Intelligent Microseismic Event Detection Using Kirchhoff Migration Style Stacking in Combination with Iterative Clustering based Segmentation

Detecting and imaging microseismic activity can be a powerful tool for reservoir monitoring, it can help in detecting and quantifying regional tectonic and production related stress buildup. However, identifying microseismic activities from individual traces/raw data is not an easy task because of their low signal-to-noise ratio. Also, the presence of numerous microseismic event locations (for example during water injection or hydraulic fracturing) can lead to erroneous identification of spurious noise as a microseismic event. In order to enhance the signal-to-noise ratio, a Kirchhoff-migration style stacking is used, where the amplitude of the event in the stacked trace gets enhanced due to constructive interferrence. To avoid the erroneous detection of these events, an Iterative Clustering based Segmentation (ICS) approach (unsupervised machine learning technique) is used. The technique is correctly able to identify the clusters of microseismic event locations in space (3D cubes/splices) at a particular instant of time for both synthetic as well as real data. The cluster identification is carried out on 3D cubes (only space) rather than 4D (space and time altogether) because of the mobile nature of the event locations, which can also be monitored using the ICS scheme.