The Behavior of Uncertainty in Transportation and Land-use Interaction

Abstract

Assessment of uncertainty in urban travel demand reinforces resilient and sustainable planning of transportation infrastructure. As transportation and land-use are believed to be co-determinant, the behavior of uncertainty when transportation and land-use models speak to each other is examined in this work. Firstly, A basic integrated framework is developed for commuting travel demand and residential location choice based on Traditional Four-Step Transportation Model and Bid-Rent utility-based model. Then, Monte Carlo and Latin Hypercube sampling-based approaches are used to examine the behavior of uncertainty across the sub-models in the integrated framework. The empirical analysis considered data sets for Boston region in Massachusetts, reduction of estimation error is observed as transportation and location choice models interact with point estimates. Moreover, uncertainty is incorporated exogenously in the starting instance as normal distributions in the input values and parameters; the point estimated values are used as mean values of the distributions. Then, the Probability Density Function of the output of each step is carried forward to the next step as an endogenously propagated uncertainty. The submodels are executed iteratively and the behavior of uncertainty is reported. Probability Density Functions of the output are generally found to deviate from the normal distribution, an assumption of approximation to lognormal distribution is considered. Uncertainty in the sub-models' output is found to correspond to the uncertainty in the previous iterations. Several limitations are reported with respect to models selection, data aggregation and assumptions, and assumptions for simplification purposes.

Date of Award	May 2017
Original language	American English
Supervisor	Abdulla Galadari (Supervisor)