The aim of this research is to find suitable level of details for environment modelling, to improve methods to reconstruct 3D objects, and to update the 3D TOP10NL dataset. The research needs to study what kind of environmental simulation would be, and what model details are required for each aim, include minimum requirement, optimal requirement. The second aim of this research is to study and develop methods to decompose a building model into parts according to their heights. The LOD1 building models in 3D TOP10NL are just one block for one building, so are not enough for many situations. A building with components will have better environment simulation than one with only one box. The third goal is to accelerate the available tools for quicker computing and to update models with photogrammetric point clouds. The AHN2 data is collected once four years, which are not enough for up-to-data models for 3D SDI. An alternative 3D source data, photogrammetric point cloud by dense matching, collected once or twice a year, is a promising data for a more frequent model updating.