Processing of multi‐well offset vertical seismic profile data acquired with distributed acoustic sensors and surface orbital vibrators: Stage 3 of the CO2CRC Otway Project case study

Abstract

ABSTRACTTime‐lapse seismic reservoir monitoring is used as a way to gain insight into subsurface processes. Yet, the application of standard 4D technology onshore faces challenges, such as high cost, significant environmental footprint and, consequently, relatively infrequent surveys. As part of the Otway Project Stage 3 CO2 injection study, continuous automated borehole‐based monitoring using distributed acoustic sensing has been paired with permanently deployed surface sources, referred to as surface orbital vibrators, as a way to monitor the spreading CO2 plume. The injection of 15,000 tonnes of CO2 in a saline reservoir at a depth of 1550 m is monitored using five boreholes instrumented with enhanced sensitivity fibre optic cables and nine surface orbital vibrators, creating an array of 45 well–source pairs. The data are processed with an offset vertical seismic profiling processing workflow developed to address key challenges of the continuous distributed acoustic sensing acquisition using surface orbital vibrators. The processing flow includes deconvolution with a source sweep recorded by a pilot geophone installed below the surface orbital vibrators. A second deconvolution with a wavelet estimated from direct arrivals compensates for the difference between distributed acoustic sensing measurements and the pilot geophone as well as near‐surface variations. Image quality is noted to be best for short offsets and decreases with increasing offsets and well deviations. As surface orbital vibrators generate unique sweeps in two rotation directions, further processing is applied to stack these rotation signals together, which further improves the images. The resulting 2D transects of each well–source pair visually provide good illumination of the subsurface, suggesting continuous monitoring of the spreading CO2 plume should be possible with some further tuning of the processing workflow for time‐lapse repeatability.