Simulating London in 1940

Initial Aim

The project has been reset a little for 2023. Simulating all of London in 1940 is about five orders of magnitude greater than the current part-time project of a single person can manage. Thus the three towns within the original coverage area:

  • Maidstone
  • Canterbury, and,
  • Ashford,

    Have been selected for simulation with the Urban Sim. This requires a slightly extended JSON file format from Urban Sim and some modification of the existing ApeSDK code for urban versus wilderness simulation.

    Final Aim simulates the south east of England from Dunstable to Rayne to Goudhurst to Chiddingfold 50 miles by 70 miles of the area around London in 1940.


    The simulation is developed from the ApeSDK's urban simulation.


    Interpreted map data coming from the Ordinance Survey maps of 1940 from the National Library of Scotland which thankfully is free to use for this kind of work. It represents very varied map data including buildings, rail, road, waterways, and orchards.

    Creative Commons License
    All maps are licensed under a Creative Commons Attribution 4.0 International License.


    The conversion effort to change these PNG map files into JSON has been achieved thanks to the assistance of Bob Mottram. The important command line processing repositories is:

    london1940 Gitlab repository:

  • creates map2json which converts the PNG files to JSON files and contains the five towns of initial interest. Also contains all the automation testing around the JSON output files. In particular can they be quickly read by the ApeSDK.

  • urban - simulation includes drawout which takes the ApeSDK compatible JSON file and generates a map showing the details captured in the JSON from the original map PNG. Also mapedit for editing the JSON files with the original PNG images backing the editing.

  • contains the original full resolution Ordinance Survey maps for the area.

    The source is licensed under the ApeSDK license (a version of the BSD license (BSD-1-Clause + Joke)).

    We are very close to getting the data-set into an agent simulation and are looking for agent modellers who are interested in this problem space.

    This web page will serve as a location for temporal updates of information. All email correspondence can be directed to Tom Barbalet barbalet at gmail dot com.

    Thank you for your interest.