Signal architecture using phase coefficient of the wavelet transform

In the petroleum exploration industry, locating and characterising reservoir rocks rely on the interpretation of a large number of data of different types and spatial resolution. Among them, wireline logs, physical measurements recorded at a regular depth sampling rate in the well, are of utmost importance. Geologists use them to evaluate the hydrocarbon saturation of the porous volume of the rock and to describe the 3D architecture of the reservoir. To perform the latter task, geologists look for a multi-scale organisation of the log, with depth as a reference, that reflects the self similarity of the geological processes (trend), punctuated by "major events" (breaks). In so doing, geologists work in a way very similar to the manner a wavelet transform analyses a signal: both look at the signal at different scales, detect breaks and heterogeneity and characterise trends. The method proposed here is based on this similarity and is meant to emulate the geologist\´s interpretative behaviour. Tests have indicated that the use of the continuous wavelet transform and the organisation of the phase coefficients are the most promising ways to document a multi-scale architecture of an analysed signal. Then, an automatic determination of the signal architecture is possible without the need of a priori knowledge and without parameters