Abstract:
The estimation of the hypocentres of seismic events is crucial in the interpretation of subsurface structure and tectonics. The principle objective was to refine estimates of these hypocentres by analysing the full seismic wave field – rather than the P or P and S waves alone – of a cluster of events. The events we examine were originally located using their P and S wave arrivals on borehole seismometers, as part of a CO2 sequestration project from the Aneth Oil field in Utah, USA. These events were then clustered based on their P wave similarity alone. We further sub-clustered these events based on their full waveform. Similar results were obtained from the full-waveform analysis in both time and frequency domains. We then compared earthquake relocation based on double-difference tomography of the original cluster, and relocation of the events, per sub-cluster. The double difference algorithm was applied in software package HypoDD. Six sub-clusters arose from the full waveform analysis on a possible “en echelon” structure with a north-west south-east alignment. Total-cluster relocations converged to the centroid of this cluster; these hypocentre locations were deemed less likely than the original locations, given prior knowledge of the geology. However, relocations were more refined when HypoDD was run for individual clusters. Individual cluster relocations appeared to condense within their sub-cluster centroid, the pre-processing step of sub-clustering retained en echelon structure and the north-west southeast alignment, preserving spatial information. This supports our assumption that the initial locations identified using correlated P arrivals were accurate at the outset. HypoDD’s inability to accept negative station elevations in conjunction with a layered velocity model prompted us to shift the station and event locations above the surface. The presence of a low velocity zone also prompted the adjustment of the velocity model such that this zone was eradicated. This was achieved by adjusting the velocity model such that the velocity increased in conjunction with the depth throughout. Shifting the location of the earthquakes above the surface may have affected our hypocentre relocations but our epicentre relocation accuracy was kept intact. Our alteration of the geological structure through the adjustment of the low velocity zone may have also affected our hypocentre locations. However, our aims of witnessing the sub-clusters converge while keeping their en echelon structure and north-west south-east alignment was fulfilled. This suggests that the errors which originate from these approximations were minimal. The application of the double difference algorithm on a known sub-cluster at a CO2 sequestration project at Aneth field refines the location of epicentre estimations, but maintains the structure of fault systems hidden in the full seismic waveforms. In the future, further knowledge of the subsurface will allow us to better reconstruct the low velocity zone. This may result in more refined epicentre and hypocentre relocations. These newly found locations could be mapped with known fault structures, in order to find new fault regimes within these structures.