Ground Model Ensemble Selection Based on Information Theory and Global Inversion of Surface Wave Dispersion Data

Abstract eng:
The purpose of this study is to obtain a ''reasonable ensemble'' of one dimensional shear wave velocity models that explain an observed surface wave dispersion curve with its corresponding measurement errors. The difficulty of finding this set of models lies in the definition of what is to be considered ''reasonable'' on one hand and that we are unaware of the correct velocity depth function description to be used. Therefore we need to be able to compare different velocity depth model parameterizations among each other in a correct way. This particular problem is approached by testing all perceived velocity depth model parameterizations until complexity (related to number of free parameters) doesn't allow the exploration of the model space (curse of dimensionality) and using a bias correction term (here AICc) for comparing the model fit to the data. Having found the overall best earth model, we can predict the maximal acceptable threshold for the misfit (including bias correction) of the mean curve. Consequently, we present a strategy for deriving a combined trans dimensional model ensemble using data uncertainties resulting in site characterization defined as a distribution of most likely models explaining our data.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket