Classification of Urban Areas for Travel Forecasting

The common practice of classifying cities into large, medium, and small population sizes, while appealing at first sight, is not adequate for travel forecasting. Two large cities with similar population sizes, for example, may turn out to have drastically different urban structures, thus generating totally different travel patterns. A more refined classification scheme involving both urban size and structure is therefore discussed in this paper. Person-trip-hours-of-travel—as determined by the product of trip frequency and trip duration—is used as a classification criterion, since trip frequency measures the rate of trip-making while trip duration indicates the length of the journey. According to this scheme, a city is either identified as large/multinucleated, large/core-concentrated, medium/multinucleated, or medium/core-concentrated. The validity of such a taxonomy is tested and proven by the data compiled from over 50% of the cities in the United States.