Transit automated data collection (ADC) systems have allowed estimation of valuable behavioral patterns, especially for multimodal transit and with consideration of the sequence of passenger trips. Mainly, the interplay with data from the ADC systems-automatic fare collection (AFC), automated vehicle location (AVL), and other geographic information systems-provides more access to individual passenger's trip chain beyond that imagined at a more-aggregate level. Studies to identify passengers' trip sequences have been expanded to include multimodal transit networks as well as much larger networks

No abstract

Advances in microsimulation approaches to modeling of urban environments have happened rather independently in three streams of research; namely, land use, travel demand, and network supply. For land use modeling, microsimulation approaches are applied to model the urban form in a region, including the land use choices of individuals, businesses, governments, and developers. Households within a region make choices about their residential location, whereas individuals within a household make choices about their fixed activity locations, including workplace location, school location, and college location (while accounting for intrahousehold interactions and constraints). Businesses make choices about locating their offices and other related facilities. Developers make decisions about development (on empty parcels of land) or redevelopment (on parcels of land with existing facilities). These land use choices, along with the sociodemographic and economic evolutionary processes, government regulations, and zoning policies, make up the urban form of a region (3-5).In the travel demand arena, the field has experienced an increasing use of activity-based microsimulation approaches to travel demand modeling and forecasting. Activity-based approaches explicitly recognize the fact that individuals travel to fulfill their need to engage in activities. The primary output from an activity-based travel demand model is the activity-travel patterns of individuals within a household along a continuous time axis (6). The model system comprises various submodels that closely interact with each other to generate household activity agendas, individual activity schedules, activity linkages, trip chaining, destination, and mode choices, subject to the different household interactions (including interactions among household members) and temporal, spatial, and monetary constraints (7). A rich body of literature describes various implementations of activity-based model systems (8). These model systems differ from each other by the underlying behavioral paradigms that they use to represent activity-travel decision-making behavior and by the various degrees to which they represent choice processes (9).Network assignment is typically the last step in any transport model. Conventional assignment methods do not recognize that transportation networks evolve continuously through the day, and the underlying assumption of static network conditions in many assignment models in practice leads to results that are unlikely to be representative of actual network conditions. With microsimulation models of travel demand now capable of generating demand at a fine temporal resolution (e.g., 1 min), interest in the deployment of dynamic traffic assignment models that explicitly account for network dynamics along a continuous time axis and that thus allow the accurate representation of people's path choices and resulting network conditions is increasing (10). Microsimulation approaches to land use and transport modeling allow realistic representations of the...

than 24 or 48 h, as in the case of a hurricane evacuation). When the time to evacuate is considerably less, people may need to be evacuated directly from their current locations. The timing and magnitude of certain types of disasters may require faster reactions. People may receive short notice about the emergency and their need to evacuate, providing little or no time to return home before evacuating. In addition, for some disasters, the spatial extent of the evacuated area may change over time. This problem may be exacerbated by congestion around the evacuated area. For example, to manage traffic flows effectively in the event of a short-notice disaster such as a wildfire or flash flood, it may be useful to know which areas must be evacuated first and the likely impact on roadway congestion throughout the region.Several studies have been conducted on estimating demand characteristics for an evacuation, particularly focusing on the factors of trip generation, trip distribution, and trip timing. Some research has focused on estimating a trip generation model, such as the logistic model and the neural network model, using hurricane survey data (1-5). These models estimate a probability of leaving an area as a function of time as the disaster progresses. Other models estimate evacuation demand using socioeconomic data (6, 7). Similar to these last two studies, the present study considers the case without existing evacuation survey data for estimating a model of trip generation; existing trip generation data from existing urban planning models are used; such is the common situation faced by most regional transportation and emergency management agencies. In addition, models used for trip distribution for evacuation include the gravity model, the intervening opportunity model, and the multinomial logit model. Finally, a type of logistic model has been used for vehicle loading, describing the time of departure (6-8).Methods of estimating demand for a short-notice evacuation have not been discussed at length in the existing literature. In this paper, an estimation process is proposed for a short-notice evacuation. The method uses "on-hand" data typically generated through existing travel demand models at many metropolitan planning organizations. In this context, the proposed trip generation and distribution models are developed using existing trip matrices based on existing, calibrated travel demand models. It generally assumes that no separate set of behavioral data is available for estimating evacuation demand. Trip matrices are suggested for use as the basis for generating trip distribution; this trip distribution approach was inspired by Southworth (7). Additionally, a time-dependent evacuee departure model is proposed that considers a multiple-zone evacuation strategy related to that of Tweedie et al. (8). Approach to Modeling Demand and Supply for a Short-Notice EvacuationHyunsoo Noh, Yi-Chang Chiu, Hong Zheng, Mark Hickman, and Pitu Mirchandani 91 As part of disaster mitigation and evacuation planning, planners ...