Urban Rain. by Holli Riebeek, design by Robert Simmon, December 8, 2006

A typical summer day at NASA Goddard Space Flight Center in Greenbelt, Maryland, dawns clear, warm, and sticky-humid, with the promise of misery by mid-day. As it climbs higher in the sky, the Sun swims through the thin haze generated by thousands of automobiles, and by the time it has dropped past its zenith, it is joined by downy piles of clouds. From an office at Goddard, the light streaming in through the window begins to dim as the clouds pile up, but that can be ignored until a distant rumble shakes the sky. By now it is late afternoon, and the office hallways are suddenly busy as people scramble to leave, hoping to reach home before Washington, D.C., traffic grinds to a halt in the sudden downpour. All of this Dr. Marshall Shepherd, an Associate Professor of Atmospheric Sciences and Geography at the University of Georgia, observed summer after summer while studying weather and climate at NASA, but while he might have sometimes worried about what rain might do to the city—water puddling on the road or overwhelming storm drains near his house—his real concern was what the city was doing to the rain.

  Page 2

Title image copyright Kevin Walter.

  Photograph of a thunderstorm over Miami, Florida

“Cities impact rainfall and can create their own rain and storms,” Shepherd explains. Though the meteorologist and climate scientist noticed the thundershowers that frequently capped off a summer day in the Washington-Baltimore region, he wasn’t aware that the cities might be influencing the rain until he noticed heavy rainfall around cities in satellite data. Reviewing the work of other scientists to determine what caused the anomaly, he learned that scientists suspected that cities were causing more rain to fall, but the connection was far from definite. He came across many theories about how cities might influence rain, but discovered that the exact process remained unknown.


Evening storms frequently strike southern cities in the summer, bringing thunder, lightning, and heavy rain. The confluence of cities and storms may not be a coincidence: more rain falls downwind of some major urban areas than in the surrounding countryside. Satellite measurements and computer models are helping scientists understand why. (Photograph copyright Eugenia y Julian.)

  Photograph of Marshal Shepherd

Shepherd realized that figuring out if and how cities affect rainfall was more than just an interesting scientific challenge. If the connection was real, weather forecasting, hydrologic, and climate models would need to take city-induced rain into account when warning people about potential flooding rains, storms, and even future climate change. It was clear to Shepherd that the problem was only going to get worse: as cities around the world continued to grow, they would be able to change weather patterns on an increasingly larger scale.

Shepherd believed NASA’s fleet of Earth-observing satellites could help him understand how cities influence rain. Previous work relied primarily on field experiments around one or a few cities. Satellite data provided a way to look at many cities under varying conditions. Shepherd decided to see if NASA satellites could confirm that more rain falls around cities in general, and then take a more detailed look at individual cities to see what the satellite data could tell him about how the city influenced rain. Ultimately, Shepherd hoped he and others could use the data to generate computer models that would provide insight into how cities influence rain today and how their influence might change in the future.

Discovering Urban Rain

When Shepherd first started studying satellite rainfall data, he wasn’t interested in urban rain; he just wanted to understand small-scale weather processes along the U.S. Gulf Coast. During the summer, weather tends to be generated by local processes: hot, humid air piles up along the face of a mountain, triggering a thunderstorm; cool moist air blows off a lake, collides with hot air over land, and rain clouds form. Though these and other local processes are still at work at other times of the year, they tend to get swallowed up by large weather fronts that bulldoze their way across a continent, leaving wide swaths of snow or rain in their wake.


Marshall Shepherd, of the University of Georgia, uses NASA satellites such as the Tropical Rainfall Measuring Mission (TRMM) to study weather and climate. (Photograph courtesy Marshall Shepherd.)

  Weather satellite image of afternoon thunderstorms near the Gulf Coast

It was the local weather-making processes that Shepherd wanted to learn about in 2001 when he started to look at coastal rainfall. NASA, in collaboration with the Japanese space agency, had launched a new satellite late in 1997 that was sending back the best estimates of rainfall in the tropics to date. By early 2001, the satellite had accumulated a three-year record. The record was long enough to begin to reveal year-to-year rainfall patterns. Shepherd was eager to see if the satellite had captured rainfall produced by local-scale events. As he surveyed the satellite data, he was looking for things like rainfall patterns created when a sea breeze interacts with coastal topography, but the pattern he saw emerging from the data of the southern United States surprised him.


Summer afternoon thunderstorms are a common occurrence near the Gulf Coast. Humid air rising off the hot ground cools as it ascends. The water vapor in the air condenses, and clouds form. These storms sprang up along the border between Texas and Louisiana on September 6, 2006. (Image and animations by Robert Simmon, based on NOAA GOES super rapid-scan data.)

  • small (9.8 MB QuickTime)
  • large (35 MB QuickTime)

Map of summer rain rate for eastern Texas showing enhanced rainfall downwind of urban areas

“I started noticing that around some of the cities, there seemed to be these anomalies in rainfall downwind,” Shepherd recalls. These downwind areas got more rain during the summer than other areas. Having spent his childhood in Atlanta, the idea that his city could be directly affecting the weather intrigued Shepherd. Had anyone else noticed this spike in rainfall? He combed through past studies and discovered that for more than a century, scientists had suspected that cities impact or maybe even create rain. Cities, Shepherd learned, are among the local factors that influence summer weather.

“You need three basic ingredients for clouds and rainfall to develop,” Shepherd explains. “You need for air to be unstable.” Air is unstable when it is warmer than the air that surrounds it. Once the air is lifted, it will continue to rise. That instability usually happens when warm air nearest the Earth is pushed up into the cooler air above it. That leads to the second ingredient: a source of lift. “You need something that will get the air rising, whether that be a cold front, or a mountain, or a sea breeze, or a city.” When surrounded by cooler air, the warm air rises naturally like a hot air balloon. “Three, you need moisture.” If there is enough moisture in the rising, cooling air, the water vapor will condense into clouds and rain. But where does the city enter the process?


Measurements from TRMM revealed elevated rain rates downwind of urban areas in Texas. Marshall Shepherd noticed the pattern in summer rainfall while exploring the interactions between sea breezes and the urban landscape of Houston. The heaviest rain (blue) occurred downwind of Houston. This image is based on TRMM and rain-gauge measurements during July, August, and September from 1998 through 2006. (Map by Robert Simmon and Jesse Allen, based on Global Precipitation Analysis data.)


The Impact of City Landscapes on Rain


“There’s a debate about how cities affect rainfall,” Shepherd answers. “There are several hypotheses about what is going on, but they primarily involve the urban land use and urban aerosols.” The first hypothesis deals with the urban “heat island” effect. Cities are made of heat-absorbing materials like concrete, steel, and asphalt. Add to that the heat pumped into the atmosphere by the machines that are concentrated in cities and a lack of cooling vegetation, and the temperature goes up. Average temperatures in a city can be as much as six to eight degrees Fahrenheit higher than surrounding rural and suburban landscapes. Called the urban heat island effect, this increased temperature may provide a source of unstable air. If air over a city is warmer than the air surrounding it, it wants to rise. As the city-warmed air rises, it cools and forms rain-producing clouds that soak the area downwind.

  Page 1Page 3
  Pair of satellite image showing the Atlanta metropolitan area and temperature on a summer day.

Second, cities may be disrupting the flow of air over the Earth’s surface. “If you blow air over a nice pristine wheat field in Kansas, that air at a very low level is going to blow in a straight line,” says Shepherd. But blow that same air over a city, and you get chaos, literally. “Cities tend to have buildings, particularly tall buildings, that cause the air to converge, or pile up,” says Shepherd. “Think of two trains colliding; their front ends go up. That’s convergence.” So the city’s buildings provide a source of lift to push warm, moist, surface air into the cooler air above it, where it can develop into rain clouds.


The roofs, concrete, and asphalt of a city absorb heat during mid-day, raising the surface temperature up to 10°C. This pair of satellite images shows Atlanta on September 28, 2000. In the true-color image (top), the urban areas are gray, and wooded suburbs and open fields are green. The map of surface temperature (bottom) shows the urban heat island. Yellow areas are relatively cool, while red areas are hot. (NASA images by Marit Jentoft-Nilsen, based on Landsat-7 data.)

  Photograph of the Atlanta skyline

The third hypothesis runs along the same lines, but instead of causing the air to pile up, the city divides the air. “Storms that are approaching Atlanta or Baltimore from the west basically split or ‘bifurcate’ around the cities because of the physical structure of the buildings or because of the thermodynamic environment [the urban heat island],” says Shepherd. When the two halves of the storm come back together downwind of the city, the air is pushed up like the two colliding trains. The rising air forms rain clouds.


The concentrations of buildings in urban landscapes are an obstacle to surface winds. Air is forced to go around or over the city center, creating a disturbance that leads to rain clouds forming downwind. (Photograph copyright adsullata.)


The Impact of Urban Pollution on Rain


City pollution may also impact cloud formation and rainfall. “Water vapor doesn’t ordinarily spontaneously condense into drops to form clouds,” says climate scientist Tom Bell, from NASA Goddard Space Flight Center. “It needs dirt to form around. All rain needs aerosols to form.” In the natural world, cloud-forming aerosols are things like sea salt, dust, and pollen, all of which are large particles. But pollution aerosols are usually smaller and more numerous than natural aerosols. With lots of particles to collect on, water coalesces into many tiny droplets instead of larger rain-sized drops. The impact on rain, says Bell, varies depending on where the clouds form. In some cases, urban aerosols suppress rain, but in others they increase it.

  Page 2Page 4
  Satellite map of cloud particle size showing aerosols reducing droplet size

What causes the difference? Scientists are still working on that question, but Bell and others believe that the temperature difference between the air near the ground and the atmosphere above may be one key difference. “Bubbles of hot air form over land and rise like a bubble in a pot of boiling water,” says Bell. The rising air cools, and many tiny droplets form around urban aerosols. These small drops are not large enough to fall, so the storm tends not to produce rain.

More rain can occur when the bubble of heated air forms over a very warm area, however. “The bubble of heated air rises faster and climbs higher in the atmosphere,” says Bell. Because of urban pollution, “droplets that would normally fall out at a lower elevation are smaller and go higher.” Now high in the atmosphere, the tiny water droplets turn to ice. To make the transition from liquid to solid, the water molecules have to release heat. “The added heat kicks in an ‘afterburner,’ and the bubble of air is pushed up higher and faster,” Bell explains.

“The extra boost makes the storm act like a vacuum cleaner,” says Bell. “Something needs to replace the rising air, so more moist air is sucked up.” This vacuum-cleaner effect allows the storm to pull in more material to work with than it would have without the urban aerosols, and more rain results. “The effect only happens when fast-rising air would form a thunderstorm anyway, and when the air near the surface is moist,” Bell adds.

Chances are, the urban heat island, city structures, and pollution all interact to alter rain storms around cities. “The real question is which combination of those is the most important factor,” says Shepherd. “Which of those takes the most precedence at the beginning? Is there one of those that takes more precedence at the end, once rainfall gets started? That’s what we’re trying to study right now.”

The key to discerning how aerosols, city structures, city heat, and weather systems interact, says Shepherd, may very well be NASA’s fleet of Earth-observing satellites. Taken together, the satellites are providing unprecedented information about land use, cloud structure, rainfall, and aerosols that can then be integrated into models to sort out the intricacies of urban rain. “We need these integrated observing systems like Aqua, CALIPSO, Cloudsat, TRMM, and eventually the Global Precipitation Measurement Mission, along with models to answer these integrated questions,” says Shepherd.

Satellites See City Rain

It was with these questions in mind—do cities influence rain and if so how—that Shepherd set out to see what NASA satellite data could contribute. The satellite data he had been looking at when he noticed the apparent urban rain anomaly in the first place were from a satellite called the Tropical Rainfall Measuring Mission (TRMM). Among its suite of instruments, TRMM (which scientists pronounce “trim”) carries the world’s only space-based precipitation radar. Like the radar that you see on the evening weather report, TRMM’s Precipitation Radar bounces radio waves through the atmosphere to measure rainfall. The difference is that all other radar systems are ground-based, and therefore have a limited range, whereas TRMM can observe everywhere between about 40 degrees north and south. When combined with GOES weather satellite and rain-gauge data, the rain data analysis can be extended to even higher latitudes. What’s more, because rainfall is measured with a single instrument, TRMM’s estimates in one city can be compared to estimates from another city, something that is difficult to do with two different ground-based systems.


Like urban pollution, the aerosols in ship exhaust change the properties of clouds. These satellite images show clouds over the North Atlantic streaked with ship tracks. The top image is similar to a digital photo, while the bottom image shows the size of cloud droplets from pink (smallest) to green (largest). Compared to droplets formed from naturally occurring aerosols like dust or sea-salt, droplets formed from pollution aerosols are smaller and more numerous. The ship tracks in the image contain droplets as small as 2 micrometers (millionths of a meter), while the clouds in the “background” have droplets that are closer to 20 micrometers. Urban pollution has a similar effect on clouds. (NASA images by Jacques Descloitres, MODIS Land Rapid Response Team, and Mark Gray, MODIS Atmosphere Science Team.)

  Map of rainfall based on data from the Tropical Rainfall Monitoring Mission (TRMM)

Pioneering though the instrument is, Shepherd feared that the rainfall data might not be detailed enough to measure the urban rain effect. Each pixel, or data point, that the Precipitation Radar detects measures four kilometers across, and the gridded (combined and mapped) data is even coarser. If the urban rain effect was small, TRMM might not pick up enough data points to detect elevated rainfall totals over and downwind of cities. “But if we could use that satellite data in conjunction with other data sets, then at least we could look at many different cities around the world if we wanted to,” says Shepherd. Spurred by that thought, Shepherd set out to take a careful look at the Precipitation Radar’s rainfall record.


The Tropical Rainfall Measuring Mission (TRMM) monitors rainfall 40° north and south of the equator, far enough north to examine cities in the southern United States. This map shows TRMM observations of average hourly rainfall rates in August 2006. Highest rainfall rates (greens and blues) occur across the Tropics. (NASA image by Robert Simmon, based on TRMM data.)

  Satellite image of rain over Texas from February 10, 1998

Working with colleagues Harold Pierce and Andrew Negri, both at NASA Goddard, Shepherd focused on five cities in the south-central and southeastern United States that were not near mountains, major rivers, or oceans—features that themselves impact rainfall. “We wanted to try to isolate that the urban environment was the only thing affecting the circulation,” he explains. He divided this area into a grid where each box measured 0.5 degrees by 0.5 degrees (very roughly a 55 by 55 kilometer square) and tabulated the average amount of rain that fell every hour in each box during the summer (May to September). He then averaged the rain rates over the three summers during which TRMM had collected data, 1998 to 2000, and identified where the most rain fell. He found that the amount of rain that fell per hour was as much as 20 percent greater in grids that were downwind of cities than it was in grids upwind of the city. These results were consistent with ground-based studies of the same regions. It seemed that the cities were generating rain and TRMM could measure the effect, but Shepherd wanted a little more confirmation. He set up a series of rain gauges in Atlanta and recorded how much rain fell around the city during a year. He compared those measurements to TRMM’s measurements. The ground and satellite measurements matched: urban rainfall was real, and TRMM could detect it.


TRMM’s Precipitation Radar maps rainfall in three dimensions along a narrow path underneath the satellite. The radar scans the storm in both the along-track (in the direction the satellite is moving, upper inset) and cross-track (across the width of the swath, lower inset). Other TRMM sensors provide visible and infrared data on the storm. Shepherd and his colleagues used these data to help understand the influence of cities on rainfall. (NASA image by Robert Simmon, based on TRMM data.)


Refining TRMM’s Picture of Urban Rain


“The pioneering work that Marshall did with the TRMM satellite data to analyze the signature of urban areas started a lot of people thinking of different ways that they can relate space-borne rainfall data to different urban signatures,” says Steve Burian, a hydrologist at the University of Utah. Burian collaborated with Shepherd to refine the picture of urban rain that TRMM had provided. When TRMM detected urban rain over a particular city, could they determine what mechanisms were triggering the rain? The question required both a model and satellite data. With the model, they could map out different scenarios—a sea breeze interacting with the urban heat island, an existing storm being split by the city, and so forth—and see how the rainfall that came out of those scenarios compared with the TRMM data. Shepherd and Burian first tackled Houston, Texas.

Along the Texas coast, a sea breeze carries moist air inland in the afternoons, and summer rain forms when the moist air piles up along the curves in the coastline carved out by various inlets. Previous studies had suggested that rain patterns in Houston were related to the shape of nearby Galveston Bay, but Shepherd reasoned that several other segments of the coast had a similar shape and should therefore display similar rainfall patterns. Repeating the gridding approach he used in his first studies, Shepherd broke the Texas coastline into seven chunks and analyzed TRMM rainfall data in each. If the coastal shape caused rainfall, then he should see the same patterns in all of the grids, but if the city of Houston was influencing rain, then there would be more rain in that grid. The grids containing Houston and the region downwind received the heaviest rain. “The sea breeze interacts with the urban circulation to create an anomaly,” concludes Shepherd.

In a related study, Burian and Shepherd also used historical rain gauge data to confirm that the percent occurrence of rainfall around Houston had increased in post-urban Houston as compared to pre-urbanized Houston. Burian and Shepherd also revealed that the city of Houston may have shifted the time of likely occurrence of storms until later in the afternoon.

  Page 3
Plot comparing summertime rain before and after the urbanization of Houston, Texas

The next step was to put these results into a model to see if they could duplicate the exact processes that created the urban rain effect and determine how much the city had influenced recorded rainfall. In the case of Houston, was the sea breeze interacting with the urban heat island, aerosols, or the city structures? If the city weren’t there, what would have happened?

“We’ve actually done experiments with the model where we’ve taken the city [Houston] out and tend not to get rainfall on a specific day. The city creates rainfall that wouldn’t have even been there had the city not been in place,” says Shepherd. When Shepherd and Burian ran the model with the current Houston “cityscape,” they found that the sea breeze was split either by Houston’s buildings or by the pool of hot air over the city.

“It’s like a pebble in a stream: the water flows around the pebble and converges on the other side,” says Shepherd. In the model, the onshore breeze encountered the city, split, and flowed back together downwind of the city, creating the convergence that generates storms—exactly where TRMM and rain gauges observed higher rain rates. Shepherd saw a similar pattern over Phoenix, Arizona. When he conducted an analysis of a 108-year rain-gauge dataset of rainfall patterns around Phoenix, Arizona, he found that the urban heat island circulation over the city may interact with air flowing down a nearby mountain to generate extra rain in a suburb northeast of the city. TRMM data also confirmed the presence of this anomaly. So the way a city influences rain may differ depending on its location and what other factors influence weather in the region.


The growth of Houston increased the total rainfall downwind of the city and shifted the timing toward late afternoon. Rain-gauge data from before Houston grew into a big city (red line) shows rainfall spread over a greater part of the day. After urbanization (blue line), more of the day’s rain was concentrated in a narrower time window that peaked around 4:00 p.m. (Image adapted from Burian and Shepherd, 2005.)

  Map of rainfall around Phoenix, Arizona

But location may not be the only factor influencing which mechanism triggers urban rain. “Smaller cities have a different effect than larger cities,” says Shepherd. “Working with a local Houston planning agency, we used a sophisticated urban growth model to make Houston much larger. With the 2025 projected surface area, a much larger heat island is generated, and we don’t see as much of that bifurcated flow around the city...You see more piling up of the air flow over the city. Cloud and rainfall concentrate over the urban areas as opposed to the downwind area.” The area most affected by city-induced rain may shift as a city grows.

To Shepherd, growth also means that the urban rain effect will likely become more pronounced in time. “We’re showing with the case of Houston that if you start proliferating more and more of the landscape with urbanized surfaces, you get a change in clouds and precipitation around that city.” Pointing to rapid urbanization—according to the United Nations Population Division, the percentage of people living in cities is expected to increase from 48.8 percent to 59.9 percent between 2005 and 2030—Shepherd continues, “Aggregate that to every city in the world, it undoubtedly has to have an effect on the overall global climate.”

But Shepherd doesn’t have to look far into the future to see a need for understanding the urban rain effect. “There are implications certainly for agriculture, urban planning, water resource management, but also for weather and climate forecasting. One of the things that we’re finding very clearly is that cities can affect rainfall, yet many of the weather forecast models that we rely on every day do not include urban land surfaces.” Understanding urban rain could improve the weather forecast for heavily populated regions, giving more accurate warning of potential flood-producing rains. And that information could save human life and property today.

  • Burian, S., and Shepherd, J. (2005). Effect of urbanization on the diurnal rainfall pattern in Houston. Hydrological Processes, 19, 1089-1103.
  • Jin, M., and Shepherd, J. (2005) Inclusion of urban landscape in a climate model: how can satellite data help? Bulletin of the American Meteorological Society, 86, 681-689.
  • Shepherd, J.M. (2005). A review of current investigations of urban-induced rainfall and recommendations for the future. Earth Interactions, 9 (12), 1-27.
  • Shepherd, J.M. (2006). Evidence of urban-induced precipitation variability in arid climate regimes. Journal of Arid Environments, 67 (4), 607-628.
  • Shepherd, J., and Burian, S. (2003). Detection of urban-induced rainfall anomalies in a major coastal city. Earth Interactions, 7 (4), 1-17.
  • Shepherd, J., Pierce, H., and Negri, A. (2002). Rainfall modification by major urban areas: observations from spaceborne rain radar on the TRMM satellite. Journal of Applied Meteorology, 41, 689-701.
  • United Nations Population Division. (2005) World Urbanization Prospects: The 2005 Revision. Accessed June 23, 2006.

Shepherd detected increased rainfall downwind of Phoenix, a rapidly growing city in arid Arizona. Rising air from Phoenix’s urban heat island may be interacting with nearby mountains to influence the summer monsoon. The map shows average hourly rain rates from July–September 2003. (Map adapted from Shepherd, 2005.)