Introduction
Semblance processing is a robust and reliable processing used to measure formation slowness that is of significant value for geophysical, petrophysical and geomechanical applications. Nevertheless, the vertical resolution of the computed slowness log is linked to the length of the receiver array used to record the data. Therefore it can be difficult to compare slowness logs with other measurement of higher vertical resolution. To overcome this problem, an improved first motion algorithm allowing the computation of a high-resolution sonic log has been developed. The technique considers that the part of the waveform before and after the first break can be modeled as an AR process. Next the use of the Bayesian information criterion combined with the computation of the envelope of the signal will allow for an automatic computation and estimation of the first break. We will demonstrate that this first motion technique does not require any human intervention and in most cases provides a robust and accurate transit times. This algorithm will then be applied on real data to produce a high-resolution log that will be compared with some others high vertical resolution wireline measurements
Formation slownesses are obtained through the processing of array sonic waveforms using techniques such as Slowness Time Coherence, STC (Kimball & Marzetta, 1984). STC and its most recent evolution (Endo et al, 1999) are robust techniques, widely used in acoustic logging and provide results that can be trusted in most cases. The processing is robust, automatic and provides easy evaluation of arrivals propagating across an array of sonic waveforms. An improved version of this processing was made a few years ago (Kimball, 1998) to incorporate the dispersion correction of flexural waves. Semblance technique produces confident answers in homogeneous formation; however the underlying assumption is that the slowness of the components remains constant across the length of the array. When this is not true, for example near bed boundaries, or in layered formations, the quality of the answers degrades and results can be erroneous or absent. Also, this entails that the vertical resolution of the resulting slowness log is approximately the array length, which is, 6 ft for the next generation sonic tool. Other porosity log data generally has a vertical resolution of 2 ft or less. An obvious solution to solve this problem is to decrease the array length, for example by reducing the number of receivers. However doing so al also reduces usable information and signal to noise ratio. In order to alleviate this issue and to have better consistency between the vertical resolutions of the various logging measurements, an automatic methodology for generating high-resolution slowness log data from sonic travel times with no human involvement was developed. The methodology is based on an improved first arrival detection scheme that generates travel times in a quick and efficient manner. This new first motion methodology provides quality control indicators that make it easier for the user to check and evaluate the quality of the obtained results.
