In lieu of an abstract, here is a brief excerpt of the content:

15 Future Vision of the Highlands Richard G. Lathrop Jr., Mary L. Tyrrell, and Myrna Hall How does one describe the landscape of the Highlands? It depends on what time period you are considering: A mid-eighteenth-century frontier of largely unbroken forest. A mid-nineteenth-century industrial landscape dotted with miners’ villages and ironworks belching smoke 24/7 amidst a scrubby cutover forest littered with the mine tailings. An early-twentieth-century landscape of second-growth forests on the uplands, the valleys a quilted patchwork of farmland and water resources just beginning to be tapped. Or fast-forward to the late twentieth and early twenty-first centuries, when the rural landscape is undergoing a radical transformation as forestland is fragmented and farmland abandoned to sprawling suburban/exurban development. And what about fifty years into the future? Examination of the implications of future land-use change helps to inform the local and regional land-use planning process before ill-advised and irreversible land-use decisions occur. In the Highlands, the main pattern of land-use change is the conversion of forest and farmland to residential land composed mainly of owner-occupied, single-family detached houses (Lathrop , Hatfield, and Tulloch 2003; and Tyrrell et al. 2010). Land-use-change scenarios and models provide a valuable tool to investigate the process of change as well as potential future landscape configurations (Botequilha and Ahern 2002; Conway and Lathrop 2005). Land-Use-Change Modeling in the Highlands Two main approaches have been adopted in modeling future land-use change in the Highlands as part of the U.S. Forest Service–sponsored Highlands Regional Study: (1) a statistically based approach that determines the rate and spatial pattern of historical land-use conversion and extrapolates that rate into the future to map areas that have the highest likelihood for future development within a set time frame; and (2) a deterministic “build-out model” to map the form of the fully developed landscape while avoiding the complexity of predicting when the changes will occur. Both approaches are spa- Future Vision of the Highlands 317 tially explicit in that they use an extensive database of digital maps to model the constraints and influences posed by the biophysical, social, political, and regulatory environment to determine the location and scope of future development. Although slightly different in details, the statistically based modeling approaches used by the Connecticut–Pennsylvania studies versus the New York–New Jersey regional studies are broadly comparable in overall form. The Connecticut–Pennsylvania study (Tyrrell et al. 2010) employed the landuse change GEOMOD model (Hall et al. 1995a, b; Pontius et al. 2001), while the New York–New Jersey study employed an econometric logistic probability modeling technique (Lathrop, Hatfield, and Tulloch 2003). This empirical modeling approach is based on the assumption that spatially explicit factors that correlated with development in the past are good predictors of where development is likely to occur in the future. The model includes physical factors that may increase the cost of building or may make a site more or less desirable, such as slope, soil stability, sunlight, and scenic view. The factors may relate to the economics of site selection or to the regional infrastructure, such as distance from towns, roads, or rail stations. Breaking the landscape up into small grid cells, the model ranks each grid cell as to its suitability (i.e., likelihood) for development. The models can be further stratified to include administrative jurisdictions such as town, county, or region. Spatial location variables were found to be strongly related to the probability of development. As might be expected, close proximity to land that was already developed or to existing roads increased the probability of development in all four state models (Lathrop, Hatfield, and Tulloch 2003; and Tyrrell et al. 2010). Amenity values such as proximity to water or lakes were found to be important in New Jersey and Connecticut. Other physical factors such as farmland or low slopes or locational factors such as distance to train stations were also important. Where the models removed jurisdictional stratification and adopted a more regional approach, demographic factors tended to explain more of the variation in development patterns across the region than did physicoeconomic variables (Tyrrell et al. 2010). Demographic variables vary across the region and appear to attract or constrain new development accordingly. These demographic factors have less explanatory power at the town level—that is, when the analysis is stratified—because most of the demographic factors do not vary much within...

Share