Aligning microseismic reflections for imaging

Introduction

The ultimate goal of our research is to use reflections of hydraulically induced microseisms in the Bonner sand of the Bossier play to image the subsurface. In our previous work (Farghal and Levin, 2012), we successfully utilized a method adapted from earthquake seismology to find events originating from closely related sources which produced almost identical seismograms. This method involved cross-correlation of a particular event waveform (called a master) consisting of a P and S direct arrival with the whole dataset to find replicas of this waveform (the collection of which is known as a multiplet). Upon identification of similar waveforms (sources), we aligned and stacked their seismograms together to decrease the data size and increase the S/N ratio. Since we were only able to identify S reflections, we were also hoping that by stacking similar seismograms we will boost companion P reflections as well. However, for reasons discussed later in this report, this first attempt was not fruitful.

In SEP-147, we noticed that reflections are weaker on the stack than they are on the individual seismograms. We proposed that this was due to misalignment of the reflections after we aligned the direct arrivals. We attributed this to misalignment of the reflections that may differ by a small amount due to small source location differences in cases of similar rather than identical/coincidental sources. We have since realized that the fracturing and pressure changes could have affected propagation velocities even if the source locations happen to be the same.

In this report, we will show how we successfully address the misalignment problems and enhance the stacked amplitude of reflections. Moreover, we attempt to find P-reflections by cross-correlating P direct arrivals with the whole seismogram in which it lies. The faintest P reflection may well be useful when we come to the migration stage.

Finally, in preparation for imaging, we apply the previously mentioned concepts (of cross-correlations, alignment and stacking) to the whole Bonner dataset.

Aligning microseismic reflections for imaging