Abstract
It is a common knowledge that proper inclinations and azimuth angle determination is a critical step in processing and interpretation of walk-away VSP data. Additionally, an in-depth analysis of the uncertainty of these interpreted values requires the introduction of measurement errors. In this contribution, we present a statistical analysis of obtained polarization angles from three-component, multi-depth level, walk-away VSP using Python 3 programming language. Our analysis is presented in the context of different processing sequences and correlation with local features of the geological medium. We show that the obtained values of polarization angles and their errors can be strongly affected by processing sequence and - when done correctly - can give addition inside into features of analysis medium. Moreover, in some cases, even a presence of saturation can be express by polarization angles variations. Additionally, we examined the impact of well-casing on interpretational values of polarization angles.
Highlights
Obtaining full anisotropy tensor using P-wave only from the walk-away VSP (Vertical Seismic Profiling) survey is not a trivial task, especially when the acquisition was gathered in challenging conditions [1]
Statistical analysis including EDA (Exploratory Data Analysis) of VSP data and obtained inclinations and azimuths are crucial steps and are not less important than proper processing of data. The purpose of this step was to find the patterns in data and prepare it for cluster analysis to decide which samples correspond to particular layers differing in acoustic properties
It was important to find the correlation between geological layers and calculated polarization angles
Summary
Obtaining full anisotropy tensor using P-wave only from the walk-away VSP (Vertical Seismic Profiling) survey is not a trivial task, especially when the acquisition was gathered in challenging conditions [1]. Statistical analysis including EDA (Exploratory Data Analysis) of VSP data and obtained inclinations and azimuths are crucial steps and are not less important than proper processing of data. The purpose of this step was to find the patterns in data and prepare it for cluster analysis to decide which samples correspond to particular layers differing in acoustic properties. The analysis presented in this paper allowed for proper preparation of data for the P-wave only inversion. The main aim of this research was to find the quantitive information about which processing scheme is the best for obtaining reliable polarization angle using PCA (Principal Component Analysis) method for components rotation. The patterns in the data were studied for further data clustering using unsupervised machine learning
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have