Testing microsimulation uncertainty of the parcel-based space development module of the Baltimore PECAS Demo Model

Abstract

A precise and stable microsimulation space development module is fundamental for supporting various policy decision-making exercises related to land development. This paper studies the dynamics or uncertainty of outputs of the parcel-based space development module of an integrated land-use and transport forecasting model—the Baltimore PECAS Demo Model. It is tested with two sub-studies: (1) running the model three times over the entire planning window from 2000 to 2030; and (2) running the model 30 times just one year ahead from 2000 to 2001. The outputs obtained are used to analyze such dynamics or uncertainty. Study results from the first sub-study show that, in general, the system is stable and consistent over runs and time, as supported by a set of paired t-tests. However, the coefficient of variation (COV) measuring the variation of estimated space quantity by category over four cross-section years indicates that the differences among runs are increasing over time through the planning window. The COV test over the second sub-study indicates the estimated space quantity is stable for most of the zones, except for a small portion of zones with a small space quantity.