Abstract:The analysis of clutter in A-scans produced by energy randomly scattered in some specific geological structures, provides information about changes in the shallow sedimentary geology. The A-scans are composed by the coherent energy received from reflections on electromagnetic discontinuities and the incoherent waves from the scattering in small heterogeneities. The reflected waves are attenuated as consequence of absorption, geometrical spreading and losses due to reflections and scattering. Therefore, the amplitude of those waves diminishes and at certain two-way travel times becomes on the same magnitude as the background noise in the radargram, mainly produced by the scattering. The amplitude of the mean background noise is higher when the dispersion of the energy increases. Then, the mean amplitude measured in a properly selected time window is a measurement of the amount of the scattered energy and, therefore, a measurement of the increase of scatterers in the ground. This paper presents a simple processing that allows determining the Mean Amplitude of Incoherent Energy (MAEI) for each A-scan, which is represented in front of the position of the trace. This procedure is tested in a field study, in a city built on a sedimentary basin. The basin is crossed by a large number of hidden subterranean streams and paleochannels. The sedimentary structures due to alluvial deposits produce an amount of the random backscattering of the energy that is measured in a time window. The results are compared along the entire radar line, allowing the location of streams and paleochannels. Numerical models were also used in order to compare the synthetic traces with the field radargrams and to test the proposed processing methodology. The results underscore the amount of the MAEI over the streams and also the existence of a surrounding zone where the amplitude is increasing from the average value to the maximum obtained over the structure. Simulations show that this zone does not correspond to any particular geological change but is consequence of the path of the antenna that receives the scattered energy before arriving to the alluvial deposits.

A complete and reliable Seismic Soil Response prediction requires a thorough knowledge of the Geotechnical and Geological structure, as well as algorithms to find the transfer function. This fact entails a significant complexity and, consequently, strengthens empiric methods, as for instance the well-known Horizontal to Vertical Spectral Ratios (HVSR). Moreover, the long return periods of earthquakes, especially in low- to-moderate-seismicity regions as Barcelona, hinder the comparison between predictive studies and observations in case of earthquake. In these sites, methods based on ambient vibration (seismic noise) become important. In this PhD thesis the Seismic Soil Response phenomenon is approached trough new theoretical contributions and applications. The main purpose has been to provide new insights and to generate new tools to improve the comprehension and the resolution of Seismic Soil Response related problems. In this way, innovative theories as Surface Sources (DSS) and Diffuse Fields (DFA) have been applied to solve the HVSR direct problem in horizontal layered media, among others. Also, specific field studies have been performed on test sites of the Barcelona plain with three main purposes: i) to improve the knowledge of the soil by means of array techniques based on ambient vibration, ii) to provide new techniques to allow an easy and fast analysis of the lateral heterogeneities and to accurately locate soil anomalies; iii) to suggest and apply smart techniques to densify HVSR measurement sites. Thus, an innovative outcome of this work has been using Ground Penetrating Radar (GPR) for preliminary detection and characterization of anomalous subsoil sections, with special emphasis in those due to paleochannels and to seasonal streams draining rainwater in mountain-sea direction. The observed HVSR curves (one station measurements) are compared with the theoretical ones obtained from DSS and DFA. Moreover, empiric dispersion curves of the Surface Waves (array measurements) are compared with the theoretical ones corresponding to the soil layered structure. Both procedures allow to check for the suitability of the new proposed tools and, to establish the reliability of the found parameters. Thus, we have been able to increase the knowledge, resolution and reliability of the soil parameters in test sites of the Barcelona plain leading to a better knowledge of the Transfer Functions. From different configurations of the DFA formulation it has been shown that, for a specific site, the HVSR response for ambient vibration is linked to HVSR response when the source is a wave-field where every spectral component is obtained by averaging the spectral components of the records of a great number of earthquakes. When lateral homogeneity assumption fails, the horizontal layered formulations are not suitable and discrepancies between empirical and theoretical responses appear. These differences have been attributed to lateral effects. In this sense, analytical results have been obtained for a constant-section valley illuminated by a two-dimensional diffuse field. This approximation enables finding specific properties of HVSR due to lateral contacts. Finally, this work has shown the need for taking into account the detailed subsoil structure in seismic micro-zonation and risk studies, as these heterogeneities may have significant importance on seismic response. So several advanced tools based on GPR have been provided, allowing, in an easy and quick way, detecting and sizing the heterogeneous areas and suggesting sections where ambient vibration measurement should be intensified. La predicción detallada de la respuesta sísmica del suelo requiere un conocimiento fino de la estructura geológica/geotécnica y el uso de algoritmos que permitan predecir su función de transferencia. Esto presenta una complejidad significativa y ha otorgado un peso importante a métodos empíricos, entre los que destacan las razones espectrales entre componentes horizontal y vertical (HVSR). Además, los largos períodos de recurrencia del fenómeno sísmico, especialmente en entornos de peligrosidad moderada o baja como Barcelona, complican el contraste entre los estudios predictivos y su observación en caso de terremoto. Entonces, el estudio del subsuelo basado en el análisis de la vibración ambiental (ruido) adquiere un especial interés. En esta Tesis se aborda el problema de la respuesta sísmica de suelos con nuevas contribuciones teóricas y aplicadas a la ciudad de Barcelona, que tienen por objetivo aportar nuevo conocimiento y nuevas herramientas para mejorar su comprensión y resolución. Se han presentado aplicaciones de las novedosas teorías de fuentes en superficie (DSS) y de campos difusos (DFA) para resolver el problema directo HVSR en un modelo de capas plano-paralelas, entre otras. Por otra parte se han realizado estudios experimentales específicos sobre zonas test del llano de Barcelona con tres propósitos: i) aumentar el conocimiento del subsuelo usando técnicas de estación múltiple basadas en el ruido ambiental ii) proponer técnicas nuevas que permitan analizar de forma relativamente rápida y simple la heterogeneidad lateral y localizar y dimensionar anomalías del subsuelo y iii) proponer y aplicar técnicas inteligentes de densificación de los puntos de medida HVSR. Así, un aspecto novedoso de este trabajo ha sido el uso de radar de subsuelo (GPR) para la detección y caracterización preliminar de secciones anómalas del subsuelo, con especial énfasis en las debidas a paleocanales y a cauces de riera que drenan las aguas pluviales en la dirección montaña-mar. Así, se ha obtenido modelos de suelo a partir del ajuste simultáneo de las curvas experimentales HVSR (estación simple) a la respuesta de las teorías DSS y la DFA, y de las curvas experimentales de dispersión de las ondas superficiales (redes de estaciones) a la respuesta teórica del modelo por las principales técnicas. Por un lado se ha comprobado la idoneidad de las herramientas teóricas usadas, y por otro la confiabilidad de los parámetros del subsuelo. De este modo se han obtenido los parámetros del subsuelo en las zonas test del llano de Barcelona con una buena resolución y fiabilidad, lo que, a su vez,repercute en un mejor conocimiento de las funciones de transferencia. La teoría DFA, en sus diversas configuraciones, ha permitido vincular experimentalmente las respuestas HVSR de un sitio concreto cuando se usa como fuente: i) la vibración ambiental y ii) el campo de ondas resultante de sumar un gran número de registros de sismo en el sitio, construyendo cada componente espectral a partir del promedio de las componentes de cada registro. Cuando el subsuelo no presenta homogeneidad lateral las formulaciones plano-paralelas no son válidas y aparecen discrepancias entre las respuestas, experimentales y teóricas, que se han atribuido a efectos laterales. En este sentido, se han obtenido resultados analíticos para un valle de sección constante, iluminado por un campo difuso bidimensional que muestran propiedades específicas de HVSR debidas a contactos laterales. Por último, se ha determinado la necesidad de tener en cuenta la estructura fina del subsuelo en estudios de microzonificación y riesgo sísmico, ya que sus heterogeneidades tienen una importancia significativa en la respuesta sísmica. Para ello se ha proporcionando herramientas avanzadas basadas en GPR que permiten, de forma relativamente sencilla, detectar y dimensionar las zonas heterogéneas y sugerir secciones donde sea aconsejable una densificación de los puntos de medida.