UCAR Staff Notes masthead
Home Our Organization Research News Center Education Community Tools Libraries
About Staff Notes
Past Issues
Favorite Photos
How to Subscribe


staff notes header

March 2006

The atmospheres of cities

With urban areas expanding worldwide, researchers want to better understand their impacts on weather and climate.

Fei Chen
Fei Chen. (Photos by Carlye Calvin.)

As a frequent traveler to some of the world's largest cities, RAL's Fei Chen knows firsthand that a downtown center can feel much warmer than neighboring suburban or rural areas.

But try explaining that to a computer model. Neither forecasting models that anticipate tomorrow's weather nor global models that simulate climate over decades or centuries pay much attention to the world's expanding urban centers.

As a result, both weather and climate models may be omitting important information. A key problem for Fei, and a number of his colleagues at NCAR and elsewhere, is how to accurately incorporate details about enormously complex metropolitan areas into simulations without overwhelming the models with too much data.

"We are just in the infancy stage of urban modeling," Fei explains. "It's imperative for us to solve these problems and include urban areas in computer weather models, especially because more people are living in cities."

Cities affect atmospheric conditions in several ways, particularly by emitting heat from paved surfaces, a phenomenon known as the urban heat island effect (see sidebar).

Here's a look at some NCAR research into the atmospheric impacts of urban areas.

Protecting New York

As the nation's largest metropolitan area and a major port, New York City presents especially formidable challenges to forecasters. Not only do researchers have to weigh the impact of the city's dense streets and concrete canyons on local temperatures, but they also have to account for how its combination of soaring skyscrapers and complex coastlines affects prevailing winds.

Given that New York is considered a top terrorist target, local and national security agencies are determined to map the city's weather patterns. If terrorists were to release radiation or another type of toxin into the air, security officials would want to determine the direction of the plume so they could evacuate people in harm's way.

Fei Chen
Daran Rife.

To that end, RAL's Daran Rife is heading up an effort to produce fine-scale weather forecasts of the New York metropolitan area. The three-year project will involve amassing an enormous amount of information about the area's topography and land use because, as Daran warns, "if you don't get the details right, a plume might go in the exact opposite direction than your model indicates."

In addition to Daran, the RAL team includes Tom Warner, Yubao Liu, Mukul Tewari, and Fei. The researchers are building on previous national security projects, including the use of NCAR models to forecast atmospheric conditions around Salt Lake City during the 2002 Winter Olympics.

One of the team's main tasks will be to amass the latest information from NASA Earth science data sets. The data sets are vital because they show the extent of development in the New York metropolitan area. In contrast, older maps that RAL researchers have examined from the early 1990s are of limited value because they predate a housing boom that has swallowed up farms and undeveloped pieces of land on the city's fringes.

The data sets, along with NASA satellite weather observations, will be fed into NCAR's Weather Research and Forecasting model (WRF), which will be coupled to a new Urban Canopy Model. The team then will run simulations with a horizontal resolution of 1.5 kilometers (just under one square mile per grid box). Standard weather forecasting models, in contrast, have a much coarser horizontal resolution of about 10 to 15 kilometers (6 to 9 miles).

These tools will enable Daran and his colleagues to capture such atmospheric details as the movement of winds through river valleys and the prevailing flows of sea breezes, as well as finer-scale temperature and wind patterns in Manhattan.

The project, which also involves Lawrence Livermore National Laboratory and the University of Alabama–Huntsville, will test the models against observed weather conditions. After that, RAL and its partners will provide the system to national security agencies, as well as publish findings about urban meteorology.

Urban heat

A world without buildings

As far as NCAR's premier climate model is concerned, cities don't exist.

The Community Climate System Model (CCSM) doesn't include data about urban areas. But that may change soon. "It's important that we get cities in there," CGD's Keith Oleson explains. "Right now, the whole U.S. East Coast is vegetated in the model."

Keith Oleson
Keith Oleson.

Keith and CGD's Gordon Bonan want to use the model to determine whether climate change may have a different impact on cities than on rural areas. In a collaboration between CGD and the University of Kansas, researchers are developing global data sets of urban characteristics that can be fed into the land component of the CCSM. The team is using data from field campaigns in Vancouver, Mexico City, and other urban areas to validate the model runs.

Within a year or so, they want to couple this information with the atmospheric component and begin testing how well the model simulates urban climate.

Getting fine-scale detail into the CCSM is particularly daunting. Unlike the regional (or mesoscale) models RAL is using, the CCSM doesn't focus on day-to-day weather. Instead, it simulates global climate on timescales of decades and centuries.

Keith's research draws on earlier work that incorporated different types of land surfaces, such as plant communities, into the CCSM. But cities, perhaps even more than land surfaces such as forests and lakes, can have subtle and hugely complex impacts on the local atmosphere. The size and density of an urban area, the darkness of its streets and sidewalks, and even the types of its construction can affect local climate.

One issue that CGD is dealing with, for example, is how to simulate the impact of building materials. Surfaces made of dense concrete can store more heat than surfaces made of stone, which means the historic core of Mexico City, with many stone structures, affects local temperatures differently than Vancouver.

"Our whole group is used to modeling vegetation," Keith says. "This is a big learning curve as we model urban areas and processes."

One of the group's ultimate goals is to help city planners prepare for an altered climate. Incorporating cities into the CCSM will guide policy makers who are weighing the creation of more parks or rooftop gardens as a way of offsetting higher temperatures.

Urban heat islands

For several decades, scientists have known that downtowns experience different weather than surrounding areas.

One of the reasons has to do with pavement. Since paved surfaces absorb more sunlight than plants and soil, dense urban areas are often 1–3°C (2–5°F) warmer than nearby suburbs or farms. The geometry of urban canyons and impervious nature of paved surfaces also act to retain heat. The so-called urban heat island effect is particularly noticeable at night, when streets and sidewalks gradually release warmth back into the atmosphere.

Skyscrapers also influence weather by disrupting air flow, creating eddies and currents. Pollutants, such as ground-level ozone and airborne particles, affect temperatures by either blocking solar radiation or trapping it. Even indoor heating and air conditioning systems can have subtle influences on weather, as poorly insulated buildings leak air at room temperatures into the atmosphere.

Some of these impacts are so profound that they can affect winds and temperatures for hundreds of miles in every direction.

Moving downwind

To find out more about what happens downwind of a large urban area, ACD's Sasha Madronich and a large team of collaborators this month are mounting one of the most complex field campaigns ever undertaken in atmospheric chemistry. The researchers are making multiple research flights in the NSF/NCAR C-130 aircraft and operating ground instruments to investigate the chemical and physical transformation of air pollution as it flows downwind from Mexico City.

Sasha Madronich
Sasha Madronich.

The team's goal is to assess the pollution's impact on regional and global air quality, climate, and ecosystems. The results are expected to be applicable not just to Mexico City, but also to other megacities (cities with 10 million or more inhabitants) around the world.

"Mexico City's pollution probably doesn't have a global impact, but all urban areas together do, and the world is urbanizing," Sasha explains.

The project, called MIRAGE (Megacity Impacts on Regional and Global Environments), is led by NCAR in partnership with a number of U.S. universities and other organizations. The researchers hope to shed light on four broad questions:

  • How far downwind does Mexico City's pollution plume extend?

  • How are the pollutants transformed by chemical reactions occurring downwind of the city?

  • How do the pollutants affect visibility, as well as regional and global climate?

  • How do the urban pollutants interact with pollutants from other sources, such as agricultural and forest fires?

Because air pollution is complicated, both chemically and physically, and evolves over time and distance, scientists have traditionally faced difficulty in quantifying its components. The MIRAGE team will use aircraft, ground stations, and satellite observations to gather data on how Mexico City's air pollution ages as it disperses in the first hours and days after emission.

The researchers chose Mexico City for MIRAGE because it is the world's third largest urban area, has some of the worst air quality in the world, and is situated in the tropics, as are most fast-growing megacities.

Current computer models for studying air pollution were developed mainly for cities in industrialized nations, Sasha says. They don't transfer well to megacities in the developing world, where people are more likely to burn coal and wood and drive vehicles that emit more harmful chemicals.

The field campaign will also gather information about aerosols, such as how long they endure in the atmosphere and how they affect clouds. These insights are important for climate modelers.

"The lifetime of organic aerosols may be longer than climate modelers have thought," Sasha says. "This could have a huge effect on climate."

Small-scale influences. CGD researchers, who want to incorporate the impacts of cities into the Community Climate System Model, are breaking down large grid cells into subgrids to reveal small-scale urban influences. To represent climate in medium-density urban areas, for example, they are adding the effects of roofs, sunlit and shaded walls, and paved and unpaved surfaces to the land component of the model. (Graphic by Keith Oleson)
grid cells

Regional impacts

Even if you live in a rural area, a city that's an hour or two down the highway may affect your weather.

In fact, the more that RAL's Fei examines the impacts of cities, the more interesting results come to light. For example, an analysis of the Houston area revealed that the city's relative warmth altered the directions of winds at least 100 miles away. "If you don't consider the urban heat island, you cannot accurately simulate weather on the mesoscale level," Fei concludes.

The reason has to do with the movement of air masses around a city. As urban air heats up, it rises and allows cool air from surrounding areas to move in, thereby affecting local air currents.
These movements can also affect precipitation: if a city is by a large body of water, it will tend to draw in moist air that's over the water. The moist air may release rain over the city or a little downwind.

City leaders, says Fei, need to be aware of how these changes can affect them.

"If you don't consider the urban heat island, you cannot accurately simulate weather on the mesoscale level."

—Fei Chen

For example, rapid growth in Beijing has pushed some precipitation away from reservoirs on the city's northern edge, which may eventually cause problems for the city's drinking water supply. And in Hong Kong, Fei and a former RAL visitor, Jeff Lo, working with colleagues at the Hong Kong University of Science and Technology, found that the city's relatively high nighttime temperatures were reducing the strength of nocturnal land breezes that used to send city air toward the South China Sea. As a result, pollution tends to hover over the city instead of dispersing.

One of Fei's goals is to determine how land use changes along a city's edges can affect weather in the area. He and Mukul are working with Christine Wiedinmyer (ACD/TIIMES) and Alex Guenther and Xue Tie (both in ACD) to look at urbanized areas outside Dallas, where new pavement and changes in land use are affecting soil moisture and heat.

It's not just new subdivisions and strip malls that can affect weather. By ripping up natural areas and planting nonnative trees and grass, developers in the area may also inadvertently alter the local composition of atmospheric chemicals in ways that can impact air quality. Different types of plants emit volatile organic compounds that can combine with emissions from motor vehicles and industrial sources to form ground-level ozone, an air pollutant and greenhouse gas.
Fei is also working on combining weather models with computational fluid dynamic models that simulate the atmosphere on an extremely fine scale. Such models can include individual buildings.

But a central question for Fei and his colleagues is whether too much detail can actually undermine the accuracy of a model.

On one hand, a model that includes the heights and shapes of individual buildings may provide more accurate simulations of local windflows. On the other hand, the mass of data can overwhelm a computer. Another concern is that the more data entered into a model, the greater the chance of introducing
an error.

"Each piece of data has to be right," Fei points out. "If you put in something wrong, the model produces wrong results."

Most researchers agree, however, that getting at least basic land use information into a model is useful. For example, model simulations tend to be more accurate if they include the location of a downtown business district (where the heat island effect is especially strong) and the location of residential areas (where lawns and parks create comparatively cooler conditions). To that end, Fei participated in a large collaborative effort with several government agencies and labs, as well as universities, to develop a database characterizing the landscapes of cities.

While Fei is focusing on local and regional scales, there is also a question of whether the world's cities have an even greater impact. Could streets and buildings affect climate on a global scale, even though they take up only a small percentage of the world's land surface? Scientists do not yet know.

"We are just at the beginning," Fei says. "There's no clear answer to this."

• by David Hosansky and Nicole Gordon

On the Web

For more about RAL's fine-scale modeling for national security
For more about CGD's land surface modeling

Also in this issue...

The atmospheres of cities
Sidebar: Urban heat islands

Remembering Andrew Crook and Diana Josephson

Let it snow

AGU, AMS presidencies

Delphi Questions: Toilet paper; professional memberships

Just One Look


Staff Notes home page | News Center