-
f 3D sheets inversion with accurate modeling of AEM systems
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 6th International AEM Conference & Exhibition, Oct 2013, cp-383-00024
Abstract
Modeling of 3D thin conductive plates most often considers a homogeneous very resistive background media, which is sufficiently accurate for a large of part of mining exploration surveys where resistive bedrock is very likely. However, cases exist where background earth can be conductive, e.g. with a sedimentary overburden at near surface. For such scenarios, it is necessary to at least consider a tabular background earth model (that could also vary along the flight lines) to limit the errors on the thin plate models estimation. Through different scenarios we show how important the consideration of this conductive overburden is, and how bad the determination of the thin plates’ parameters can be if this near-surface layer is not accurately determined. The inversion tests of synthetic data show that a too resistive considered overburden results in a too deep thin plate model, and a too thin overburden in a both too deep and badly oriented thin plate. Thanks to our previous experiences on accurate AEM system modeling, we are able to accurately model AEM systems to avoid misinterpretation with this new 3D sheet inversion code. We particularly insist on the importance of the system calibration. Synthetic inversion tests show that if an error in amplitude level has a negligible impact, a bad timing of ~ 5 μs leads to bad orientation estimates. The 3D sheet modeling algorithm, based on surface integral equations, has been implemented within our in-house inversion program AarhusInv. Optimization for fast solving of the scattered fields and for managing a large number of source-receiver positions have been made to compensate the computation drawback of considering a tabular background (Green functions with Hankel transforms). To improve the efficiency of the code even more, efforts have also been made to parallelize it through the OpenMP library to benefit from today’s computers, which can cheaply hold several processor units.