The open black box: the role of the end‐user in GIS integration

We would like a hydro layer that doesn't compromise detail for completeness ... Being able to cross the federal boundaries with like structure is important for watershed approaches to things (T. Schindler, GIS analyst, Clallam County, Washington, personal communication, 12 November 1999). Introduction Complex and pressing environmental problems tend to expose the gaps in our scientific knowledge, our technologies and our theories. Attempts by local, tribal, state and federal government agencies in the US Pacific Northwest to respond quickly to the 1999 listing of nine salmon species under the Endangered Species Act (US Department of Commerce 1999) make it apparent that current spatial-data infrastructures (Chan et al. 2000) are neither readily available nor transparent to use (Star et al. 1997). The county GIS employee quoted above bemoans the lack of an infrastructure of hydrography data--that is, a consistent, publicly available layer of data depicting the region's network of rivers and streams. Such a network could be used to integrate other types of environmental information critical to salmon survival. Before local watershed-level data on salmon habitat and spawning conditions can be linked to the river and stream network, the watershed-level hydrography network itself must be built. Based on participant observation and interviews, this paper examines the day-to-day work practices of several local government employees in the Upper Cedar River watershed in Puget Sound, Washington, USA. These employees are mapping the rivers and streams in their watershed. Later, they will integrate these high-resolution hydrography data into a regional hydrography database that is being cooperatively assembled by a group of organizations in the Pacific Northwest. The participants in the Oregon/Washington Hydrography Framework Project have established a shared database of surface water features for the region using standards and data models that have been agreed upon through negotiation with a number of organizations. This database, which will be maintained through the Internet by local partners, will serve as a framework for the long-term sharing of data on stream conditions and salmon habitat across the region (Washington State Department of Natural Resources 1999; Regional Ecosystem Office 2001a). This effort builds on such recent developments as the US National Spatial Data Infrastructure (US Executive Office of the President 1994), the Global Spatial Data Infrastructure (Global Spatial Data Infrastructure 2001), and federated database movements in Europe (DeVogele et al. 1998). Long-term sharing of distributed databases by multiple organizations is only one context in which to examine integration issues in GIS. However, it is a context with a long history, stretching back to the early days of land-records integration in local government (National Research Council 1980, 1983; Chrisman and Niemann 1985; Nyerges 1989a), and it is a context that has become increasingly important with the emergence of spatial-data infrastructures in the 1990s. Spatial-data infrastructures are typically seen as a problem of coordination among organizations (Nedovic-Budic 1997; Rajabifard et al. 1999; Chan et al. 2000; Rajabifard et al. 2000). Recent studies have argued that to effectively provide large-scale data to meet the needs of local users, these infrastructures must be built from the local level upwards, rather than being planned from the top down, as has been the case with the national spatial data infrastructure in the United States (Harvey et al. 1999). This paper enlarges on the bottom-up perspective by arguing that the organizational approach is not nearly a fine enough sieve to catch many important details in the integration of local data into regional and national spatial-data infrastructures. Integration must be looked at from the perspective of the individual end-user and the local environment in which she works, because the process of integrating data provides critical feedback to the infrastructure itself. …