Measuring, Predicting, and Understanding the Impacts of Landscape Transformations

In many ways planet Earth faces the future millennium as an urban planet. Between 1990 and 2025, population in urban areas will double to 5 billion people. The increase in urban population for this period will be greater than the planet’s entire population 50 years ago. This population explosion will be concentrated over limited, but growing, territory, with unknown consequences. Without an understanding of the local consequences of the urbanization process, there can be no hope of constructing a sustainable urban future.

Despite decades of satellite and computer mapping, and even despite the new era of global observation of the planet, basic information on the global geographic footprint of urbanization and its change over time is lacking. City growth is documented statistically with census records, but these don’t measure and map the geographic extent of growth. Work by the U.S. Geological Survey (USGS) on a group of cities in the United States has shown how urban growth can be mapped over hundreds of years. It is mapped by the consistent interpretation of a multitude of source materials, including satellite and aircraft imagery, historic maps, and even written descriptions and landscape paintings. Databases from these sources form the basis for making future projections and assessing how urbanization affects the landscape.

Two of the United States’ largest metropolitan regions, San Francisco-San Jose-Sacramento and Washington-Baltimore (each with current populations of approximately 8 million), have been mapped over 140- and 200-year time periods. Current work is extending this mapping to additional metropolitan areas, including Greater New York City, Philadelphia, Chicago, Portland, Albuquerque, and Denver. These regions are mapped at intervals corresponding to available sources and historical events to form temporal geographic databases consisting of the themes of urban extent, major transportation routes, and water features.

These temporal geographic databases documenting landscape transformations in urban areas can be visualized as a sequence of maps or as an animation. Sequential maps show urbanization as a static pattern that grows with each period mapped. Animation reveals the temporal dynamics, showing patterns and trends that cannot be discerned from tabular or graphical data or sequential maps.

These databases are essential for calibrating a cellular automaton simulation model to predict urban growth. Topographic relief and lands protected from urbanization (parks and greenbelts) were added to the database. Behavior rules for the model considered spontaneous new growth, diffusive growth and spread of a new growth center, organic growth, and road-influenced growth. Factors controlling the behavior of the system were calibrated with the temporal geographic database and were then self-modifying during model runs. Researchers used a Monte Carlo randomization approach in making future urban extent predictions for each region by running the model a thousand or more times. They combined the results of all model runs and calculated the probabilities for each year of each grid cell being urbanized.

Resulting maps showed probability levels of urban extent 50 or more years from now. Animations were used to transform the past seamlessly into the future. Possible urban growth was mapped in color intensity that was proportional to its probability for each year on the basis of the model results. The urban growth model, all databases, and model results are available at no charge over the World Wide Web on the Internet for use in geographic information systems (http://urban.wr.usgs.gov/urban.html). The model can be run in a loosely coupled fashion with any GIS.

The preliminary model results indicate a continuation of trends noted by Gottman and others, with merging urbanized areas dominating extensive landscapes. In California, the cities that have already spread around the margins of San Francisco Bay (San Francisco, Oakland, and San Jose) are being linked to Sacramento, Stockton, and other growing cities in the fertile Central Valley of California. Accelerated urbanization of this important world agricultural region has global consequences. In Washington–Baltimore (a part of Gottman’s Megalopolis) continued urbanization is joining those two urban systems with new nuclei that are growing on their periphery (Frederick, and Harford County Maryland, and much of northern Virginia).

Concurrent with the accelerated growth in both of these regions during the last 50 years has been the creation of an urban form that is less dense than in the past. This is attributed to the spread and popularity of single-family housing and low-density employment and retail centers made possible by reliance on automobile transportation. As data from other United States cities are compiled, more quantitative statistics and comparisons of these factors will be made. As these trends spread worldwide because of rising living standards, tracking their impact on surrounding environments will be important to track.

In future work, the USGS hopes to use the same model in a non-North American context to examine the differences in urbanization worldwide. Extensions of the techniques will be necessary to accommodate the more diverse data sources that will be available, to adopt uniform standards for defining urban land use, and to consistently classify mass settlement on the peripheries of large cities in developing nations.

Work is progressing at implementing the strategy at a lower resolution to map recent urban growth and project future urban growth in the United States. A design for extending the methodology globally is anticipated.

Applications of the temporal databases and model projections are quite diverse. In every case, the fact that data are available over decades makes it possible to measure changes in land uses and estimate their impacts. As an example, habitat loss and fragmentation can be measured over time and compared to inventories of species richness. Planners can measure loss of agricultural land, forestland, and wetlands for the past, project it for the future, and estimate the impact that loss of these lands has for the production of food and fiber. They can apply estimates of energy and resource use to urban land to measure the land’s ecological footprint. This would be a first step in conceptualizing the challenges that we face in building a global sustainable world. Indeed, the most important set of applications are those that challenge policymakers to use data on temporal land use change and the urban growth model to anticipate the impacts of current trends and then examine how alternative policies may alter those initial projections.