easy to spot the cities because of what’s called the urban heat island effect, which is caused by several factors. “One of the key reasons cities are hotter is that cities don’t evaporate that much,” says Li. “We use a lot of impervious surfaces, and impervious surfaces don’t evaporate as compared to soil and trees, which can hold water from rain and then evaporate it back into the atmosphere—this is kind of a cooling effect—while cities just don’t have this mechanism because we use a lot of concrete, asphalt, those types of things.”
Cities also have unique properties, he says: for example, tall buildings create long shadows that influence temperature, making it hard to get an accurate read on climate models. Until recently, many modelers didn’t even try. Heat islands can also skew older climate simulations. Older models might, for example, paint all of northern Ohio as unusually cooler because they neglected the influence of Cleveland.
One way to test a model intended to project the future is to test it on one from the past, so Gu has been plugging in historic climate data from across the continental United States from 1981 to 2000.
“Some people argue that if you can simulate the historic period correctly, it doesn’t guarantee that you can simulate the future correctly,” Li says. “This is true, but if you get the historical period wrong, then it is more likely that you will get the future wrong.”
Gu says that he can simulate the past urban heat island effect down to less than a month. Plug in a date, and out pops the simulation: a map of the entire continental United States broken up into little shaded squares—or cells—of cold blues and warm reds. Each cell looks smaller than the head of a pin, but that’s deceiving.
“They’re actually huge,” says Gu—each square is about 50 kilometers across, which is considered very fine detail in climate models, sufficient to separate out the cities from the rural areas. Their model, developed for the National Oceanic and Atmospheric Administration’s Geophysical Fluid Dynamics Laboratory, will help form the sixth report of the Intergovernmental Panel on Climate Change, an international organization for assessing the science related to climate change.
He emphasizes that this does not mean the new model can predict weather, “but we can use our model to more or less predict what the temperature will be,” he says. “We can never give an exact account of what the future will be.”
That’s because these simulations have an inherent catch-22, says Keith Oleson, a project scientist at the federally funded National Center for Atmospheric Research, who also works on urban climate models. When you have more data “it becomes more difficult to understand, because you have so many processes that they’re competing against each other, canceling each other out,” Oleson says. “There is value in having a simpler model where you can really understand what is going on, but the danger is that you can miss some important processes.”
“There’s no absolute truth,” says Gu. “We just try to be as accurate as possible with the model we have.”