Enhancing Model Reliability from TEM Data Utilizing Various Multiple Data Strategies

Over fifteen years, we have developed and utilized forward and inversion techniques to interpret electromagnetic data collected with various commercial systems. A wide range of survey configurations have been utilized including in-loop and outside-loop measurements with both moving and fixed source configurations and with arbitrary location and orientation of receivers. A variety of different inversion strategies have been developed based on either overdetermined or underdetermined approaches utilizing approaches similar to those that other researchers have adopted. These algorithms have been used extensively in a range of applications including mining exploration and groundwater applications. This experience leads us to the belief that a more comprehensive approach must be taken to ensure reliable results. We have developed inversion algorthms that simultaneously incorporate data from both multiple data components or multiple data locations. Incorporating various data into an inversion process provides better signal-to-noise ratios within the inversion. Applying the inversion on carefully selected data that contain information about different geological structures may enhance the resolution of the inverted models and result in more meaningful models. In this paper, we begin by performing an underdetermined Occam inversion on synthetic data simulated with the configurations where the receiver is inside a transmitter loop (in-loop) or outside a transmitter loop (outside-loop). The inversion technique essentially generates smooth models that fit the data within a prescribed tolerance. We built synthetic layered earth models to generate impulse responses plus Gaussian noise upon which we ran inversion. Specifically, we built the first layered earth model by inserting a conducting layer into a relatively resistive host medium. Our inversion results of this model show that the inversion on either the in-loop data or the outside-loop data can resolve the conducting layer. Further, a joint inversion of both the in-loop and the outside-loop data leads to an improved inversion model. Our second synthetic layered earth model was built by adding a thin conducting overburden to the first model. In this case, our inversion results show that the in-loop data may resolve the top overburden layer better than the outside-loop data. However, the inversion on the in-loop data did not resolve the basement, that is, the lower half-space. Moreover, the application of inversion on the outside-loop data may detect the lower half-space, and a joint inversion of both the in-loop and the outside-loop data gives rise to an overall improved model with enhanced resolution of both shallower and deeper layers. In short, we utilized synthetic examples to demonstrate that the inversion on the in-loop data tend to resolve the top layers better than the inversion on the outside-loop data while the outside-loop data may see the deeper structures better than the in-loop data, and inverting both the in-loop and the outside-loop data simultaneously may lead to layered earth models of enhanced resolution. We also performed a overdetermined least-squares inversion on a ground data set with a large loop from the Hornby Bay basin in western Nunavut of Canada.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.