The need for more sophisticated simulations of urbanization has prompted the development of interdisciplinary land-use models. Integrated land-use models feature distinct yet related submodels requiring substantial data input representing the broad range of elements that drive land-use patterns. Further, the relationships between and among the submodels can be treated as dynamic with multiple pathways of influence. Dynamic models are linked to a database which manages and updates source data during model runs. While data constraints may force a static representation of certain parameters or processes, dynamic interactions between individual agents, grid cells, and model components more realistically simulate land-use changes. Like other system models, feedbacks can be positive or negative, lead to growth, decline, or stability, and establish steady states or devolve into chaos.
Although several large-scale integrated models have been developed for specific locations, a growing number are intentionally flexible and can be applied to multiple locations across a range of spatial scales. The use of open source software and programming languages for model development eliminates proprietary constraints on model and source code distribution while facilitating a cooperative development environment. This has greatly enhanced the ability of integrated land-use models in not only addressing problems specific to particular locations but also in the formation of modeling scenarios with the cooperation of stakeholders. Further, it allows the development of modular utilities that add functions based on planning needs (e.g., water allocation or air quality modules). However, as such models become increasingly complex, it is imperative that they be well documented so as to ensure maximum transparency for stakeholders who rely on the outputs. While it is beyond the scope of this article to provide detailed descriptions of an exhaustive list of land-use models, Table 2 characterizes several
Table 2 Land-use models and their characteristics8
Pautuxent landscape model
CUF&CUF II Clarke UGM SLEUTH
Model type: agent-based, CAs, statistical
Model description: dynamic, highly complex, modular, spatially explicit, open source Internet reference: http://www.urbansim.org
Model type: CAs
Model description: dynamic, intermediate complexity, interdisciplinary, combination of open-source and proprietary tools Internet reference: http://www.uvm.edu/giee/
Model type: agent-based, statistical
Model description: spatially explicit, intermediate complexity, proprietary Model type: CAs, statistical
Model description: dynamic, spatially explicit, low complexity, open source Model type: CAs, statistical
Model description: dynamic, spatially explicit, low complexity, modular, open source Internet reference: http://www.ncgia.ucsb.edu/projects/gig/
Model type: CAs, statistical
Model description: dynamic, spatially explicit, low to medium complexity, modular, proprietary Internet reference: http://www.clarklabs.org/
Model type: agent-based, statistical, gravity model Model description: spatially explicit, low complexity aThis list of land-use models is not intended to be exhaustive. Rather, the model characteristics are provided only for those land-use models referenced in the text.
land-use models according to the terminology presented in this paper.
One recent example of an integrated model is UrbanSim, an agent-based behavioral simulation model that allows for prediction of land-market responses to policy alternatives. UrbanSim predicts the spatial distribution of urban residential and commercial development based on its own models of household and employer behavior and through the integration of an external four-step travel-demand model. Predictions are based on statistically derived coefficients estimated from empirical data. It operates under dynamic disequilibrium in which supply-demand imbalances are addressed at each time step but are never fully satisfied, as they would be in a model assuming full equilibrium (e.g., DRAM/EMPAL, Disaggregated Residential Allocation Model/Employment Allocation Model). The disequilibrium approach accounts for process-specific timescales of change such as daily transport decisions (short-term process) or real estate development (long-term process).
UrbanSim can endogenize factors that other models take as exogenous, such as employment location or land values. Model features include the ability to simulate the mobility and location choices of households and businesses, developer choices for quantity, location, and type of development, fluxes and short-term imbalances in supply and demand at explicit locations, and housing price adjustments as a function of those imbalances. Because it is open source, other modular capabilities can be freely added based on model outputs.
Was this article helpful?