Project Topics
Get Complete Project Material File(s) Now! »
Seismic Wave Field Analysis of Argostoli Dense Array Network
This chapter presents the description of the seismic wave field analysis from dense seismic array data and the corresponding results. First, an overview of the seismic wave field analysis and relevant techniques has been provided. Then the two dense arrays (Array A and B) are presented. Next, the array analysis technique MUSIQUE, the methodology used in this part of the work, along with the data processing procedure has been described. Finally the post-processing criteria and the results of the analysis as well as the possible interpretations of the results have been provided.
Introduction
Site-specific characteristics of the observed ground motions are considered important for the estimation of seismic design parameters in engineering applications.
Seismological observations have indicated that effects of surface geology and geometry (e.g. sedimentary valleys, topography) significantly contribute to ground-motion amplification and variability. These effects are generally associated with a substantial proportion of surface waves in the seismic wave field. Among them, the surface waves diffracted by the basin edges do contribute significantly to the site effects in modifying the wavefield and the resulting ground motion(e.g. Moczo and Bard, 1993; Field, 1996; Chavez-Garcia et al., 1999; Cornou and Bard, 2003; Bindi et al., 2009; Scandella and Paolucci, 2010). An understanding of the seismic wave field crossing the site, hence, is the key aspect to characterize and quantify these effects.
Studies investigating the properties of the wave field have shown that seismic arrays are very useful to characterize the fine-scale structure of Earth’s interior and the variations of the material properties. A single seismometer is unable to determine both velocity and direction of the incident seismic waves while arrays of seismic sensors enable us to study the phase delays that normally cannot be identified in seismograms of single stations. Yet, a one dimensional seismic array can only determine the component of the wave vector which lies in the array direction. Therefore, twodimensional arrays are needed to retrieve the back-azimuth and velocity of the incoming waves.
In this study, the two-dimensional dense seismic arrays, deployed during the seismological experiment in Argostoli basin, as a part of the FP7 EU-NERA (Network of European Research Infrastructures for Earthquake Risk Assessment and Mitigation) 2010-2014 project, was used to analyze the seismic wave field composition. As presented in Chapter 2, the principal dense array (Array A) was located close to the south-western edge of the basin and consisted of 21 velocimeters located in concentric circles with radii of 5 m, 15 m, 40 m and 80 m around the reference station. Another smaller array was deployed near the north-eastern edge and consisted of 10 60 3.2 Seismic wave field analysis velocimeters with interstation distances ranging from 5 to 60 meters. A subset of 46 events recorded by Array A was used for the MUSIQUE analysis. This subset was carefully selected so that the corresponding events are characterized by a homogeneous back-azimuth distribution, local magnitudes ranging between 2 and 5, and epicentral distances ranging between 3 and 200 km from the array center (see chapter 2).
Among the various available array techniques, we have chosen and applied the MUSIQUE algorithm (Hobiger et al., 2012) to analyze our selected events. This algorithm, combining the two algorithms MUSIC (MUltiple SIgnal Characterization; Schmidt, 1986; Goldstein and Archuleta, 1987) and quaternion-MUSIC (Miron et al., 2005; Miron et al., 2006), offers an advanced three-component seismic array processing technique. In addition to the estimation of slowness of the incoming waves, MUSIQUE allows identification of Love and Rayleigh waves, and estimation of the polarization parameters, i.e., ellipticity and sense of rotation of the Rayleigh wave particle motion.
The array analysis was performed over a frequency range of 1 to 20 Hz considering entire duration of the signals. The present chapter focuses on the results of array analysis which include identification and characterization of the diffracted wave fields.
Seismic wave field analysis
When an earthquake occurs, seismic waves are generated as rupture occurs along the fault and propagate towards the site or point of observation after being diffracted, reflected, or scattered, through regional and local earth’s structure. Especially, nearsurface geology plays an important role in modifying earthquake ground motion and, hence, the damage distribution, as witnessed by most of the past destructive earthquakes (e.g. Mexico, 1985; Loma Prieta, 1989; Kobe, 1995; Izmit, 1999). Generally, four types of seismic waves form the seismic signal, namely, pressure (P) and shear (S) waves, Love and Rayleigh surface waves. P- and S-waves are linearly polarized. P-waves are compressional waves propagating along the wave propagation direction. S-waves propagate perpendicularly to the wave propagation direction and consists of two components, SH and SV. SH waves have a motion parallel to the horizontal plane while SV wave motion is perpendicular. Love and Rayleigh surface waves travel along the free surface of the earth and that is the reason they are known as surface waves. Love waves are horizontally polarized shear waves resulting from the superposition of multiple reflected of SH waves at the free surface. Rayleigh waves originate from the superposition between P and SV waves at the free-surface. The coupled P-SV type displacement of Rayleigh waves results in a phase-shift of ±π/2 between the horizontal and vertical components of particle motion which can be represented by an ellipse. The ratio of amplitudes between horizontal and vertical axes of the ellipse defines the ellipticity. In uniform half-space, at shallower depths (at and near the surface), horizontal movement advances the vertical movement by π/2 in phase and the motion (fundamental mode) becomes retrograde. Since P-wave component of the motion decays faster than the SV motion, at sufficiently greater depth, SV motion dominates, and the particle motion becomes prograde corresponding to a phase difference of -π/2.
When the direction of wave propagation is from left to right, counter-clockwise particle motion is retrograde while clockwise is prograde. For multi-layered structure, ellipticity of Rayleigh waves depends on the velocity structure and frequency. Amplitude (energy) of surface waves decays exponentially with depth within a medium. Due to geometric spreading in 2-D, surface wave energy decays with distance r from the source as 1/r whereas it is 1/r2 for body waves making them less prominent on a seismogram. It is also to be noted that different types of waves travel at different velocities depending on the material characteristics of propagation media. P-waves are the fastest and marked by early arrival on a seismogram followed by the S-waves and then surface waves. Love and Rayleigh surface waves involve different mode of propagation (fundamental and higher modes) that propagate at different frequency-dependent velocities whose asymptotic values for the low and high frequencies are close to the shear-wave velocities, in the deep and surficial layers, respectively. The later arrivals on a seismogram are called coda waves that are caused by multipathing/scattering of waves through a heterogeneous structure. In terms of frequency content of a seismogram, the high frequency seismic wavetrain is usually dominated by the arrivals of P and S waves while the low frequency content is dominated by surface waves.
Although initially it was believed that the body wave trapped in the sedimentary layers are responsible for amplification of the ground motion, the contribution of locally generated surface waves in amplification, duration lengthening and spatial variation of ground motion and related damage has been recognized (Kawase, 1996). . At present many urban areas have been developed on sedimentary basin structures, hence, the effects of surface waves need particular attention in the seismic hazard assessments and risk estimation of such sites. In other words, knowledge of the local soil structure and complex wave field propagation of a given site is of utmost importance in order to investigate the physical causes underlying spatial variation of earthquake ground motion and estimate associated risk.
Table of contents :
Chapter 1: Ground-motion variability from finite-source ruptures simulations
1.1 Introduction
1.2 Ground-motion simulations
1.2.1 Kinematic source models
1.2.2 Station layout
1.2.3 Synthetic ground-motion computation
1.2.4 PGV calculation
1.3 Analysis of PGV within-event variability
1.3.1 Variability considering bilateral ruptures only
1.3.2 Variability considering unilateral ruptures only
1.4 Discussion and conclusion
1.5 Data and resources
Chapter 2: Review of Argostoli site and dense array network
2.1 Introduction
2.2 The site: Argostoli, Cephalonia
2.2.1 Seismotectonics
vii Table of Contents
2.2.2 Seismicity
2.2.3 Geology and geomorphology
2.2.4 Argostoli Valley
2.2 Seismological experiment
2.3 Data acquisition
2.4 Catalogue preparation
2.5 Selection of subset of events
2.6 Example wave forms
Chapter 3: Seismic Wave Field Analysis of Argostoli Dense Array Network58
3.1 Introduction
3.2 Seismic wave field analysis
3.3 Argostoli experiment and dense array characteristics
3.3.1 Dataset
3.4 Methodology
3.4.1 MUSIC
3.4.2 Quaternion-MUSIC
3.5 Data processing
3.6 Post-processing
3.7 Results from single dominant source : example event
3.7.1 Event characteristics
3.7.2 MUSIQUE results
3.7.3 Identified back-azimuth
3.7.4 Identified slowness
3.7.5 Energy repartition
3.7.6 Results from Array B
3.8 Robustness of the results
3.9 Summary results for all events
3.9.1 Back-azimuth distribution of the diffracted wave field
3.9.2 Dispersion curve (slowness)
3.9.3 Energy repartition between Rayleigh and Love surface waves
3.10 Results from double source identification
3.10.1 Array A results
3.10.2 Array B results
3.11 Interpretation of the energy partition between Rayleigh and Love waves
3.12 Interpreting observed site amplification
3.13 Discussion and conclusion
Chapter 4: Coherency analysis of Argostoli dense array network
4.1 Introduction
4.2 Short review on coherency models
4.3 Causes of incoherency
4.4 Coherency- a stochastic estimator
4.4.1 Complex coherency
4.4.2 Lagged coherency
4.4.3 Plane-wave coherency
4.4.4 Unlagged Coherency
4.5 Evaluation of coherency
4.5.1 Smoothing parameter
4.5.2 Selection of time window
4.5.3 Statistical properties of coherency: distribution, bias and variance
4.5.4 Prewhitening
4.6 Dataset
4.7 Selection of time-window for coherency estimation
4.7.1 Sensitivity test of the time-window selection
4.8 Estimation of Coherency from the Array Data
4.8.1 Verification of the algorithms used for coherency estimation
4.8.2 Sensitivity of lagged coherency to duration of time window
4.9 Results of coherency analysis from single events
4.10 Statistical analysis considering all the events
4.10.1 Estimation of Confidence Interval (CI)
4.10.2 Coherency estimates from the subset of events
4.10.3 Variation from different time-window selection approaches
4.10.4 Variation associated to the orientation of horizontal components
4.10.5 Variation from the array geometry
4.10.6 Variation from the site-axes orientation
4.10.7 Variation from source back-azimuth
4.10.8 Variability from coda windows
4.10.9 Magnitude Dependence
4. 10.10 Hypocentral Distance Dependence
4.11 Discussion and conclusion
