TY - GEN
T1 - On The Presence of Correlated Noise in Transient Electromagnetic (Tem) Monitoring Data
AU - McLachlan, Paul Jack
AU - Khare, Smith Kashiram
AU - Grombacher, Denys
AU - Larsen, Jakob Juul
AU - Christensen, A.
AU - Zamora Luria, Juan Carlos
PY - 2022
Y1 - 2022
N2 - This work investigates a novel approach for investigating correlated measurement errors in TEM data. Generally, measurement errors are assumed uncorrelated. However, noise sources include very-low-frequency (VLF) signals such as radio waves, which are not random. TEM monitoring, where systems are installed semi-permanently, offers the unique opportunity to investigate measurement errors more thoroughly. The signal-to-noise ratio in TEM data is generally improved by data stacking; however, continued stacking of correlated noise sources may introduce bias. Such biases are significant when the target of interest is subtle variations such as groundwater table variations. This work provides some background on the general treatment of errors in TEM data before an approach utilizing covariance matrices for characterizing correlated noise is introduced. The noise in a monitoring TEM data set indicated that errors were strongly correlated; furthermore, patterns in covariance matrices change between daily surveys showing that the proportions of VLF signals are not stable over time.
AB - This work investigates a novel approach for investigating correlated measurement errors in TEM data. Generally, measurement errors are assumed uncorrelated. However, noise sources include very-low-frequency (VLF) signals such as radio waves, which are not random. TEM monitoring, where systems are installed semi-permanently, offers the unique opportunity to investigate measurement errors more thoroughly. The signal-to-noise ratio in TEM data is generally improved by data stacking; however, continued stacking of correlated noise sources may introduce bias. Such biases are significant when the target of interest is subtle variations such as groundwater table variations. This work provides some background on the general treatment of errors in TEM data before an approach utilizing covariance matrices for characterizing correlated noise is introduced. The noise in a monitoring TEM data set indicated that errors were strongly correlated; furthermore, patterns in covariance matrices change between daily surveys showing that the proportions of VLF signals are not stable over time.
U2 - 10.3997/2214-4609.202220119
DO - 10.3997/2214-4609.202220119
M3 - Article in proceedings
T3 - EAGE Conference and Exhibition, Proceeding
SP - 1
EP - 5
BT - NSG2022 28th European Meeting of Environmental and Engineering Geophysics
PB - European Association of Geoscientists and Engineers
ER -