Causes and consequences of land use change in the north carolina piedmont: the scope of uncertainty
The Triangle Landscape Change Project is an on-going effort at regional assessment centered on the Triangle region of North Carolina, a region framed by the cities of Raleigh, Durham, and Chapel Hill. Like many regions of the eastern United States and elsewhere, the Triangle has an agricultural and industrial past, while its current status is defined by high-tech industries of Research Triangle Park, three major universities, and a growing retirement community. The Triangle is one of the fastestgrowing regions in the United States, with some portions experiencing 30-50% population growth in the 1990-2000 decade (Triangle J Council of Governments, public comm.). As a case study for the patterns and consequences of land use change, the Triangle is compelling because its period of explosive growth is rather recent and thus coincides with the period of record of satellite imagery. The availability of imagery is augmented by the inclusion of Duke Forest as a NASA SuperSite; specialized imagery flown for the Forest also encompasses much of the larger region. In addition, a wealth of ancillary ground-based data are available (including the Duke Forest data archives, with monitoring data originating in the 1930?s), and so there is a rich geospatial data infrastructure to support large-scale studies of landscape pattern and landscape change. The Triangle Landscape Change Project embraces a set of related research themes under the umbrella of land use/land cover change (Figure 13.1). Land use pattern provides a framework and template in which we are studying various consequences of changing landscape pattern. These themes include forest dynamics, forest bird communities, and watershed impacts. These themes are coupled in that forests affect watershed hydrology via transpiring and intercepting leaf area as well as via protective ground cover typically associated with intact forests. Forests also provide the template of bird habitat, in terms of forest composition and structure, while land cover provides a larger context via edge effects on nesting success and potential dispersal limitations for habitat patches isolated by human land uses. Coincidentally, forest bird communities are coupled to watershed impacts indirectly because many forests are preserved as riparian buffers and these buffers represent a significant amount of forest habitat for wildlife. We believe that the Triangle Landscape Change Project is typical of many large-scale programs in integrated assessment, which increasingly rely on a shared geospatial data infrastructure and various models to interpolate field data and extrapolate the assessment to the regional scale. Two implications of this approach are that the projects within the larger program tend to be loosely coupled (i.e., studies done by people with different objectives), and that there is no single model that represents the program (i.e., there may be several models). Often, the end-users of the models are not the people who developed the models initially, as illustrated by the increasingly common use of institutionalized models such as Century (Parton et al. 1987), BASINS (US EPA 2001), and other models. These aspects of integrated studies pose some challenges when we attempt to account for uncertainty in the model projections. Our goal in this chapter is to explore issues related to uncertainty encountered when attempting to conduct integrated, regional-scale assessments using coupled models. Specifically, we will (1) describe sources of uncertainty in scaling to regional applications with increasing reliance on remotely sensed data, and illustrate how these sources of uncertainty are often "lost in translation" in loosely coupled applications in integrated assessments; and (2) explore methods for propagating these sources of uncertainty and communicating this information to a client audience of fellow scientists as well as city and regional planners. We will focus on the wood thrush (Hylocichla mustelina) as a case study and illustration. The wood thrush is an example of an area-sensitive forest bird species, a species that seems to prefer intact woods and is sensitive to nest predation and brood parasitism by cowbirds (Molothrus ater) (Brittingham and Temple 1983, Roth and Johnson 1993, Hoover et al. 1995). Thrushes have recently been exhibiting a regional decline in Triangle-area Breeding Bird Censuses and are consequently a species of some concern (Sauer et al. 2002). For our present purposes, the goal of forecasting regional patterns of abundance of the wood thrush is appealing because this represents quite a stretch for our data (indeed, perhaps the worst possible case), and thus introduces a number of issues related to uncertainty and error propagation in integrated assessment. © 2006 Springer.