Premium
One‐dimensional non‐linear inversion of magnetotelluric data: the importance of data errorrs 1
Author(s) -
Belay R.,
Rasmussen T.M.
Publication year - 1995
Publication title -
geophysical prospecting
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.735
H-Index - 79
eISSN - 1365-2478
pISSN - 0016-8025
DOI - 10.1111/j.1365-2478.1995.tb00288.x
Subject(s) - magnetotellurics , inversion (geology) , data point , regional geology , economic geology , environmental geology , limit (mathematics) , geology , mathematics , geophysics , algorithm , hydrogeology , mathematical analysis , telmatology , seismology , physics , electrical resistivity and conductivity , geotechnical engineering , quantum mechanics , tectonics
Inversion of noisy magnetotelluric data over a horizontally stratified earth has no unique solution. Instead, the resistivities and thicknesses of the layers can only be specified to lie within certain bounds at a particular confidence level. The significance of having correctly estimated data errors when calculating the parameters and parameter bounds is discussed. Emphasis is put on the very corrunon problem of not being able to obtain models which can be considered acceptable with a reasonable certainty from a statistical point of view. A method is presented to obtain a pragmatic data error description by adjustment of wrongly estimated data errors while keeping the suppression of 1D data features at a minimum. A comparison is made between calculation of the parameter bounds based on approximating the confidence surface by hyperellipsoids and a calculation based on a full non‐linear most‐squares analysis. In general it is found that the approximation works very well when the confidence limit is small. However, significant deviations are found in some cases.