00457 ðc9724a5349b9ca020fa0b800a81195

462

Russell

The Geophysical Sonic Log Data Set

The author's first experience with the univariate phase map movie was with geophysical sonie log data. Sonic logs are a measure of the transit time of sound through rock surrounding a well-bore, and they are generally plotted against depth. In generaÅ‚ these logs are used by geoscientists, through yisuai signature extraction, to track formations in the earth from well to well. Quite often there exist a large number of layers with visually distinct data character, making this practice possible.

In the initial examinations of any sonie log data, it is elear that if it is possible to fit a model to the data, the model has to change frequently. So, at best piecewise fitting is all that is possible. In fact we might typically expect to see a change in the fitted model every 20 to 30 data points. It is also expected that sonie log data should be nearly 100% signal - there is almost no noise. And it is elear that there should be a strong autocorrelation in the data. This is due to the manner in which the data is collected, the measurements in the case studied here were averaged over a two foot distance with a sample taken every 0.5 foot in depth.

The author eventually accessed a number of sonie logs, from the same region, that had been declared as uninterpretable. As was the standard practice in this particular part of the world, there was also 3000' of core taken from the well borÄ™ itself. The core was fully interpreted by geoscientists as to type of depositional environment by depth in each of the wells. It was this information from the interpreted core that was being used to track layers from well to well in the region.

The equivalent of a time-order plot is presented in FigurÄ™ 24 below. This plot clearly shows the ever-changing naturÄ™ of this data over depth.

When viewed as a univariate phase map movie, the rich naturÄ™ of the autocorrelation models becomes readily apparent. It is also elear that the models do change rapidly with depth as expected for a geological setting. It is also seen in the movie that some of the pattems clearly repeat at different depths. This is also expected in a geological environment.

Upon comparison of some of the morÄ™ outstanding repeating pattems, such as the offshore bar in Figures 25 and 26 and the sharp-based shelf ridge in FigurÄ™ 27, to the interpreted core it was noticed that there was a 1-1 correspondence between the pattem and the interpreted depositional environment. In addition the same signatures were found to hoÅ‚d with the same interpretations in nearby wells.

Based on this, it appears that not only was the sonie log data incorrectly labeled as uninterpretable, it actually contained features that could