Geometrical gravity inversion to refine geological models

Jeremie Giraud and Mark Lindsay and Mark Jessell. ( 2021 )
in: 2021 RING Meeting, ASGA

Abstract

We present a geophysical inversion technique designed for the recovery of the geometry of rock units in 2D and 3D. It relies on the iterative adjustment of the location of interfaces between homogenous rock units implicitly defined by the signed-distance to each contact. We formulate the inverse problem using a generalized level-set method capable of dealing with any number of rock units. Our implementation uses a deterministic least-squares inversion framework. It is flexible and permits to incorporate prior geological information such as the location, shape and aspect ratio of rock units in the regularisation function and to adjust the thickness of the interface between them. The utilisation of the algorithm we make here focuses on adjusting pre-existing geological models to honour geophysical data. We first explore the capabilities of the proposed inversion approach using noisy synthetic gravity data and perform the proof-of-concept. We then model field gravity data from the North-Eastern part of the Yerrida Basin (Western Australia) to model the geometry of a prospective greenstone belt that previous studies highlighted as requiring further investigation. The synthetic example demonstrates the capability of the proposed method to improve the recovery of the geological bodies' geometry and to change the initial geological model's topology to honour the geophysical measurements. The application to real world data allows tuning existing interpretations of the tectonic setting. In particular, it suggests that the undercover mafic greenstone belt might be shallower than anticipated by geology alone and thinner overall.

Download / Links

BibTeX Reference

@INPROCEEDINGS{GIRAUD_RM2021,
    author = { Giraud, Jeremie and Lindsay, Mark and Jessell, Mark },
     title = { Geometrical gravity inversion to refine geological models },
 booktitle = { 2021 RING Meeting },
      year = { 2021 },
 publisher = { ASGA },
  abstract = { We present a geophysical inversion technique designed for the recovery of the geometry of rock units in 2D and 3D. It relies on the iterative adjustment of the location of interfaces between homogenous rock units implicitly defined by the signed-distance to each contact. We formulate the inverse problem using a generalized level-set method capable of dealing with any number of rock units. Our implementation uses a deterministic least-squares inversion framework. It is flexible and permits to incorporate prior geological information such as the location, shape and aspect ratio of rock units in the regularisation function and to adjust the thickness of the interface between them. The utilisation of the algorithm we make here focuses on adjusting pre-existing geological models to honour geophysical data. We first explore the capabilities of the proposed inversion approach using noisy synthetic gravity data and perform the proof-of-concept. We then model field gravity data from the North-Eastern part of the Yerrida Basin (Western Australia) to model the geometry of a prospective greenstone belt that previous studies highlighted as requiring further investigation. The synthetic example demonstrates the capability of the proposed method to improve the recovery of the geological bodies' geometry and to change the initial geological model's topology to honour the geophysical measurements. The application to real world data allows tuning existing interpretations of the tectonic setting. In particular, it suggests that the undercover mafic greenstone belt might be shallower than anticipated by geology alone and thinner overall. }
}