Final Report Summary - QUEST (Quantitative Estimation of Earth's Seismic Sources and Structure) The Initial Training Network QUEST (Quantitative Estimation of Earth’s Seismic Sources and Structure, www.quest-itn.org) was a project in the area of computational seismology focussing on all aspects of extracting information from seismic observations in terms of Earth’s internal structure and sources of seismic energy on all scales. The network consisted of a wide spectrum of institutions with expertise in computational modelling, geology, source physics, mathematical aspects of inverse problems, and geophysical exploration. The inclusion of one of the largest geophysical contractor (Schlumberger Research Cambridge, UK) as partner provided access to industry related problems (e.g. hydrocarbon exploration) and direct application of the tools that were developed in QUEST.QUEST had a strong methodological approach. The reason is that, due to the computational resources nowadays available, a new approach to seismic inverse problems is possible. Instead of reducing the seismograms to a few bytes of information one now tries to extract a maximum amount of information from the observations by trying to model and fit the complete seismograms. In addition, the use of ocean-generated ambient vibrations for the imaging of Earth’s interior had just started shortly before the onset of QUEST and turned out to be another central methodology to be applied. The goal of QUEST was to bring those research tools from innovative leading-edge methodologies to a standard procedure used to do seismic tomography and assess model uncertainties. Within QUEST tremendous progress was achieved in the sense that some of the first 3-D tomographic images based entirely in the modelling of 3-D wave propagation tools emerged. These include models on local scales as well as recently new models of the European continent that will mark a new era in tomographic reconstruction. Substantial progress was also possible in the area of ambient noise studies. The use of observations that “only” contain noise and no earthquake related information was to a large part pioneered by one of the QUEST partners. The correlation techniques that allow the tomographic reconstruction without having to rely on earthquake data could be further extended and new images are emerging for the central European region and other areas. In addition - as indicated in the project objectives – the domains of application could be extended to include other wave types. In a related study the origin of the “noise” signals could be modelled for the first time as wave fields generated in the oceans by interfering water waves. The QUEST project was heavily influenced by the occurrence of several major earthquakes within the life time of the project. These include the M8.7 Maule earthquake in Chile in February 2010 and the M9.0 Honshu earthquake in March 2011. QUEST researchers used the data from these earthquakes in particular to advance the understanding of the earthquake source process and the observations and interpretations of the Earth’s free oscillations that are generated by such large earthquakes. The considerably refined tools in the whole spectrum of waveform inversion methodologies – combined with the adaptation to supercomputer infrastructure - has led to some exciting new discoveries in the final project years. Key results include the emergence of body waves from the cross-correlation of ocean-generated seismic noise recordings. The analysis of ocean-generated ambient noise continues to revolutionize seismology and some major new routes where opened with QUEST funding. This also includes the new opportunities for recovering temporal changes of Earth’s seismic properties in the vicinity of active faults or inside volcanoes. In the following we highlight progress in the various project domains:Computational seismic wave propagation: 3-D forward modelling is the key ingredient to waveform inversion technology. An unprecedented extensive analysis of many different numerical forward modelling schemes allowing for the first time a quantitative comparison as to the efficiency of various methods. Major progress could be made in understanding the modelling in heterogeneous media using homogenization approaches. This may - in the near future – lead to a revival of simpler regular-grid methods for the simulation of waves in heterogeneous media. Advances were also made using the discontinuous Galerkin method on unstructured tetrahedral grids extending it to the use for industrial and regional applications. Several QUEST partners co-initiated the first e-infrastructure project VERCE (www.verce.eu) in computational seismology aiming for a community infrastructure that allows to considerably widen the access of the scientific community to high-performance computational tools on supercomputer infrastructure. QUEST associated researchers provided some of the software applications (e.g. ObsPy, seissol, specfem) that are currently being integrated into the VERCE platform. Seismic inversion technology. This work package saw the emergence of new approaches to quantify uncertainties in seismic inverse problems one of the major challenges for the coming years. The domain of applications for the analysis of ocean-generated noise was considerably extended. For the first time body waves emerged in noise correlations (Science paper) with applications on regional and global scales. Major progress was achieved in understanding and quantifying the generation of noise in the oceans. Analysis also of rotational motions led to a major open research question concerning the origin of Love waves in the ocean-generated noise. For the first time Earth’s free oscillations were observed on a rotation sensor. Industrial applications. Industrial applications focused on the application of full-waveform tools to exploration type problems. This included the inversion of near surface structure for terrestrial data as well as the full 3-D acoustic inversion of the North Sea Valhall data set. The latter project led to one of the first published applications of the source-encoding method to real data. The stacking of multiple source leads to considerable speed-up of the inversion procedure and is likely to play an important role in future applications. Seismic tomography and geodynamics. With both central methodologies (full waveform inversion and passive imaging using noise) a number of new models emerged for the first time using these tools. These include new models of Europe as a whole, as well as regional applications to Turkey, Italy, Romania, Scandinavia, the Alpine Region. As stated in the original proposal, these new images with much improved spatial resolution are now leading to a better understanding of the Earth’s internal structure and the associated dynamic processes. Earthquake source processes. This work package was tremendously influenced by the best-ever recorded Mega-earthquake that happened in March 2011 in Japan and a further large subduction earthquake in Chile in 2012. For the first time 3-D rupture and wave propagation schemes could be applied to a large earthquakes in order to understand in detail the rupture behaviour. Further progress was possible in understanding seismic source processes inside volcanoes. It is fair to say that the activities in QUEST have allowed European Earth scientists to continue and extend a leading role in computational wave propagation and seismic inverse problems with a strong impact on the understanding of natural hazards (earthquakes and volcanoes) as well as the imaging and monitoring of Earth’s natural resources.