Filter Kl/Svd For Ground Roll Noise Attenuation

In geophysics there are several steps in the study of Earth, one of them is the seismic records processing. These records are obtained through observations made on the surface, being useful to obtain information about the structure and composition of inaccessible zones in high depths. Most of the tools and techniques developed for such studies have been applied in academic research. The major problem in seismic processing is the undesired energy, registered by receivers that do not bring any information about the reflectors, that could mask the information and / or generate mistaken information about the subsurface. This unwanted energy is known as seismic noise. Attenuate this noise to improve a signal that indicates a reflection without losing desirable signal is quite complicated. This project aims to the attenuation of noise generated from surface waves, especially the ground roll, which shows a pattern characterized by low frequencies, low speeds and strong amplitudes. The Karhunen-Loève Transform is a great tool for identification of standards based on the eigenvalues eigenvectors. The Transform will be used with the singular value decomposition as it is a great technique for mathematical manipulation of data. The algorithm of the filter KL/SVD was designed to run within the graphics card, using the architecture CUDA.