Abstract
Focusing on a single vertical-well microseismic monitoring dataset, the aim here is to objectively look at the hypocentre location determination workflow and compare results from another independent processing effort of the same dataset. To that end, we use the Akaike information criterion-based picker followed by the cross-correlation-based refinement for P- and S-wave arrival times. We apply a particle swarm optimization algorithm to calibrate a one-dimensional velocity model using a single ball-drop event. Due to the absence of any a priori information, we obtain three different local solutions for the velocity model using the particle swarm optimization algorithm with different upper and lower bounds on the search space. We also use the particle swarm optimization algorithm to determine hypocentre locations for microseismic events. In addition, we perform a waveform-similarity-based analysis to identify clusters of closely located microseismic events. Our results show that the hypocentre locations from all three local solutions exhibit similar fracture orientations and dimensions, as compared to the results from another independent processing effort. However, random differences with mean and standard deviation values in x, y, z as (53.9, -23.2, -6.2 m) and (35.5, 42.6, 17.6 m) exist between the two processing versions. These differences can be explained by the combined effect of errors in arrival times, back-azimuths and velocity models, and the use of different algorithms in the two processing efforts. We also find that the comparison of event clusters other than full event distributions is an effective way of identifying any systematic differences between the two processing results.