Automatic Generation of Levels of Detail : A Study on the Swedish National Road Database

Sammanfattning: When creating a map, the amount of data and which geographical information chosen to bepresented are decided based on the purpose of the map. Trafikverket is responsible for theNational Road Database that is a reference road network with a large amount of dataconnected to it. The database is built at a carriageway level where several links can be usedto represent a single road. In this thesis the database has been studied as well as thepossibilities to create an automatic workflow that creates a generalisation to road level. Someapplications and analyses that use this database are requesting input data of a higher level ofdetail. It is also found important from a cartographic point of view. Possibilities to create alinear referencing for the data between the different levels was studied as well, to make theupdates between the levels easier and not having the need to maintain two differentdatabases. It would make it possible to inherit attributes and the generic features connectedto the network.The first fully automated workflow for a generalisation of an entire map was developed in theNetherlands in 2014. Several other agencies in Sweden and Norway are currently working ongeneralisations as well and using different approaches. Statens Vegvesen, which is theNational Road Agency in Norway, has already implemented the different levels of detail intheir road database. Lantmäteriet currently has a project where their goal is to create a fullyautomated map to easily change to a larger scale.After studying the programs used in other generalisations and which programs Trafikverketnormally use, FME and ArcGIS were chosen to be used in this thesis. A small study area waschosen, and the proper data was collected from the National Road Database. The study wasconducted by a set of experiments and trying different parameter values in order to obtain asatisfactory result.Three different areas where studied, where the goal was to find a sequence of functions thatsuccessfully generalised all the areas. The areas were parallel roads, complex intersectionsand roundabouts. The result consists of a suggested workflow, but alterations hade to bemade manually since it was not possible to find a fully automated generalisation for the areachosen. Therefore, a set of functions and tools that could be developed were collected as wellin the end to make a more automated workflow possible.The result found in this thesis show that if the cartographic generalisations are more studiedand examined it should be possible to get a more automated workflow to create ageneralisation between the different level of detail. However, it might be difficult to get theworkflow fully automated due to the complexity of the intersections.