Introduction

"data! data! data!" he cried impatiently, "I can't make bricks without clay."

Sherlock Holmes
A. Conan Doyle, The Adventure of the Copper Beeches (1892)

The Earth's mantle, a silicate rich layer from depths of $\sim$30 km to $\sim$2900 km, holds the answer to many open questions of the dynamics of the Earth, e.g. plate tectonics, intraplate volcanism, mid ocean ridges and orogenesis. The engine of all these processes is the convection of mantle material, the creeping flow of hot material rising, which cools at the surface and subsides back into the mantle, driven by buoyancy forces (Oxburgh and Turcotte, 1978). The mantle convection is the result of the heat stored in the Earth's core, which heats the core-mantle boundary, and the heat generated by the decay of radioactive isotopes in the mantle material. The lower mantle, excluding the D''-layer directly at the core mantle boundary, is probably very homogeneous as a result of continuous mixing processes of the mantle material, but the upper mantle shows more structure.
The layered structure of the Earth is known since the beginning of the 20th century. The early, rough division into crust, mantle and core has been refined and new subdivisions have been found ever since (Oldham, 1906; Mohorovicic, 1909; Gutenberg, 1926; Lehmann, 1936; Adams, 1968; Hales, 1969).
There are several geophysical and geological methods which enable the investigation of the upper mantle structure. Most methods are restricted to the shallowest part of the upper mantle. Therefore, the shallow depths are well known, but the deeper upper mantle is not resolved in great detail yet. However, the deeper upper mantle can be studied by various methods, e.g. by seismology or magnetotellurics. Especially seismologists have developed many methods to study the structure and composition of this part of the Earth's interior.
A prominent feature of the upper mantle structure, down to depths of $\sim$700  km, is the existence of discontinuities, changes of the parameters of the mantle material, e.g. density, electrical conductivity, and seismic velocities. Seismology is ideally suitable for the study of the current structure of the discontinuities, because seismic waves are reflected, refracted and converted at the boundaries. Numerous methods have been applied to recordings of earthquakes to use the information contained in these seismic phases to resolve the structure of the discontinuities.
The results of seismology are most valuable in combination with high-pressure and high-temperature laboratory experiments of mantle materials and the study of xenoliths, mantle material fragments carried up from depths smaller than $\sim$200 km by basaltic magmas (Birch, 1952; Bina and Wood, 1987; Karato, 1997; O'Neill and Palme, 1998). A combination of seismology and laboratory experiments solved the origin of at least some of the discontinuities as the result of solid-solid phase transitions of the mantle materials. The mechanisms for the generation of other discontinuities, e.g. chemical boundaries, compositional boundaries or thermal boundaries, are still not understood in detail. Sometimes the laboratory experiments and seismological studies produce inconsistent results about formation and dynamics of upper mantle discontinuities (Anderson, 1967; Adams, 1968), indicating that our present picture of the upper mantle might be too simplified (Jeanloz and Thompson, 1983; Helffrich and Wood, 1996).
The knowledge of the structure of the upper mantle discontinuities down to depths of $\sim$400 km and the discontinuities of the mantle transition zone between depths of $\sim$400 km and $\sim$700 km is important to determine the thermal and chemical structure of the mantle and to solve controversial questions of mantle chemistry and dynamics.
Because of the importance of the upper mantle and transition zone discontinuities, the investigation of these structures started early in the history of seismology (Byerly, 1926; Gutenberg and Richter 1934; Jeffreys, 1936) and is still in the focus of several recent studies (Flanagan and Shearer, 1999; Gaherty et al., 1999; Li et al., 2000).

This thesis studies the structure of the upper mantle and transition zone discontinuities in a corridor located in the Pacific stretching from the Hawaiian Islands to the Sea of Okhotsk. The reflections of compressional waves from the discontinuities recorded at a permanent seismological array in Canada are used to resolve the depth and structure of the discontinuities. This source-receiver combination enables the detailed study of the geodynamically interesting regions of the postulated mantle plume beneath the Hawaiian Islands (Wilson, 1963; Morgan, 1971), as well as the Kurile subduction zone south of Kamchatka. The use of high quality short-period array data offers the possibility of a high resolution image of the discontinuity topography and more detailed information on other parameters of the discontinuities than possible with other datasets and methods previously used. The aim of this study was to prove that the reflected compressional phases are detectable by short-period arrays, and to produce detailed topography maps of the different discontinuities. The improvement of standard array techniques applied to the dataset enabled both. The emphasis lay on the major discontinuities of the mantle transition zone at 410  km and 660  km depth, but the study revealed evidence for other, previously undetected, discontinuities at shallower depths in the northwestern Pacific.
In a minor part of this thesis the anisotropic structure of the upper mantle was studied using data from a temporary array in Canada and shear-wave splitting studies.
Chapter 2 describes the present knowledge about the structure of the upper mantle discontinuities and discusses the possible explanations for the different discontinuities. In chapter 3, the methods and techniques previously used to study the upper mantle are introduced and existing shortcomings are discussed. The two arrays and the earthquake dataset used in this study are described in chapter 4. In chapter 5 the different processing methods applied to the data are explained in detail, resolution tests for the newly developed sliding-window fk-analysis are presented and possible error sources are discussed. The results of the thesis are presented in chapter 6 and they are discussed in the following chapter, considering geodynamical implications of the results and the results are compared with previous studies. Finally, chapter 8 concludes the main results of this thesis.