Seismic velocity monitoring using ambient seismic noise Interferometry in Northern Thailand

Abstract

This thesis tests the feasibility of seismic velocity variation (dv/v) monitoring using ambient noise interferometry in 26 seismometers across Northern Thailand including IU.CHTO station from the IU network, MU.BHSP from the MU network, 11 stations from the TM network, and 13 stations from the T2 network from 2004 to 2021. The study is divided into two parts including (1) single-station ambient noise monitoring using the three-component records at an individual seismometer from 14 stations in the IU, TM, MU, and T2 networks, and (2) station-station ambient noise monitoring using vertical records of 12 stations in the T2 network covering the Chiang Mai Basin (CMB). All ambient noise data are processed by the Msnoise package (Lecocq et al., 2014) to achieve the dv/v information. The four key processing steps are (1) pre-processing by demean, detrend, bandpass filtering, resampling, clipping, and spectral whitening, (2) correlation calculating of any ambient pairs and filtering to generate daily noise correlation functions (NCFs) in the limited frequency bands, (3) averaging all daily NCFs in selected times used as the reference function, and (4) traveltime shift (dt/t) estimating between the daily NCFs and the reference functions then converts by taking negative dt/t to obtain dv/v. The processing parameter justification is initiated at the station JU.CHTO by the single- station approach, and the resulting optimal parameters are expanded to process the other stations. After dv/y retrieval, the feasibility of ambient noise monitoring at each station is assessed by comparing with rainfall, groundwater level, and seismicity for interpretation of seismic velocity variations underneath the stations. Alternatively, the objective of the station-station approach is to investigate spatial co-seismic velocity changes associated with the 2014 Mw 6.2 Mae Lao earthquake, which is the largest earthquake that is instrumentally recorded in Thailand. From the single-station approach, the solutions from six seismometers, including IU.CHTO, MU.BHSP, TM.CMMT, TM.MHMT, TM.LAMP, and T2.CM3 1, are capable of identifying dv/v temporal evolutions in 1 - 6 Hz frequency bands. The dv/v solutions varying between + 2 % reveal strong seasonal variations related to rainfall. At the stations where groundwater data are available such as IU.CHTO, TM.CMMT, and T2.31, the excellent correlations in the dv/v patterns with annual cycles of groundwater levels are observed. The dv/v change (Adv/v) is used to estimate groundwater level change (Ah) via the linear empirical formula Ah = -a(Adv/v). The solutions yield conversion factors a in a range of 1.50 - 2.10 m/%o. The finding shows that this noise-based dv/v monitoring can be a promising tool for tracking groundwater levels with long-term temporary high resolution in areas with permanently installed seismometers. Besides, the dv/v monitoring in 0.12 - 1 Hz frequency bands from the station-station approach in the CMB also shows the seasonal variations between + 0.5 % related to rainfall. However, significant dv/v change associated with the Mae Lao earthquake and local seismicity is not reported due to large distances from epicenters and the low magnitude of earthquakes.