The need to pay attention to intrazonal trips in macroscopic travel demand modeling has long been recognized but the representation of all trip ends within a zone by a centroid limits the determination of intrazonal interactions. Methods such as the nearest-neighbour technique and functions related to zone sizes have been the usual practice of estimating intrazonal trip attributes but these are generally proxies that do not give accurate estimates of intrazonal attributes. Recent modeling practices tend to use small zones to obviate the need to estimate intrazonal trip attributes. However, for many areas, structural data such as demographic and socioeconomic information are usually available at levels of spatial resoultions coarser than these sizes.
In this paper, an approach is presented that adapts node centrality measures to the most consistent way of defining intrazonal distances where trip and distance data are disaggregated within each zone and the average trip distance is computed. Nodes are selected to represent trip ends within each zone and are assigned weights based on their centrality. The average travel distance within each zone is computed from the weighted paths where the weight of each path is the product of the weights of the origin and destination nodes.
Using the 102-zone network model of the city of Dachau Germany, the model zones are aggregated into 51, 25, 12, 6, 3 and 1-zone systems with different combinations of zones. For each zone system and combination, intrazonal travel distances are computed and compared across the following scenarios.
i. Valid Case: Using the zone centroids
ii. Proposed Approach: Using the nodes (unweighted, weighted by degree centrality, weighted by closeness centrality)
iii. Existing Methods: the nearest neighbor-technique, zone-size function
The results show that the proposed approach, whiles consistent with the estimation of interzonal distances, provides better estimates of intrazonal distances than the existing methods. Though the paper presents the estimation of intrazonal travel distances for car travel, the proposed approach can be used to compute other intrazonal travel attributes like travel times. The approach can also be adapted for the estimation of intrazonal attributes for public transport, walk and bicycle travel.