Integration of Geographic Information Systems and Agent-Based Models of Land Use: Prospects and Challenges

Cellular Landscapes. Many ABM/LUCC researchers begin by developing models that operate over an abstract landscape, even when their longer-term goals include application to real-world landscapes. Roles for abstract ABM/LUCC models are discussed by Parker and colleagues (Parker, Berger, and Manson 2002 a,b; Parker, Manson, and Berger 2002; Parker et al. 2003), and the advantages of a progression from abstract to empirical are discussed by Berger and Parker (2002). Balmann and colleagues (Balmann 1997; Balmann et al. 2003) have developed spatially explicit models of structural change in agriculture. The Repast-based (repast.sourceforge.net) SLUDGE model (Parker and Meretsky 2004) examines the relationship between negative spatial externalities, landscape fragmentation, and economic efficiency. The Swarm-based (www.swarm.org) FEARLUS model explores the evolution of cropping patterns and land holdings in an agrarian setting (Gotts, Polhill, and Law 2003). GIS Initialization /Output Display. Many currently developed ABM/LUCC models input either abstract or real-world raster landscapes and export output to files suitable for viewing in raster GIS. Berger (2001) models the influence of information networks and resource markets on adoption of new efficient irrigation technologies in Chile. The LUCIM model (Evans and Kelley 2004) focuses on the relationship between land-owner decision making, landscape heterogeneity, and the process of deforestation and reforestation. Landscape statistics similar to those produced by FRAGSTATS (McGarigal and Marks 1994) are calculated for the raster-based model output. The Repast-based LUCITA model (Deadman et al. forthcoming) focuses on the behavior of heterogeneous agricultural land owners in the Brazilian Amazon. The Swarmand