We implement two different inversion strategies for solving the inversion of full-waveform induced polarization data (FWIP inversion): the first is a local inversion based on the Levenberg-Marquardt (LM) algorithm, whereas the second is a global inversion driven by the Particle Swarm Optimisation (PSO). We limit the attention to synthetic data with the aim to maintain the inversion at a simple level and to draw essential conclusions about the suitability of the two approaches for solving the FWIP inversion. However, realistic noise is added to the simulated data in order to better simulate a field dataset and a tailored processing sequences is applied to the synthetic, noise-contaminated data to enhance the signal-to-noise ratio. Moreover, the analysis of residual function maps and the sensitivity analysis of the inversion kernel allow us to gain a better insight into the ill-conditioning of the FWIP inverse problem. It turns out that the FWIP is a well-posed problem in case of double-peaked resistivity spectra, whereas it becomes a hopelessly ill-conditioned problem when the subsurface model generates single-peaked resistivity spectra. The LM and PSO methods give comparable results, although the gradient-based approach achieves faster convergence and more stable predictions.

Inversion of FW-IP data - Sensitivity analysis and synthetic inversions through local and global optimization strategies

Vinciguerra Alessandro;Aleardi M.;Costantini P.
2019-01-01

Abstract

We implement two different inversion strategies for solving the inversion of full-waveform induced polarization data (FWIP inversion): the first is a local inversion based on the Levenberg-Marquardt (LM) algorithm, whereas the second is a global inversion driven by the Particle Swarm Optimisation (PSO). We limit the attention to synthetic data with the aim to maintain the inversion at a simple level and to draw essential conclusions about the suitability of the two approaches for solving the FWIP inversion. However, realistic noise is added to the simulated data in order to better simulate a field dataset and a tailored processing sequences is applied to the synthetic, noise-contaminated data to enhance the signal-to-noise ratio. Moreover, the analysis of residual function maps and the sensitivity analysis of the inversion kernel allow us to gain a better insight into the ill-conditioning of the FWIP inverse problem. It turns out that the FWIP is a well-posed problem in case of double-peaked resistivity spectra, whereas it becomes a hopelessly ill-conditioned problem when the subsurface model generates single-peaked resistivity spectra. The LM and PSO methods give comparable results, although the gradient-based approach achieves faster convergence and more stable predictions.
2019
978-94-6282-263-4
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11568/1022829
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact