The Mobility Spectrum: A Data Driven Strategic Transport Model for the Whole of the Netherlands

In the Netherlands there is an increasing need for nation-wide insights on the present-day mobility of travellers. For mobility planning and as a starting point for regional transport models it is desired to combine nation-wide data on social economic data, infrastructure and empirical data on mode-, and destination- and route-choices of travellers. To this end DAT.Mobility and Goudappel Coffeng are developing the Mobility Spectrum: a data fusion application that describes the number of travelers per mode on any location in the Netherlands. The Mobility Spectrum will be expanded to also describe the number of travellers per mode for each day of the week and for each hour of the day. Allowing it to be used for tactical and operational mobility planning as well. For the Mobility Spectrum the social economic data is determined per address and all public roads and public transport lines in the country are incorporated. The model is completely data driven and based on open data. It is designed as an automated process from creating the network until the actual matrix fusion and assignment. When new data is available, an updated version of the model can be created. In this way it can be updated periodically when new infrastructure is realised or if the social economic data has been updated for the current time period. The end goal for the Mobility Spectrum is that it updates itself whenever a new version of a data source becomes available, but currently the authors aim at a bimonthly updating frequency. The Mobility Spectrum currently fuses data that describes infrastructural networks and socio-economic activity, with empirical data on mode-, and destination- and route-choices of travellers like modal splits, trip frequency distributions and distributions patterns. This data is derived from traditional nation-wide survey data (‘OViN’), mobile phone data (MPD), GPS-tracking-survey data (‘Dutch Mobility Panel’) and vehicle count data. The survey data consists of 35.000 respondents, the mobile phone data has 5.000.000 active users, with the GPS tracking survey 10.000 people are tracked 24 hours a day during the whole year and the vehicle count data contains 1.000 locations. New is that the Mobility Spectrum can fuse and parametrize any data source that relates to (aggregates of) mode-origin-destination combinations. The next step is to also fuse data sources like parking capacities and Automatic Number Plate Recognition (ANPR) data into the Mobility Spectrum. Combining different sources of observed data can lead to difficulties. For example, each data source has a different spatial detail. Furthermore, inconsistencies between data sources need to be detected and removed. Lastly, combining data sources often needs weighting and normalizing. To overcome these difficulties the Mobility Spectrum uses the multi-proportional gravity model as a data-fusion tool because: 1) when fed with consistent datasets- it is guaranteed to provide the most likely set of OD matrices (i.e.: the OD matrices that adhere to conditions of maximum entropy); 2) it allows for automatic detection of inconsistencies between data sources and/or observations; 3) it does not require to weigh or normalize different data sources; and 4) it allows for easy parametrization into a synthetic demand (gravity) model, that can be used for traditional strategic transport model applications (applying a pivot-point on the fused OD matrices). This paper will address the different data sources that are used. It will elaborate on how these data sources are used and what problems have been overcome. Furthermore, it will give an insight on the use of the multi-proportional gravity model and the possibilities to parametrize a multimodal strategic transport model from it.


  • English

Media Info

  • Media Type: Digital/other
  • Features: Bibliography; Figures;
  • Pagination: 12p
  • Monograph Title: European Transport Conference 2020

Subject/Index Terms

Filing Info

  • Accession Number: 01768542
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Feb 19 2021 2:41PM