Analysis of Modelling 3D Bedrock Surfaces using Interpolation Methods : A Comparison between Models of 3D Bedrock Surfaces using Different Interpolation Methods

Sammanfattning

Geological and geotechnical studies are a critical component of most projects, and it has a significant impact on the economy, project timelines, and subsequent project design. Today, the development of GEOBIM eases this analysis, thus saving costs, speeding up processes and avoiding risks. The recently developed Tyréns algorithm generates 3D bedrock surfaces using kriging. Commercial programs (such as ArcGIS and Civil 3D which are the programs used) can generate them as well. So, this project aims to compare how well Tyréns’ algorithm creates surfaces in contrast with commercial programs. 3D models of the bedrock have been developed using different interpolation methods. Furthermore, true bedrock data is available; they had to make an excavation of an area inside the model boundaries, i. e. directly the real bedrock could be measured. So, a thorough and complete scrupulous analysis is performed. Kriging, nearest neighbor, triangulation and inverse distance weight, have been the interpolation methods used in this project for the generation of surfaces, using ArcGIS. The comparisons made are: • between generated surfaces (Kriging with respect to the rest of the methods), • between the surface generated with Kriging in ArcGIS and the surface generated by the algorithm, • between the Kriging and TIN (Triangulated Irregular Network) surface with respect to the real surface (the one excavated). • and finally the surface of the algorithm with respect to the real surface. Taking into account all the above comparisons, different results have been achieved. On the one hand, inverse distance weight is not a suitable method for bedrock analysis since it does not operate properly in areas where there is no data. Next, nearest neighbor, which has a similar functioning as inverse distance weight, has a better results, but it is not the best solution. Kriging provides the best performance: it resembles reality. In addition, the model generated by ArcGIS and the algorithm are very similar, especially in those areas where there is a significant amount of data. Finally, the triangulation method has obtained results that are similar, but not better, to Kriging. Although it is a simple system, it is effective and fast.