This paper develops a spatio-temporal data model to support activitybased transport demand modelling in a GIS environment. This so-called mobilityoriented spatio-temporal data model conceptualizes the spatial and temporal interaction of travel and activity behaviour using the concept of mobility. In other words, activity patterns are conceptualized as a sequence of staying at or travelling between activity locations. The model can support the analysis and queries of activities from diOE erent

more »

... rspectives, i.e. queries can be location-based, time-based, and person-based. It can also support activity-based modelling by identifying spatial and temporal opportunities for activity participation. The conceptual and logical designs of the data model are presented. A prototype system based on the data model is implemented in ArcView. The prototype is illustrated and tested by a case study based in Hong Kong. Activity-based transport demand modelling Since the late 1950s, transportation models have played an important role in forecasting transport demand and evaluating the impacts of plans and policies. Planners use transportation models to learn about the behaviour of transport systems. Over the past decades, the development of particular modelling approaches has closely followed planning needs, which in turn are closely related to dominant policy issues. The rst generation of transportation models was developed during the late 1950s and early 1960s. Their purpose was to facilitate the prediction of future transport demand, such that road capacity programs could be based on predicted demand. The rst generation models are commonly referred to as four-step models. These models are typically formulated and calibrated at the level of the tra c zones. Individuals are aggregated by tra c zone. Tra c is considered to be the result of four sequential decisions: trip generation, trip distribution, mode split and tra c assignment. These decisions are modelled separately at successive stages. Although the four-step models have been widely used, even institutionalized in literally thousands of applications (Stopher et al. 1996) , the major shortcoming of these largescale, aggregate and supply-oriented models is their lack of behavioural content. Consequently, researchers could not evaluate alternative policies such as tra c demand management measures that are unrelated to investment proposals for major facilities, because the evaluation of these policies requires the prediction of individual behavioural responses. Consequently, in the 1970s and 1980s, this aggregate approach was gradually replaced by a disaggregate modelling approach. The disaggregate approach focuses on decision-making and choice processes at the level of individuals or households. This shift was stimulated by a policy change from long-term supply strategies to short-term market-oriented tra c management initiatives. The major premise underlying this more decentralized and demand-oriented approach is that individual decision-making mediates the impact of transport policies on travel patterns. If the impact on an individual's behaviour can be predicted, the eOE ects of transport policies can be easily derived by aggregating across individuals. The disaggregate approach was developed from random utility theory. Individual travel decision-making is represented by discrete choice models. Mode and destination choices are the two types of travel decisions that have witnessed most applications. Although the theoretical underpinnings of the disaggregate approach were certainly stronger than those underlying the aggregate approach, the disaggregate approach also did not escape from major criticisms. Like the aggregate approach, the disaggregate approach is a trip-based approach which decomposes travel behaviour into a set of simple problems related to single trips, ignoring the interdependencies between individual trips. More importantly, the approach treats travel as a demand in its own right, avoiding the question why people travel. It has become increasingly evident, however, that travel choices fundamentally depend on choices to participate