Back to search

University of British Columbia

Parameterized level-set inversion of sharp-boundary targets using radial basis functions

Abstract

dc:description

In mineral exploration and mining, the boundary of a target often matters more than a smoothed image of it. The edge of an ore body informs drill targets and volume estimates, and the extent of a block cave determines where ground support is needed. Conventional inversion recovers a property value in every cell of a discretized model subject to a smoothness penalty. This results in blurring of the anomaly over many cells, forcing an interpreter to set the target/host boundary with an arbitrary threshold. The parameterization developed and applied in this thesis leads to a more accurate representation of the target boundaries. I parameterize the subsurface as a binary partition. The target boundary is the zero contour of a level-set function built from a linear combination of radial basis functions (RBFs), and the background and target property values are parameters recovered alongside the geometry. The parameterization is independent of the survey method and the physical property. The target can split or merge during the inversion, so the number and shape of the targets need not be specified in advance. Only an approximate initial location is required. I implement the parameterization as a custom mapping in the SimPEG framework with the sensitivities to the RBF weights derived analytically. I demonstrate the method on three synthetic problems. In 2D direct-current resistivity examples, it recovers compact conductors at approximately their true conductivity and reshapes a flat initial interface into a layered basement topography. In 3D muon tomography over a block cave, it recovers a sharply bounded low-density region matching the true cave geometry where a smooth inversion does not. In a 3D time-domain electromagnetic example, the smooth inversion stalls above the target misfit and pushes the recovered conductivity toward the survey edges, whereas the RBF inversion fits the data and places a compact conductor at the correct location. The method is demonstrated on synthetic data with a single property contrast between target and host, and I discuss extensions to multiple contrasts.

Degree

thesis:*
Name thesis:degree_name
Master of Science - MSc
Level thesis:degree_level
master's
Discipline thesis:degree_discipline
Geophysics
Grantor dc:publisher
University of British Columbia
Year dc:date
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Pokar, Parth
Contributors dc:contributor
  • Heagy, Lindsey

Rights

dc:rights
Statement dc:rights
  • Attribution 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/95277

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Pokar, Parth. Parameterized level-set inversion of sharp-boundary targets using radial basis functions. master's thesis, University of British Columbia, 2026. http://hdl.handle.net/2429/95277