A wide range of land use planning and modeling tools and methodologies have been applied in various locations across the country in recent years. Some of these tools focus on providing land use forecasts based on a consistent methodology grounded in social science and econometric methods. Other tools offer sophisticated GIS mapping and 3D visualization capabilities to allow planners to present land use information to the public and engage them in developing alternative scenarios. This presentation presents a new two level approach which attempts to integrate econometric land use modeling techniques with a GIS/3D visualization tool in a single system.
The Evansville, Indiana, Metropolitan Planning Organization (EMPO) leads a coalition which received a grant from HUD to develop a regional plan for sustainable development. As part of this planning effort, EMPO proposed to develop land use planning and modeling tools. They desired both the capability for easily developing multiple land use scenarios in collaboration with the public as well as the capability to show the public and decision makers the most probable land use which would develop given various policy scenarios, including no changes to the status quo.
After reviewing available tools and techniques, it was decided to develop a new two-tiered approach which would integrate an available GIS/visualization tool with parcel level detail with econometric forecasting techniques at a more aggregate zonal level. The existing GIS/visualization tool could be used by planners to develop and present land use scenarios by itself, or it could be run together with higher level spatial discrete choice models to predict the likely land use that would result from various zoning and infrastructure policies.
The two level system works essentially as a set of spatial nested logit models. The attractiveness of the lower level choice of parcel level building locations informs the upper level choice allocating population and employment growth among zones. The upper level model is applied analytically to the zones while the lower level is applied to the parcels using Monte Carlo simulation. This system results in both static, repeatable estimates of zonal population and employment as well as a probabilistic allocation of buildings at the parcel level which can then be visualized in 3D.