CFE 2019: Start Registration
View Submission - CMStatistics
B1856
Title: Statistical inference in hydraulic tomography with wavelet-based priors Authors:  Philipp Wacker - FAU Erlangen-Nuernberg (Germany) [presenting]
Peter Knabner - FAU Erlangen-Nuernberg (Germany)
Abstract: Hydraulic tomography is a technique for inferring subsurface properties like hydraulic permeability as a function of the domain. Probes are alternatingly used as pumps and pressure measuring device. The process can be compared to medical imaging technology (only that we are interested in the subsurface's inner composition, not some patient's). This constitutes an infinite-dimensional statistical inverse problem and is usually solved by a geostatistical approach. We will try to tackle this problem by setting a prior distribution which is governed by random superpositions of wavelet functions (similar to more common priors which are random superpositions of sine and cosine functions).