New process leads to smaller estimate for global urban area

December 16, 2009 By David Tenenbaum

Urbanization is one of the most important trends of the 21st century, yet we don’t have a good idea of how much land cities occupy. Different studies vary by as much as a factor of six, which confuses efforts to understand how urbanization is affecting farming, conservation, climate and essential “ecosystem services” like watersheds. More than half of the global population now lives in cities (according to recent United Nations estimates), making the impact and importance of urban areas greater than ever before.

In a presentation today (Dec. 16) to the American Geophysical Union fall meeting in San Francisco, Annemarie Schneider, an assistant professor at the Center for Sustainability and the Global Environment at the University of Wisconsin–Madison, described results based on 2001 data from the Moderate Resolution Imaging Spectroradiometer (MODIS) on the Terra and Aqua satellites.

Cities, defined as places dominated by the built environment (including all non-vegetated, human-constructed elements, such as buildings, roads and runways), covered 657,000 square miles in 2001. That estimate is up to six times smaller than other calculations, Schneider says. Still, based on recent trends, she expects the area of urban land to double between 2000 and 2050, and notes that 90 percent of population growth during the next 30 years is expected to occur in cities.

Schneider uses seven wavelengths of visible and infrared light from the MODIS instruments. “For the global scale of this study, we can’t just rely on spectral data,” Schneider says. “Instead, we use temporal information to pull out the urban signal. We have 46 snapshots from MODIS over the years, and we can use these to look at how vegetation changes from season to season.”

Vegetation reflects strongly in the near infrared, Schneider says, but she is less interested in pure vegetation than in the mix of vegetation and impervious, highly reflective surfaces, such as roofs and pavement. “Every city has tree cover for shade and parks, and we can capture this mixed signal with satellite data. The core of the city — downtown areas that may be only 10-15 percent of a city’s land footprint — are the only locations that are fully impervious or ‘bare.’ The rest of the city has a unique mixed vegetation-urban signal that is quite distinct from crops and forests outside the city.”

The study starts with detailed images from Google Earth, where each pixel records roughly 16 square meters of land surface, and then jumps in two steps to the 0.25 square kilometer pixels recorded by MODIS. “This is a ground-up, not a top-down approach,” Schneider says. “We start with the highest resolution data available, so we can be certain that the samples we collect are truly built up or urbanized land.”

Schneider, with Mark Friedl at Boston University and David Potere at Princeton University, then search the MODIS data for patterns resembling those seen in the views with more details, and generate a map of urban areas around the world.

A key issue is encroachment on farmland, says Schneider. “Cities are almost always located on our most productive agricultural lands, and as a result, rapid or unplanned urban growth quickly eats up some of our most precious natural resources. Although cities are expanding, I don’t think we are ‘paving over the world.’ With a global map of built-up lands, we can look at where cities may threaten our best farmland in different parts of the world and start to plan accordingly.”

A second topic involves global warming, Schneider says. The United Nations’ Intergovernmental Panel on Climate Change is developing new global scenarios for economic and population growth, on which a wide range of climate change analyses and modeling will be based.

“Tracking historic and future changes in urban growth will be an important component of future IPCC scenarios, so that growth must be accurately assessed,” Schneider says. “We can do a better job of understanding where urban areas are now and where they may be in coming decades. If we understand how cities expand, we can start estimating the impacts of urbanization at regional levels, and provide information for policymakers who want to influence city growth in a more sustainable direction.”