Variational Mode Decomposition-Based Processing of Time-Domain Induced Polarization Data

TY - ABST

T1 - Variational Mode Decomposition-Based Processing of Time-Domain Induced Polarization Data

AU - Naveed, Khuram

AU - Meldgaard Madsen, Line

AU - Grombacher, Denys

AU - Larsen, Jakob Juul

N1 - Conference code: 29

PY - 2023

Y1 - 2023

N2 - Time domain induced polarization (TDIP) method suffers from a plethora of undesired responses which prominently includes powerline harmonics, electric spikes, self-potentials of the earth and random noise. These undesired responses restrict access to the spectral content which is vital in characterizing rocks, soil and minerals present in the subsurface. Hence, mitigating the impact of these noisy responses measured along with the IP potentials is among the core processing challenges faced by TDIP method today. To this end, we propose the use of variational mode decomposition (VMD) that can decompose time series data into its intrinsic frequency modes (IMFs) which are essentially time series containing a narrowband signal component. Such a decomposition is vital in TDIP signal processing because IP decay is known to have a dominantly low frequency spectrum while noise has considerably higher spectral range. Hence, we employ successive use of VMD on the raw TDIP signal to gradually extract and accumulate the low frequency components of the IP decay in initial IMFs (while noise is distributed among latter modes). We validate the performance of the proposed VMD based approach through synthetic as well as field examples.

AB - Time domain induced polarization (TDIP) method suffers from a plethora of undesired responses which prominently includes powerline harmonics, electric spikes, self-potentials of the earth and random noise. These undesired responses restrict access to the spectral content which is vital in characterizing rocks, soil and minerals present in the subsurface. Hence, mitigating the impact of these noisy responses measured along with the IP potentials is among the core processing challenges faced by TDIP method today. To this end, we propose the use of variational mode decomposition (VMD) that can decompose time series data into its intrinsic frequency modes (IMFs) which are essentially time series containing a narrowband signal component. Such a decomposition is vital in TDIP signal processing because IP decay is known to have a dominantly low frequency spectrum while noise has considerably higher spectral range. Hence, we employ successive use of VMD on the raw TDIP signal to gradually extract and accumulate the low frequency components of the IP decay in initial IMFs (while noise is distributed among latter modes). We validate the performance of the proposed VMD based approach through synthetic as well as field examples.

UR - http://www.scopus.com/inward/record.url?scp=85182928508&partnerID=8YFLogxK

U2 - 10.3997/2214-4609.202320133

DO - 10.3997/2214-4609.202320133

M3 - Conference abstract for conference

T2 - NSG2023 29th European Meeting of Environmental and Engineering Geophysics

Y2 - 3 September 2023 through 7 September 2023

ER -