Validation techniques for region-level microscopic mass evacuation traffic simulations

Document Type

Article

Publication Date

12-1-2011

Abstract

The experiences of several recent evacuations have demonstrated how a mass evacuation of a major city can affect traffic throughout an entire region. This realization has brought the need for analyzing and evaluating evacuation plans at a regional level. Numerous recent studies have devoted themselves to the topic of simulating large-scale evacuations. However, few studies have developed procedures for the validation of large-scale models. This paper discusses validation within the context of the recent development of the regional multimodal evacuation model for New Orleans, Louisiana. The New Orleans model is unique because it is among the first ever to incorporate qualitative and quantitative model validation procedures based on field data collected during an actual mass evacuation. The paper discusses the various statistics considered for validation, including their inherent advantages and disadvantages. It also presents the results obtained from the validation exercises of the New Orleans model. The study concluded that regression analyses were the most appropriate for statistically analyzing the spatial and temporal data correlations between the traffic patterns produced within the simulation and those actually observed during the Hurricane Katrina evacuation. From a qualitative standpoint, colorized spatiotemporal maps were also found to be quite effective for visualizing traffic speed and volume patterns. The maps were also invaluable for quickly identifying and analyzing bottleneck areas at both the local and regional levels.

Publication Source (Journal or Book title)

Transportation Research Record

First Page

66

Last Page

74

This document is currently not available here.

Share

COinS