The objectives of this research are to:
identify the significant factors influencing land use conversion from agriculture, forests, and natural land use states to residential use using historical data from the Rocky River watershed study area;
) predict land use pattern changes at a regional scale under different policy scenarios using the estimated parameters from the land use conversion model developed in Objective 1; and
develop the model from Objective 2 as a planning tool and educational resource for informing the decision-making process regarding growth management and environmental impacts.

Spatially explicit models that link policies to predictions of land use changes are necessary for evaluating the costs and benefits associated with existing and proposed policies and for guiding watershed planning on a regional scale. In addition, such models have the ability to yield predictions of spatially explicit land use changes over time and at a variety of spatial scales, which is a necessary component of integrated economic-ecologic models that link policies, land use changes, and changes in environmental functioning at different spatial scales. By establishing these linkages, the environmental and economic trade-offs of policies that impact water quality, development patterns, and economic growth can start to be elucidated. While this research will focus only on developing a model that links policies and land use changes, it represents an important step towards developing an integrated economic-ecologic regional model. In addition, results from this model can be used to inform planners and policymakers of the likely changes in future land use patterns resulting from specific policies. For example, the model could generate predictions of how land use patterns may change if zoning regulations were changed in a region or if public sewer were extended or withheld from a particular area. Such predictions can aid policymakers in assessing the current and future cumulative impacts of development within an area. Potential users have been identified and include: Ohio Coastal Management Program, Ohio Environmental Protection Agency, Medina County Soil and Water Conservation District, Greater Cleveland Growth Association, Cuyahoga County Planning Commission, and others. In addition, several Sea Grant agents will work with the researcher to help guide the policy applications of the model and extend the results to local officials and organizations.

The research proposed here is comprised of a two-step modeling approach. First, historical land use change data at the parcel level will be compiled and used to estimate a discrete choice model of land use conversion in which the probability of a parcel=s conversion is treated as a function of parcel-level characteristics and policy variables. Second, estimates from this model will be used in a spatial simulation model that predicts changes in future land use patterns under the current policy regime and alternative policy scenarios. This is done by simulating the model over multiple Arounds@ of development. For each round, all Adevelopable@ parcels are identified, their landscape attributes calculated, and the estimated parameters from the discrete choice model used to calculate each parcel=s probability of conversion. Probabilistic measures of conversion are translated into actual conversion by picking off the highest probability parcel in each round and the model is then simulated in this way for multiple rounds for each policy scenario. Model predictions for each policy scenario can then be shown either in map format with snapshots of the predicted changes through time or as a dynamic simulation file on the computer.

The first two phases of this project have been completed:

Phase One: Describe historical pattern of land use change in study area (Medina County). A variety of spatial landscape measures were used to quantify different aspects of the change in the spatial pattern of the following land uses from 1956-96 in Medina County: residential, commercial, industrial, and undeveloped (including agriculture and forests). Results from the application of several different landscape measures show that the location of residential development has been the primary cause of a fragmented land use pattern. Specifically, we find that (1) on average, forested and agricultural areas have become progressively smaller in size over time, while the number of separate "patches" of forest and agriculture has increased; (2) larger undeveloped patches of land have become fragmented into smaller ones specifically due to intervening residential development; and (3) residential development has become less clustered over time, while commercial and industrial development have become more so.

Phase Two: Statistical model of residential land use change. Results from a discrete-choice model of residential land use conversion show that areas of low-density population are more likely to experience residential growth, but also that residential development is less likely to occur in completely rural areas. This implies that people prefer to live closer to existing urban areas, presumably because of the services and accessibility that these locations offer, but at the same time dislike higher density areas that may be characterized by congestion and other negative effects of urbanization. These opposing positive and negative effects of urban areas explain why residential development has spread out, rather than become more clustered, over time. We also find that the minimum size that a lot is zoned has a significant effect on the likelihood that it is converted to a residential land use. Specifically, land that is zoned with a minimum lot size of three acres or greater is less likely to be converted. This implies that larger minimum lot sizes have increased the dispersed pattern of residential development.

Phase 3: Policy scenarios. Results from the statistical analysis of residential land use change were used to predict changes in land use patterns under different zoning regulations. The purpose of this was to perform 'what if' policy scenarios to to examine how residential land use patterns are predicted to change under different zoning policies. In particular, we were interested in testing the hypothesis that zoning implememted at a local level (e.g. township) could have spillover effects on development in neighboring counties. We found that an increase in the minimum lot zoning of one township deflected development from that township to other surrounding townships. As a result, a more scattered or sprawled pattern of residential development emerged. We concluded that lack of coordination of zoning policies among local jurisdictions can contribute to a more sprawled pattern of development at a regional level.