Creario Documentation

Learn how Creario prepares all the different datasets that make up a MATSim scenario.

The input data and processes for creating each part of a MATSim scenario are documented in the following chapters. The individual chapters go into many details; if you are more interested in a shorter abstract, read the summary below.

Summary

Network: The initial road and public transport networks in Creario are created from OpenStreetMap data, but are then enhanced with data from AWS Terrain Tiles (for node elevations and link gradients) and GHSL (for determining if a link is in a rural or urban context). Combining data from all these datasets allows estimating very detailed link capacities and speeds based on a link’s classification, curvature, gradient, and location. The generated multimodal networks are thus not only suited for the simulation of private car traffic, but also for bicycle trips and electric vehicles where gradients and the road surface have a higher impact on route search.

Public Transport: Public transport data in the form of GTFS datasets can be uploaded by users. Alternatively, Creario provides access to more than 1.500 GTFS feeds from all over the world and allows users to simply select those covering their model area.

Population and Demand: For the synthetic population, data from GHSL and WorldPop is used. The combination allows for a realistic spatial distribution of agents as well as reasonable age and gender attribute values. To improve the accuracy of the generated population, users can upload custom spatial data that describe the distribution of the population within the model area along with optional age and gender distributions. Demand (i.e. agent plans) is created based on data from National Household Travel Surveys. Currently, users can select between NHTS data from France and the USA, which are both available as open data. Additional surveys might be included in the future, allowing for richer and better adapted travel behavior in the generated models. Location choice for the activities in the agents’ plans is done based on facilities extracted from OpenStreetMap.

Limitations: The models are created by a fully automated pipeline based on datasets that provide a worldwide coverage. They will thus not be able to compete with manually created, hand-calibrated models. While the models might not be used for detailed transport planning without further refinement, they provide a good, fast and cost-effective base for such models. In addition, the created models are typically good enough for research projects that want to demonstrate certain effects applied to a specific region. To assist transport planners working with the generated models, each model comes with an extensive report that includes the results of numerous validation steps performed, highlighting potential problems or improvements that can be made to the models.