4th International Conference on Integrating GIS and Environmental Modeling (GIS/EM4):
Problems, Prospects and Research Needs. Banff, Alberta, Canada, September 2 - 8, 2000.

A Comparison of Databases for Modeling Rural Land Use Change

GIS/EM4 No. 76

David M. Theobald
N. Thompson Hobbs

Abstract

Accommodating rapid growth of the human population while sustaining natural systems has emerged as a fundamental challenge confronting society in the United States and the world. For example, there is concern throughout the nation about effects of urban sprawl and development on agricultural and wildlife habitat. Recent statistics from the Natural Resource Conservation Service show that the rate of conversion of farmland and other open space to development has more than doubled in recent years (1992-1997). Consequently, there is an urgent need to develop ways to forecast changes in land use, to evaluate alternative planning actions that seek to alter development patterns, and to examine the consequences of those patterns for natural and social systems. Although urban sprawl is much on the minds of urban planners, it is particularly important to address changes in land use across the full urban to rural gradient. The growth in the service economy and progress in telecommunication technology has fueled rapid development of rural areas, often at low-densities. These changes in the economy and work patterns have caused increased urbanization and a shift towards low-density, "rural sprawl." As a result, agricultural land is increasingly developed. Understanding the patterns of rural sprawl and identifying ways to cope with it are an important issue for local and national planning. To respond to this issue, we have developed a model that forecasts land use change in both urban and rural areas. We use nationally available data sets (from the US Census Bureau) that can also be refined using finer-scale local data (parcels). We have developed a simulation environment that allows alternative land use planning actions, both regulatory and incentive-based, to be played out into the future. This allows the suite of alternative approaches to managing sprawl, including re-zoning, transfer of development rights, and conservation easements, to be examined. We then can evaluate the environmental consequences of those changing land use patterns, for example on critical wildlife habitat and important agricultural areas. We will demonstrate the model and interactive simulation environment for the Estes Park, Colorado area, a gateway to the Rocky Mountain National Park and discuss how the model could be used in other areas experiencing rapid growth and development.

Keywords

Forecast modeling, alternative scenarios, land use planning, rural sprawl

Introduction

We are challenged as a society to both accommodate growth and to conserve our natural resources. There is widespread concern at the local and national level over urban sprawl and development consuming agricultural and wildlife habitat. Recent statistics from the Natural Resource Conservation Service show that the rate of conversion of farmland and other open space to development has more than doubled from 1992-1997. Consequently, a foremost research need is to develop ways to understand past land use change, to forecast changes in land use, to evaluate alternative planning actions that seek to alter development patterns, and to examine the consequences of those patterns for natural and social systems. While research activities (NCGIA 1997) and the popular press have focused on urban sprawl, it is paramount to address changes in land use across the full urban to rural gradient. The growth in the service economy and progress in telecommunication technology has fueled rapid development of rural areas, often at low-densities. These changes in the economy and work patterns have caused increased urban sprawl and a shift towards low-density, "rural sprawl." Indeed, understanding the patterns of rural sprawl and their social and ecological consequences is an important local and national challenge.

Objectives

The central question we address here is how useful are commonly-available datasets for modeling rural sprawl? In particular, our objectives are to:

1. illustrate the need for examining the full urban-to-rural gradient of development by computing area of urban development in each map; and
2. compare data sets that are the foundation for modeling efforts.

Methods and Analysis

There are a number of data sets that are commonly used to quantify historical land use change and that serve as a basis for modeling future land use change. For broad-scale land use/cover changes, the US Census of Agriculture and USDA?s National Resource Inventory data are cited frequently. These data are collected at the county-level, however, and are not appropriate for spatially-explicit modeling efforts.

Remotely-sensed imagery

There are a wide variety of datasets that are derived from either satellite imagery or aerial photography. A useful source of data is the North American Land Characterization project, which provides NDVI-classified imagery for three dates (1970s, 1980s, and 1990s) at 60 m resolution (NALC 2000). Land cover data also can be classified directly from Landsat TM data (e.g., image above). Also, development patterns can be derived from an analysis of fine-scale aerial photos.

USGS LULC

This data set provides land use/cover data classified by the Anderson land use codes at 1:250,000 scale (some states also have 1:100,000). Most maps depict the late 1970s or early 1980s (see left). Occasionally, land use/cover change has been mapped at a finer-scale and for multiple dates. For example, the goal of the Colorado Front Range Resources Infrastructure project is to develop historical data for 1930s, 1950s, 1970s, and 1990s to determine trends and patterns of development, and to serve as a basis to forecast future development (USGS 2000).

US Bureau of the Census

Housing and population densities can be mapped at the census tract, block-group, or block level to map development patterns. These data can be further refined using dasymetric mapping techniques to eliminate public and undevelopable lands (Theobald 2000). The resulting maps depict the full urban to rural gradient, at decadal increments (from 1940).

County assessor data

County assessor databases provide the basis for property tax collection. Though many metropolitan counties have mapped the parcel boundaries as well, most counties have not (e.g., in Colorado only 25% have a complete parcel map). However, parcel-level data can be summarized on their parcel ids and then geo-referenced using the Public Land Survey System (a.k.a. township/range/section). This type of data is useful and formed the basis for the forecast model in Theobald and Hobbs (1998).

Again, the resulting maps depict the full urban to rural gradient, at yearly increments. If available, parcel-level maps provide the finest-scale of land use data available. These data provide the opportunity for many types of modeling, such as modeling urban land use change or to conduct a build-out analysis (Theobald and Hobbs, in review) (left - 1975, center - 1995, right - build-out ~2040).

Conclusion

The amount of urban land identified in the Landsat TM data is substantially underestimated compared to the parcel dataset. The developed area identified in the LULC data is close to the urban and suburban areas in the parcel data, but misses exurban development. The block-level data likely overestimates the extent of development. Future research is needed to investigate whether these results are consistent in other areas. Efforts to quantify the effects of development and forecast future development need to incorporate low-density development patterns.

Acknowledgements

This work was supported by a grant from the US Environmental Protection Agency's Science To Achieve Results (STAR) program (#R827449-01-0), Great Outdoors Colorado, and the Colorado Division of Wildlife. Although the research described in the article has been funded wholly or in part by the U.S. Environmental Protection Agency's STAR program, it has not been subjected to any EPA review and therefore does not necessarily reflect the views of the Agency, and no official endorsement should be inferred.

References

NALC 2000. North American Land Cover dataset. http://edcwww.cr.usgs.gov/glis/hyper/guide/nalc.

NCGIA 1997. The Land Use Modeling Workshop. EROS Data Center, Sioux Falls, SD, June 5-6. http://www.ncgia.ucsb.edu/conf/landuse97/.

Theobald, D.M. 2000. Fragmentation by inholdings and exurban development. In Forest fragmentation in the central Rocky Mountains, R.L. Knight, F.W. Smith, S.W. Buskirk, W.H. Romme, and W.L. Baker, editors. University Press of Colorado: Boulder, Colorado. Pgs. 155-174.

Theobald, D.M. and N. T. Hobbs (in preparation). Using build-out scenarios to evaluate land use planning alternatives: Protecting biodiversity on private land.

USGS 2000. Colorado Front Range Resources Infrastructure project. http://rockyweb.cr.usgs.gov/frontrange/index.html.

Authors

David M. Theobald, Scientist, Natural Resource Ecology Lab
Colorado State University, Fort Collins, Colorado, USA 80523.
Email:davet@nrel.colostate.edu , Tel: +1-970-491-5122, Fax: +1-970-491-1965.

N. Thompson Hobbs, Research Scientist, Natural Resource Ecology Lab
Colorado State University, Fort Collins, Colorado, USA 80523.
Email:nthobbs@nrel.colostate.edu.