In the last two decades there has been a steady rise in the gathering and use of 3D geo-information. A common way to store and use 3D data is by using 3D city models. In 3D city models, geo-information can be stored at different levels of detail. CityGML, the most commonly used data model and encoding for 3D city models, uses five levels of detail in order to model increasing geometric and semantic complexity. These levels of detail may be interpreted as a model quality measure, and as a guideline for users that need data for a certain application. CityGML consists of several thematic modules, each with their own level of detail specification. Some of these modules have a more further developed level of detail specification than others. Recently, several authors have proposed improvements to the Transportation module. This has led to proposals for various changes, especially concerning road data. However, these proposed changes have not been encoded yet. The main critiques are the lack of a level of detail specification for linear representations of roads, no ability to model networks, no representation of intersections and general ambiguity in the level of detail specification. Many road data use cases might potentially benefit from improvements on these points. In this thesis I attempt to improve the current level of detail specification of roads in the CityGML data model. The improvements are encoded in CityJSON, a JSON encoding of this data model. I assess the shortcomings in the current CityGML transportation module. After, a road data needs analysis is performed on three use cases: transport modelling, navigation and road maintenance. The data needs are compared to modelling approaches of other road data standards. This has resulted in several encoded improvements. A topological structure has been added to CityJSON. This includes the addition of two new modelling classes: Nodes and Edges. This structure is general such that it can be used by other thematic modules as well. Moreover, the level of detail specification for roads has been further developed to include both the linear representations and less ambiguous areal representations. This includes a prescription on how to model intersections and roundabouts at different levels of detail. Finally, I provide a structure which enables one to link the linear and areal road data together. This link is made at the scale level of the object, which data providers may choose themselves. This way one object can be modelled in two representation types without needing a one-to-one mapping between linear road segments and areal triangulated surfaces. These concepts are then tested by creating a CityJSON road data file for all new levels of detail.

Vario-scale is a new mapping technique which automatically generalizes maps from a baselayer of faces. Applications of vario-scale are continuous, smooth zoom in web maps,multi-scale representation in one map and being able to generate maps at arbitrary scale. Also,this would only require having to maintain the dataset at the highest scale level, since all otherscales are derived from it.Potentially, vario-scale could be an alternative for current web maps and generalizationalgorithms. The Dutch national mapping agency, Kadaster, currently employs its owngeneralization process. However, they would like to know whether the users of theirtopographic datasets are interested in vario-scale. At this moment, there is a workingimplementation of vario scale (made by dr. ir. Martijn Meijers). This implementation,however, is still lacking in, for example, cartographic quality. Therefore the research questionin this project is: how can the implementation of vario-scale be improved to better meet theneeds for end users of Kadaster topographic data?This question is answered by questioning surveying users of Kadaster data on what theywould like to see improved about the existing implementation. Combining this with anexploration of the current software leads to an attempt at improving the currentimplementation. The project goal is set as enabling the road network visualization and mobilemap adaptation. Road network visualization is achieved by building the roads space scalecube and overlay with the background area at the front-end. Mobile map adaptation is realizedby creating the touch screen interaction between the device and the user. Finally, a validationsurvey is conducted to examine the difference between the original vario scaleimplementation and the adapted one.