Surface wave tomography and assessing the uncertainty of ambient seismic noise data across Thailand

Traditional passive seismic methods to image the Earth’s interior require energy from earthquake sources. An alternative approach uses the quasi-random signal of ambient seismic noise that is ever-present in the seismic record. Dispersion characteristics of surface waves are extracted from the cross-correlation of the noise between two stations, and inverted to produce tomographic models.

This study investigates ambient noise tomography beneath Thailand. Vertical component data from 39 vertical broadband and short period seismic stations operating between 2011 and 2015 across Thailand and adjacent countries were used. A specific challenge for this region is that, unlike most data from global seismic networks, the data set has a significant proportion of non-continuous recordings. The impact of these gaps in data, on the recovery of reliable surface wave information is investigated.

Initially, synthetic tests were applied to test the effect of discontinuous data and results indicated that data with total gap lengths less than 5 minute per 1-hour window could be used. Subsequently, the stability of the surface wave dispersion was investigated with respect to the signal to noise ratio (SNR). After stacking, the reliable data were selected based on the cut-off SNR values which equal 5 for the precursory SNR and 9 for the trailing SNR. In this stage, a phase-weighted stack was introduced to reduce persistent-incoherent local noise however in the final stages of the thesis it was discovered that the application of the PWS was destroying the information at the longer periods. Finally, a least-squares tomographic inversion was applied to the dispersion data to create Rayleigh wave group velocity maps at period range 5 – 50 s. Three different durations of data (2011-2012, 2014-2015, and 2011-2015) were inverted to further assess the uncertainty in final results.