Estimating Uncertainty in 3D Models using Geostatistics

Student: Maria Rota
Supervisors: Prof. A. Pecker, Dr C. Strobbia


This work proposes a methodological approach for the evaluation of uncertainty in 3D models, with the aim of investigating how uncertainty in the model propagates to the results. It is a general approach for generating soil models that can then be applied to any type of analysis, such as for example site response analyses.

The method consists in exploiting the capabilities of both a classical geotechnical site characterization and the more advanced and refined geostatistical approach. The latter allows to take into consideration the spatial variability of the property of interest and derive a set of equiprobable stochastic representations of reality, reproducing the original structure of the data, through the use of simulation.

After a theoretical introduction of the main instruments necessary to perform a geostatistical analysis, a case study site is described, where a detailed geotechnical and geophysical test campaign has been carried out. An accurate exploratory analysis of the tests results and of all the information available for the site is necessary in order to be able to apply the entire geostatistical procedure to some geotechnical properties of interest. The results of such approach will be reported and discussed.

You may download a digital version of this MSc dissertation here.