Building a transportation network and finding the shortest route
In a dedicated course on GIS for Transportation, we started by getting to know what network datasets are and how they work. In the first exercise, the task was to create, validate and present a multimodal network dataset for Västra Götaland in the southwest of Sweden. The modes to be implemented were car travel, walking, and cycling, as well as the usage of public transit.
A network dataset models a transportation network that is comprised of lines (also referred to as ‘edges’), points (also referred to as ‘nodes’) and turns. The connectivity of the network is configured with respect to the network dataset properties that account for costs (e.g. impedances of travel time; length, etc.), restrictions (e.g. no walking on motorways; oneway streets for cars, etc.), and policies (i.e. how features connect), among other things.
The most common network analysis is finding the shortest path between two points, which is what this exercise was about.
Throughout the process of modeling the network with rules such as costs and restrictions, it was important to regularly test the network to see whether it followed the rules correctly. In doing so to find the shortest route, I could simply digitize hypothetical start and endpoints and then run a route solver. A good way to do this was to see whether the one-way directionality was being respected. A roundabout served well to see if the direction of travel follows the restrictions correctly. See the illustrations below.
The last thing was to incorporate General Transit Feed Specification (GTFS) data into the network model. GTFS-based modeling makes use of public transit data such as schedules, locations of stops, lines, etc. in conjunction with walking. Converting one or more GTFS datasets to a set of feature classes and tables representing the transit stops, lines, and schedules can then be assigned the ‘Public Transit’ evaluator for the walking travel mode, meaning that the route solver will only consider the GTFS data if the mode of transport is by foot. In other words, the network dataset is being set up to model the pedestrian network to also include public transit connections.
The map below shows the final result of the shortest route between the arbitrarily chosen starting point (1) and the destination point (2) with respect to the rules of the network dataset.