Fast Computation of Continental-Sized Isochrones

Abstract

GIScience 2016 Short Paper Proceedings Fast Computation of Continental-Sized Isochrones Paolo Bolzoni 1 , Sven Helmer 1 , Oded Lachish 2 Faculty of Computer Science, Free University of Bozen-Bolzano, 39100 Bolzano, Italy Email: {firstname.lastname}@unibz.it Dept. of Computer Science and Information Systems, Birkbeck, University of London, London WC1E 7HX, United Kingdom Email: oded@dcs.bbk.ac.uk Abstract We propose an approach to speed up the computation of isochrones, which are maps showing the reachability of locations given a starting point and a time constraint. The core idea of our technique is to materialize large parts of an isochrone, demonstrating how this can be achieved for multi-modal transport networks in a scalable way. We illustrate the effectiveness of our method with the help of an experimental evaluation. 1. Introduction Isochrone maps, which given a starting location show the reachability of places within a certain time span, have been around for more than a hundred years. Before the introduction of computer sys- tems and the digitization of map data, creating these artifacts was a very time-consuming task. The easy availability of geographical information systems (GIS), such as PostGIS, and map data, such as OpenStreetMap, has sparked a renewed interest in isochrones. The applications of isochrone maps are manifold. For example, in urban and regional planning they can be used to determine the location of public services, such as hospitals, schools, police and fire stations, making sure that catchment areas cover an adequate part of the population. When planning new transport links, isochrones can help in identifying zones that are not well-connected. Establishing evacuation routes for emergency situations can be facilitated as well. In real estate applications potential buyers and tenants can check which ac- commodations are within easy reach of workplaces and schools. Finally, planning trips on the fly to quickly determine which places are reachable in an acceptable time frame is also made easier. Currently, there are no isochrone algorithms that scale to very large and detailed maps. Our goal is to compute isochrones very efficiently, in the ideal case in real-time. While there are efficient al- gorithms for route planning, most of them are uni-modal, i.e., only one mode of transportation, e.g. by car, is used. Usually, people change their mode of transportation a few times when moving from one location to another, though. There are only a few sophisticated algorithms for multi-modal route planning, e.g. Bast et al. (2015), but they only compute shortest paths from point to point. While a lot of the early work on algorithms for computing isochrones, and also some of the more recent work, assumes that the underlying network is fairly homogeneous, Gamper et al. (2012) specifically investigate isochrones for multi-modal transportation networks. These approaches are implemented on top of database systems utilizing geographical information system features. On the one hand, this makes it easier to implement the computation of isochrones, as some general-purpose database system functionality can be re-used. On the other hand, not even geographical information systems support or are optimized for the direct computation of isochrones, so there is still a lot of code outside of the database system that needs to be written. Our technique offers a highly scalable approach for computing isochrones on multi-modal trans- portation networks efficiently. In summary, we make the following contributions: we develop a data structure that precomputes and materializes a large part of an isochrone map; as a result, our algorithm can assemble large parts of the answer by sequentially scanning the data structure and in the case of a