Dynamic Models
of Land Use Change

Developing Tools, Techniques, and Talents for Effective Conservation Action

The Yale Global Institute of Sustainable Forestry and the State University of New York College of Environmental Science and Forestry (SUNY-ESF) have recently released a study, finished in 2004, entitled “Dynamic Models of Land Use Change in Northeastern USA: Developing Tools, Techniques, and Talents for Effective Conservation Action.” With funding from the U.S. Forest Service, this project aimed to assist community leaders, conservation organizations, and citizens in understanding and predicting land use change dynamics, in particular fragmentation and loss of forests.

Forests in the northeast are increasingly under threat from urban sprawl and other unplanned development. Nationwide, eighty percent of new housing construction is occurring on the rural landscape. The challenge is increasingly one of keeping forests as forests—preventing their permanent loss to development or from being fragmented into such small parcels that sustainable management is no longer feasible. Our research focused on detecting the patterns and trajectories of the cumulative changes causing forest fragmentation and permanent loss of forestland in the region.

We are studying two areas in the eastern United States: the Thames River Watershed in Connecticut and Massachusetts; and the Catskill/Delaware water supply watersheds in New York (see map). These largely forested places are under tremendous pressure from development and the sprawling metropolitan areas of New York City, Boston, Hartford and Providence. As the largest unfiltered surface water supply in the country, the New York City Watershed is extremely vulnerable to potential changes in land use. Protecting the remaining forested landscape is a high priority for both the local communities and the urban population of New York City. The Thames River Watershed, in northeastern Connecticut and south-central Massachusetts, known as the “Last Green Valley” between New York and Boston, is home to the Quinebaug-Shetucket National Heritage Corridor, the Norcross Wildlife Sanctuary and the Yale Myers Forest. The smallest towns in this valley have experienced an average population growth of 17% from 1990 to 2000.

At the New York study location, we found that topography, roads, and development were the most significant predictors of future development. We also found that parcelized land was 1.5 times more likely to see fragmentation than non-parcelized land, but forest loss was high throughout the study area. Forest cover dropped from 87% to 79% in the Catskill/Delaware region between 1992 and 2001, and forests in 2011 are predicted to be twice as fragmented there as they were in 1992. In the Thames Watershed, distance from agricultural lands, soil type, and distance from urban areas were the most significant drivers of forest loss. Conversion of forestland was high in this area as well, with forest cover dropping from 79% in 1985 to 74% in 2002. The towns with the least amount of forest cover were also found to be losing the greatest amount of forest each year.

The core of the project is the application of a dynamic land use simulation model, developed by Systems Ecologist Charles Hall and graduate students at SUNY ESF, to the problem of loss and fragmentation of forestland due to unplanned growth and sprawl. This model, GEOMOD, predicts the rate and spatial pattern of land conversion, particularly that which is anthropogenically-derived. GEOMOD is extraordinarily effective in helping people understand the dynamics of land use change, see where forests are most at risk of fragmentation and conversion to development, visualize future conditions, and plan strategic approaches to the mitigation of harmful trends. Knowing how, where, and why those changes are likely to occur can be a powerful tool for conservation organizations, community leaders, and citizens.

In partnership with local and regional conservation organizations, we gathered commonly available geographic, economic and social data on factors believed to influence land use change. We statistically analyzed these drivers to find out which are the best predictors of land use change in the study areas. We then used GEOMOD to analyze changes in land use classification data, determining a background rate of conversion from forest cover to non-forest cover. That rate was validated with data from 2001 and was used to predict development until 2011.

GEOMOD will be publicly available in the next release of Clark Labs’ Idrisi GIS/imaging processing system. Although there are many tools now being used by the conservation and land use planning communities, we believe that dynamic simulation, with its ability to portray the importance of cumulative effects, change over time, and driving forces, will be an enormous enhancement to the static GIS mapping or build-out analyses currently in use. The study report “Dynamic Models of Land Use Change in Northeastern USA” was released in 2004 and distributed to the communities in the study areas. The report is available as a PDF file or in hard copy from the GISF office.

Download PDF file:

• complete report (15.9 MB, high speed connections only)
• front cover
• 1st section - introduction
• 2nd section - Catskill/Delaware Results
• 3rd section - Thames Results
• 4th section - Reference and Appendices A and B
• 5th section - Appendices C and D ( MB)

GEOMOD will be publicly available in the next release of Clark Labs’ Idrisi GIS/imaging processing system. With this project, we will create the methodology for using GEOMOD to as a land use planning tool, and test its predictive power when incorporating socio-economic factors into the model. Although there are many tools now being used by the conservation and land use planning communities, we believe that dynamic simulation, with its ability to portray the importance of cumulative effects, change over time, and driving forces, will be an enormous enhancement to the static GIS mapping or build-out analyses currently in use.

While the initial benefit will accrue to the two study sites, the project will have far-reaching effect through the development and testing of new and improved analytic tools and techniques that can be published and distributed for both teaching and application, as well as providing real-world training for students, government agencies, and conservation organizations that will benefit future efforts. Part of the research design will be to test for commonalities among the project study areas in forces driving land-use conversion from forest to development. If such commonalities are found, the results will have broad implications for conservation strategies throughout the country.

We are currently beginning the next phase of this project, which will attempt to predict water quality from land use changes in the Catskill/Delaware study region. With funding from the NY State DEC’s New York City Watershed Protection Program, we hope to improve the predictions from the initial parcelization study and also examine ways in which forest conversion may negatively impact water quality in the New York City Watershed.