Nonhyperbolic multiparametric travel-time approximation for converted-wave and OBN data

• Ano de defesa: 2021
• Autor(a) principal: Zuniga, Nelson Ricardo Coelho Flores
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: converted wave; mudança de fase; multicomponent; multicomponente; não-hiperbólico; nonhyperbolic; OBN; onda convertida; phase shift
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/14/14132/tde-11052021-202215/

To obtain a more accurate stratigraphic model, it is important to perform a reliable velocity analysis by fitting the calculated travel-time curve to the recorded one. The main challenge to perform this step in the offshore seismic survey is the fact there are many factors which present characteristics that generate nonhyperbolicity in travel-time events. For this reason, it is necessary to use an approximation which is able to control the effects of the nonhyperbolicity so that performing an efficient velocity analysis is possible. The main proposition of this thesis was to develop a generalization of a nonhyperbolic multiparametric travel-time approximation, which can control the nonhyperbolicity associated to heterogeneity of layered media and long offsets, wave conversion, and difference of datum between source and receiver. This work proposes a numerical study which aims to analyse the complexity of the objective function, the quality and efficiency of the travel-time curve fitting with different approximations, using different optimization algorithms and using the L1- and L2-norm. Three of the five models studied in this thesis were elaborated from well logs data of the Santos Basin and showed different characteristics making the proposed analysis more complex. The other two models were adapted to test some limits of the nonhyperbolic approximation (including the proposed one). Another important proposition was the development of an automated picking technique to obtain the reflection events related to the interface between the bottom of the salt and the top of the reservoir (target reflection). That technique was also part of a set of tests to obtain information of phase variation. It was possible to determine how complex each approximation is with the set of information obtained by the residual function maps analysis. The quality of the fitting and the efficiency of each approximation were also analysed. The optimization algorithm which showed the best results was determined. The difference between the use of L1-norm and L2-norm was also studied and it was determined which one is better to work for each kind of inverse problem. With all these analyses, it was possible to identify which approximation along with which optimization algorithm present the best results for each reflection event of each model after the inversion. The automated picking technique proved to be able to obtain the target reflection seismic events and information of phase variation.