Transport accessibility, a component of the more general concept of accessibility, captures how easy or difficult it is to travel to different locations at different times by different means of transport. Alternatively, it can be viewed as the capability of the transport system to provide fast, inexpensive or otherwise valuable means of transport between locations.
As such, transport accessibility is one of key concepts in transport and urban planning. The analysis of transport accessibility serves both evaluating the performance of transport systems and proposing measures for improving it.
Accurate and detailed analysis of transport accessibility is, however, a computationally challenging problem requiring large amounts of data and efficient algorithms. Because of the lack of data and high computation costs, traditional approaches thus work with highly simplified representations of the transport system and omit many factors (e.g. service connectivity at transfer points and/or walking segments at the beginning and the end of a trip) that can have a crucial impact on the accessibility subjectively perceived by the travellers. As such, they only provide limited insights into the real-world performance of transport systems and, consequently, can only support transport system optimization to a limited degree.
Fortunately, with the wide-spread adoption of standard data formats for describing transport systems (in particular OpenStreeMaps and General Transit Feed Specification), the lack-of-data issue has greatly diminished. That is why in the Intelligent Transport and Logistics group I and my students have developed novel techniques for efficient fine-grained analysis of urban transport accessibility. Leveraging our extensive experience in multimodal trip planning, our approach works with a full-detail representation of the transport system using exact timetables and complete maps of road, footpath and cycleway network. Because of the fine-grained, detailed model of the transport system, our accessibility analysis calculates a more comprehensive and accurate transport accessibility metrics. For example, when analysing public transport accessibility, our method takes into account pavement network as well as constraints on number of transfers and maximum walking distance.
Transport Accessibility Metrics Definition
Our method calculates seven primary accessibility metrics, some of which are common for modes while others are mode-specific:
- Public transport: Besides travel time, which is considered as the most important metrics and the only one typically used for determining the quality of public transport, we compute two additional primary metrics – the number of transfers and the frequency of connections.
- Car: For car journeys, we define and calculate three primary metrics – travel time, travel distance and fuel consumption.
- Bicycle: For bike travel, two additional metrics besides travel duration and travel distance are supported – elevation gain and physical effort. Elevation gain is the minimum amount of altitude that must be ascended to reach the destination (such a route can e.g. go around a hill so it need not be the shortest). The physical effort corresponds to the amount of energy consumed along the least energy-expensive route.
All of the above accessibility metrics can be calculated at three levels of aggregation: location-to-location accessibility forms the basis of more aggregate location analysis which in turn forms the basis of the most aggregate area analysis.
Transport Accessibility Metrics Calculation
Unsurprisingly, the calculation of fine-grained accessibility metrics is computationally costly. In fact, calculating all location-to-location metrics corresponds to executing tens of thousands of journey plan searches.
Fortunately, we have managed to significantly reduce such high computational requirements by developing a novel accessibility analysis technique based a modified Dijsktra’s algorithm extended to work on contextual graph views for time-dependent graphs. The Dijkstra’s algorithm is used because of its ability to efficiently compute shortest paths from the analysis origin node to every other node in the transport graph. This way, the calculation of all location-to-location metrics between an analysis origin and all other destinations can be done using just several runs of the modified Dijkstra’s algorithm configured to use the respective evaluation metrics as the search optimization criteria.
We have integrated our accessibility analysis algorithms into several practical tools. For interactive transport accessibility analysis, we have developed a web graphical user interface (see Figure 1). In the interactive frontend, results of accessibility analysis are shown as a heatmap, with colours corresponding to the accessibility level for the selected accessibility metric. Upon hovering over another location on the map, an infobox is shown with information about the value of the location-to-location metric between the location and the selected analysis origin (denoted by the red placemark). The frontend allows the user to specify additional analysis settings, such in particular the analysis time for public transport accessibility analysis. The computation of location metrics on the servers takes about one second and it takes approximately two seconds to visualize the analysis results in the user frontend. The interactive frontend is publicly available at http://transportanalyser.com.
In addition to providing an interactive frontend for location accessibility analysis, we have also employed our method to calculate and visualize location accessibility for all location in the city of Prague. Such an analysis is computationally more demanding but provides an accurate global picture of transport system coverage and performance. Importantly, it can be also used to evaluate the impact of planned changes in the transport system design, such as evaluating proposed changes to public transport services routes and timetables. An example result of such analysis is shown in Figure 2.
Conclusions and Outlook
Fine-grained analysis of multimodal transport accessibility enables gaining deep insights into the performance of transport systems. The analysis enables systematic approaches to the optimization of transport system structure and operation, including the design of public transport timetables. Until a few years ago, such a fine-grained analysis was difficult to perform due to lack of data and high computational costs. This situation has now changed and this post provides a brief description of an approach that can be efficiently used for fine-grained analysis of transport accessibility. More details about our approach can be found in our recent paper Efficient fine-grained analysis of urban transport accessibility.