Upon receiving the satellite data, Imhoff realized there was one big problem they’d have to overcome." The raw image overestimates urbanized areas by as much as seven or eight times," he says. The problem came mostly from the effects of the relatively bright city lights on the satellites’ sensor array. The sensors on these satellites are made up of photoelectric cells organized into in a grid-like pattern, like pixels on a computer monitor. When light emanating from the Earth’s surface hits one pixel on the array, it is registered in the satellite data as a 2.7-km by 2.7-km square area of well-lit land surface.

Yet sometimes, bright lights trigger the initial pixel and inadvertently set off neighboring pixels in the array as well. If this happens, then an area the size of a city block will appear to be the size of three or four square blocks on the raw satellite image. It’s somewhat similar to what happens when a flash photo is taken of a mirror. Though the flashbulb itself may not be more than a couple of inches across, the light from the flash reflecting off the mirror would likely cover an area the size of a person’s head on the photograph.

To correct for this "blooming" effect, the Goddard team zoomed in on the lights emanating from individual cities, effectively isolating them from the larger, continental image. Using computers, they then lowered the overall brightness levels of the city image. The blob of lights representative of the given metropolis would begin to shrink on the outside in a manner similar to an evaporating puddle of water." We scale back on the brightness levels of the imaging data, until the perimeter stops shrinking on the outside and the interior lights of the city begin to break up," Imhoff says. "At that point we stop."

The researchers classified the lights left on the image, after this dimming process, as urban area. The previously lit areas on the image that shrank back were classified as peri-urban (low-density suburban areas or farmland). Any areas that had no lights to begin with were labeled as non-urban. They compared these classifications to the boundaries on the actual urban areas of the city and found there was a close match. Imhoff and his team now had a set of numbers (threshold values), which told them to what extent the lights from any portion of the United States should be dimmed to get an accurate and spatially explicit representation of urbanization.

Using the threshold values, Imhoff’s group categorized the entire continental United States into urban, peri-urban, and non-urban areas. To make sure the classifications were correct on a nationwide basis, they checked each state on their map against the 1990 U.S. Census population statistics. Imhoff explains the Census Bureau doesn’t map urban areas. However, it does classify urban areas as any region where there are 1000 people or more per square mile, and it takes a tally of who lives in these areas. By merging the city urban map with the Census data, the researchers could calculate population density for the urban, peri-urban, and non-urban lands. They found that the number of people per square mile on his map measured up to the Census’s definition of an urban area (1000 people and up per square mile). "After the thresholding we had an almost perfect match, which is amazing since we didn’t use any Census data to create the satellite map," Imhoff said. "We thought this is good; this is working."

For peri-urban areas Imhoff found there were roughly 100 people per square mile, and in non-urban areas, roughly 10 people per square mile. While the Goddard team’s map still couldn’t give them the exact density of the population for these classifications, it presented them with a picture of where the landscape had been transformed to the point where it no longer resembled the natural ecosystem. The researchers could be fairly certain that any area classified as urban on their maps had at least a few subdivisions, strip malls, and parking lots.

Imhoff and his group could now overlay this map of the United States with other maps showing where the best soils are, where fragile ecosystems exist, and where plant life is the most robust. With such comparisons, the NASA scientists could determine exactly how urbanization is affecting our planet, our natural resources, and even our climate. By repeating the entire process for other countries, they could get an idea of what was happening all over the world.

In the coming weeks in the Earth Observatory, we will be bringing you the results Imhoff and his team obtained by comparing his urbanization map of the United States with a map of the most fertile soils in the nation. We will follow this with an article on how urban sprawl may be contributing to the greenhouse effect in the Northern Hemisphere.

References
U.S. Census Bureau, 1995: Urban and Rural Population 1900-1990, Suitland, MD.

World Resources Institute, 1996: World Resources 1996-97, Washington, DC.

 Seeing the Light