RUcore Resource Object
RUcore Resource Object
PDF-1(4.24 MB)
TitleAlternative inversion strategies to resistivity data for targets with sharp boundaries
NameElwaseif, Mehrez H. (author), Slater, Lee D (chair), Keating, Kristina (internal member), Ntarlagiannis, Dimitrios (internal member), Nyquist, Jonathan (outside member), Rutgers University, Graduate School - Newark,
Degree Date2012-05
Date Created2012
SubjectEnvironmental Science,
Geophysics,
Archaeological assemblages
DescriptionEstimating the geometry and resistivity of archeological structures using resistivity models produced as a result of applying smoothness constraints in most inversion algorithms is difficult, especially when structures are closely spaced. However, such quantification is important to facilitate conservation and to minimize the potential of damage when excavations are undertaken. Alternative inversion approaches more appropriate for imaging such targets require either a priori information about the subsurface (e.g. disconnected inversion) or require two or more geophysical datasets to be collected at the same site (e.g. joint inversion). The research outlined in this dissertation presents three novel approaches to improve resistivity imaging of discrete targets without the need to incorporate a priori information in the inversion. The first approach combines an initial 2D smoothness constraint inversion coupled with a digital image processing technique known as a watershed algorithms and a second inversion step incorporating a disconnect in the regularization based on the output of the watershed algorithm. This approach has improved estimate of the geometries of individual targets, but it was not very effective at predicting the resistivity of the targets or resolving closely spaced targets. The second approach combines an initial 2D smoothness constraint inversion coupled with the watershed algorithm and a trained Artificial Neural Network (ANN). Although this approach has been proven effective for resolving widely and closely spaced archeological targets, the results depend largely on the quality of ANN training and on the accuracy of the watershed algorithm geometry prediction. Finally, the third strategy is an iterative approach that combines an initial 3D smoothness constraint inversion that is used only at the first iteration to recover a resistivity model that is fairly consistent with the measured data, from which an initial target location is estimated using an edge detector method and from which a disconnect in the inversion is identified. The disconnect defining the target outline is then progressively improved following each iteration of the inverse procedure. This approach has been proven more effective for resolving widely and closely spaced archeological targets over other approaches, but it is partially sensitive to artifacts in the initial smoothness constraint model.
NotePh. D.
NoteIncludes bibliographical references
NoteIncludes vita
Noteby Mehrez H. Elwaseif
Genretheses
Persistent URLhttp://hdl.rutgers.edu/1782.1/rucore10002600001.ETD.000065030
Languageeng
CollectionGraduate School - Newark Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.