Speaker: Loïs Letellier

Date: Thursday 10th of April 2025, 1:15pm.

Abstract:

Well log interpretation is a time consuming task, which, moreover, depends on the interpreter's experience, as there exists different admissible interpretations for a same input log. To account for the uncertainty of the interpretation, we propose to use a signal processing technique, the Continuous Wavelet Transform (CWT), that gives information on the signal considering both depth and scale parameters. The result of this transform is then processed to extract some features of the signal (i.e. the local extrema of the signal at different scales), and propose different geological scenarios. We then compare the different solutions with experts' interpretations. 
These different scenarios are destined to be included in a correlation process, in order to study correlations with a multi-scale approach, which allows to explore the range of possibilities, while reducing the computational cost using a hierarchical process.