Since earthquakes can neither be prevented nor predicted, in order to reduce damage to the built environment and human losses, the fundamental goal is to improve the design of buildings. For this purpose, variations in ground motion over short distances during earthquakes arising from both site effects (modifications of the ground motions due to changes in the shallow geological layers) and soil-structure interaction (SSI, influence of built structures in modifying the ground motion during earthquake shaking) have to be better understood and integrated in seismic hazard and risk assessment. In urban areas, SSI is not limited to interactions between a single building and the soil, but is extended to interactions between the city as a whole and the soil, the so-called site-city interactions (SCI). Until now, these effects were mainly studied by 2D and 3D numerical simulations, which are usually based on very simplified representations of the distribution of the buildings and their coupling with the soil, with only a limited number of analyses based on real data sets being carried out in order to investigate this topic. Studies of the wave propagation in buildings and through the soil based on real data sets were usually carried out separately. Studying and understanding the wave propagation through building-soil layers and investigating the interactions that take place is possible only if real data recorded simultaneously by sensors installed in a building and borehole are analyzed jointly. In this thesis, in order to tackle the problem of understanding soil-structure interactions (i.e., modifications of the wave field generated actively through structural vibrations and passively through scattering and diffraction by the presence of a built structure) in more detail, first, the wave propagations through the soil and buildings are analyzed separately in order to identify their main influences on the wave field and to help with the interpretation of the results derived from the innovative joint analysis proposed here. Note that this analysis is shown to provide a more comprehensive view of the wave propagation through the building and soil and the effects of their interaction. The novel approach outlined in this study, the so-called joint deconvolution, analyzes jointly recordings of sensors installed in a borehole and a nearby building. It makes it possible to study the wave propagation through the soil to the building and back under different levels of shaking. In particular, it allows the real seismic input and the wave field being radiated back from a building to the soil to be reconstructed and thus, the energy that is released back from the building to the soil to be quantified. This capacity to extract correctly both the incoming energy and that radiated back using the joint recordings inside the soil and in the building is, amongst others, the main theoretical contribution of this work. After a validation of the novel approach using synthetic data, three studies (in Bishkek, Kyrgyzstan; Istanbul, Turkey and Mexico City, Mexico) of real data sets recorded by downhole installations and building sensors have been carried out and the results are presented in this thesis. These three test cases include different soil conditions and building construction types. Since the impedance contrasts between the buildings and the soil are different, as well as the wave propagation velocities in the subsurface and through the built structures, different building-soil couplings should be expected. For all three test cases, the energy radiated back from the building to the soil was estimated by the new method, and the estimated amount of energy radiated back was found to be not negligible (e.g., for the Bishkek case, at 145 m depth, 10-15% of the estimated real input energy is expected to be from the building, for Istanbul at 50 m depth, this value is also 10-15%, while for Mexico City at 45 m depth, it is 25-65% of the estimated real input energy). The presented study shows the influence of buildings in modifying ground motion during earthquake shaking. If energy is radiated back from a building into the soil, it will also affect the behavior of the buildings located nearby. The potential of the method in analyzing the interactions between individual buildings and the soil by real data sets was demonstrated. The presented method and its appropriate modifications in the future also appear promising for studying more complicated cases, from building-building interactions through the soil up to site-city interactions. This offers the possibility to investigate how the waves generated by each individual built structure can interfere with each other, resulting in increased/decreased hazard. This can depend on the positions of the buildings and the urban geometry, therefore, offering the opportunity to optimize them in order to reduce seismic risk in the future. Furthermore, once the full physical mechanism of soil-structure interactions is understood, if included and integrated in a statistical framework consistent with probabilistic seismic hazard and risk assessment, improved estimations of seismic hazard and risk would be possible.