Hunting for microseismic reflections using multiplets

Discussion

We identified clear multiplets which, disappointingly, when stacked did not produce reflection events as strong on the stack as they appeared on specific individual event windows. Using the velocity log in Figure 1, a back of the envelope calculation showed three deficiencies in our algorithm:

• Our bulk shifts to the nearest sample were too coarse for the majority of arrivals. We should be using fractional sample shifts.
• Our linear moveout corrections were similarly too coarse.
• The misalignment of later reflections due to small differences in the microseismic source locations is generally larger by a factor of 2 to 4 greater than that of the direct arrivals due to the longer travelpath of reflections. So we need to include time-variant alignment, i.e. warping.

In addition to these effects, we remark that the RMS value of the noise in the stack and the direct P-arrival have almost the same value, an additional reason why P reflections were not clearly revealed after projection even though the direct P first arrival was enhanced.

On a positive note, we can use the misalignments in reflected arrival times to constrain differences in source locations, supplementing double-difference hypocenter analysis for the microseism locations (Waldhauser and Ellsworth, 2000).

Hunting for microseismic reflections using multiplets