Origin–Destination Models

Definition

The OD model can be defined as a way to represent the flow of travelers between their origin and destination for any specific transportation mode or region. The major difficulty of OD modeling is obtaining accurate estimation to represent these flows in origin-to-destination tables. Thus, the OD model is the basis of trip distribution, destination choice, or zonal interchange analysis, which is the second step in the four-step transportation forecasting model. Information on trip distribution is presented in a two-dimensional square OD matrix denoting volume of demand flows. Depending on the specific research purpose, trip distribution of an OD table can be represented by numbers of trips, numbers of passengers or freight volumes, or numbers of traffic counts.

Data

To estimate OD demand is a challenging and costly task because it requires a large amount of OD-specific data. Data can be obtained in different ways, ranging from a large-sample OD survey and traffic monitoring system to physical road counters. The monitoring system, sometimes also called an automatic vehicle identification system, collects OD information of vehicles through two or more cameras placed separately at certain distances, or by local street sensors or monitoring cell phone signal (CPS) movement and distribution.

The OD information of each vehicle can be identified through vehicle plate recognition. In addition, vehicle information such as travel time and distance can be automatically calculated based on the information recorded by cameras or cell phones. In recent decades, because of the development of Intelligent Transportation System (ITS), traffic data can be obtained in a much more timely and efficient matter. Traffic counts can be observed at very short time intervals by traffic monitoring systems. This information can then be used for time-dependent OD modeling.

Despite the use of both surveys and the advanced traffic monitoring system, which provide solutions for OD data collection, the financial constraints of using a sampling survey and traffic monitoring system still prohibit collecting data for the whole population. In order to complete the OD matrix of the population, some mathematic algorithms are developed to estimate missing data.

Methods

Classic transportation forecast modeling is conducted in four steps: trip generation, trip distribution, mode choice, and traffic assignment. The OD model is used at the second step to generate information on trip distribution in the form of trip tables.

The OD table is the central goal of OD modeling because it displays the distributions of trips going from origins to destinations. The conceptual framework of an OD table is illustrated in Table 1. The table is an n × n matrix. The row of the table represents place of origin, while the column represents place of destination. Each cell in the table indicates [Page 1011]the numbers of trips or passengers from origin i to destination j. The task of OD modeling is to fill in the cells for the tables headed 1 to n.

...