Skip to content
Technical University of Munich

Concept - Integration of data and databases for (geo-) statistical modeling of uncertainties

Contact:

M.Sc. Andreas Wiegel

Dr.-Ing. Andrés Peña Olarte

 

 

Concept - Integration of data and databases for (geo-) statistical modeling of uncertainties

Conventional process for the generation of a deterministic subsurface model
with (linear) interpolation

Integration of drilling data, soundings and laboratory tests in a probabilistic subsoil model
using (geo-)statistical methods [1]

Geostatistical of simulation methods for quantifying spatial uncertainty and model validation

One realization of Soil Behavior Type (SBT) with measurements from CPTs of Sequential Gaussian Simulation (SGSIM) (left) and one of Sequential Indicator Simulation (SISIM) (right) [2]
CPT derived Soil Behavior Type (SBT)
result of Leave-One-Out-Cross-Validation (LOOCV) example at 55 m in SGSIM (left)
and one of SISIM (right) model; (black line: excluded test data, dark grey: 50 % confidence interval, grey: 90 %, light grey:100%) [2]

Prediction for spatial distribution of soil layers and corresponding uncertainty

a) Most probable soil behavior type after 500 Sequential Gaussian Simulations;
b) Corrected tip resistance qt and uncertainty displayed as entropy related to the most probable soil type [3]

Subsoil model located in Munich
a) most probable facies;
b) Entropy (certainty) of the prediction of the clay layers after 100 simulations [3]

To top