Investigation in New Methods for Passive Seismic Exploration with Application to Simulation of Strong Motion Scenarios in Campo de Dalías (Almería)

Abstract

The aim of this thesis is to obtain the seismic velocity modelling of Campo de Dalías basin (CDB). This work contributes to filling the lack of investigations on site effects characterisation and opens the possibility of forecasting earthquake damage scenarios in this sedimentary basin. Given the importance of the S-wave velocity (VS) as a relevant parameter influencing the strong ground motions and some of the seismic-induced wave phenomena, various methodologies based on seismic ambient noise were employed to accurately estimate the VS structure of CDB. A total number of seven array measurements of ambient noise were carried out to obtain the surface wave dispersion curves of the fundamental Rayleigh wave mode by applying the Spatial Autocorrelation (SPAC) method. To properly characterise the sedimentary units of CDB, which reach thicknesses of 1000 metres in the deepest zones of the basin, the apertures of these arrays varied from tens to hundreds of metres. The lack of borehole information within the limits of these arrays impeded a reliable interpretation of their shear wave velocity profiles in terms of geological units more than the geological basement. The ground models obtained after the inversion of the experimental dispersion curves indicate that the VS of the geological basement of CDB is ranging between 2040 and 2800 m/s. The shallower ground structure of the two biggest urban areas of CDB, El Ejido and Roquetas de mar towns, was also studied by the analysis of geotechnical borehole reports and the application of the active Multichannel Analysis of Surface Waves (MASW) method. The frequency bandwidth covered by the fundamental dispersion curves obtained from MASW varied widely in both towns, but it spans from 10 to 50 Hz in the majority of profiles. In terms of VS30 and attending to the Eurocode 8 classification, the ground in El Ejido urban area can be classified as type B (360 < VS30 <800 m/s), while in Roquetas de Mar grounds type B and type C (VS30 < 360 m/s) are found. The seismic microzonation aimed at estimating the frequencies of the S-wave resonance was carried out through the densification of single station observations of seismic ambient noise, which were distributed on a square grid of 1 x 1 km projected over the coastal plain. El Ejido and Roquetas de mar towns were sampled with smaller grid sizes of 250 and 400 metres respectively. The total number of single-station measurements performed in these grids amounts to 388. The Horizontal-to-Vertical Spectral Ratio (HVSR) was the methodology employed to analyse the single station observations. The range of resonance frequencies obtained in CDB goes from 0.2 Hz, seen in the lowest fundamental HVSR peaks, to 27 Hz in the highest secondary peaks. The anticline and syncline forms with an ENE-WSW trend that fold the older sediments and the basement of CDB are revealed by interpolation of the fundamental HVSR frequencies. The spatial trends observed in the fundamental resonance frequencies are concordant with the depocenter and anticlinal positions, which vary from 0.2 to 0.4 Hz and between 0.6 and 1.2 Hz respectively. A quantitative analysis of the HVSR shape variability was performed by using up to two years of continuous seismic ambient noise data in three permanent seismic stations. Despite not observing changes that compromise the robustness of the HVSR to identify the site's predominant frequency, the correlations observed with the groundwater cycles add to this methodology the ability of a monitoring tool particularly sensitive to the S-wave modifications. The capacity of the HVSR in conjunction with inversion methodologies to estimate 1-D seismic velocity structures made the pseudo 3-D modelling of CDB feasible by using the 1 x 1 km grid of single-station measurements. The HVSR modelling based on the Diffuse Field Assumption (DFA) applied over the main grid of measurements enabled the estimation of individual velocity models by inversion of the HVSR curves. The issue of non-uniqueness from HVSR inversions has been alleviated by the numerous borehole information gathered, which also helped to build a relationship between the fundamental HVSR frequency and depth to the geological basement. In order to test the 3-D model built after HVSR inversions an earthquake ground motion was simulated. Such test was based on the comparison between the observed and the synthetic waveforms generated by full 3-D modelling with finite differences. The waveforms of three accelerometer stations deployed in CDB by the time of the control event (MW 4.1, November 2010) were used to compare simulated and real seismograms. The agreement between the observed and modelled power spectral amplitudes in the three control stations is satisfactory. This result confirms the capacity of the seismic ambient noise methodologies to build three-dimensional ground models through cost-effective strategies