Assessing and integrating uncertainty into land-use forecasting

Abstract

Uncertainty in land use and transportation modeling has received increasing attention in the past few years. However, methods for quantifying uncertainty in such models are usually developed in an academic environment and in most cases do not reach users of official forecasts, such as planners and policymakers. In this paper, we describe the practical application of a methodology called Bayesian melding and its integration into the land-use forecast published by the Puget Sound Regional Council, a metropolitan planning organization. The method allows practitioners to assess uncertainty about forecasted quantities, such as households, population, and jobs, for each geographic unit. Users are provided with probability intervals around forecasts, which add value to model validation, scenario comparison, and external review and comment procedures. Practical issues such as how many runs to use or assessing uncertainty for aggregated regions are also discussed.

Keywords

• Uncertainty
• Land Use Forecast
• Bayesian Melding
• UrbanSim
• Agent-based Models
• PSRC