Towards real-time assessment of convergence criteria in seismic interferometry: Selective stacking of cross-correlations at the San Emidio geothermal field

The Edge of Exploration: An Edge Storage and Computing Framework for Ambient Noise Seismic Interferometry Using Internet of Things Based Sensor Networks

“…Results of assessments, if they are available in quasi-real-time, allow us to reconfigure the seismic deployment’s geometry or acquisition parameters, if necessary, to ensure a study’s success. Keeping track of data characteristics can help us address the important questions “How long does it take to extract seismic arrivals?” and “Have the virtual source gathers computed so far converged to an accurate estimate of the subsurface Green’s functions?” Additional processing schemes, including selectively stacking the “best” data windows, can be devised and implemented with RaPiERs (and other WSNs), as well [ 31 ]. By enabling real-time computing of seismic data at the edge, RaPiERs allow us to optimize data acquisition by maximizing the quality of results while also minimizing effort and cost in ANSI studies.…”

“…In a typical ANSI example, a potentially large set of sensors is deployed to record continuously for days or weeks and the time series recorded by a given station is cross-correlated with that of every other station in a series of time windows. Results for each time window are “stacked” (summed) to increase the signal-to-noise ratio and produce a “virtual source gather”, which is an estimate of the Green’s function for subsurface structure beneath the sensor array [ 27 , 28 , 29 , 30 , 31 , 32 , 33 ]. Recent deployments of sensor networks have comprised thousands of stations in remote areas with limited access to the Internet [ 31 , 34 ].…”

“…Results for each time window are “stacked” (summed) to increase the signal-to-noise ratio and produce a “virtual source gather”, which is an estimate of the Green’s function for subsurface structure beneath the sensor array [ 27 , 28 , 29 , 30 , 31 , 32 , 33 ]. Recent deployments of sensor networks have comprised thousands of stations in remote areas with limited access to the Internet [ 31 , 34 ]. Such deployments could benefit greatly from IoT-enabled wireless sensor networks capable of processing data in the field due to the fact that, while traditional seismic techniques need only record a specified time window following a known seismic source, ANSI results require the processing and stacking of many hours (or days or weeks) of data to reveal low-amplitude waves.…”

Optimal Stacking of Noise Cross-Correlation Functions

“…This concept, called selective stacking, has been implemented in previous studies (e.g. Olivier et al 2015;Thangraj & Pulliam 2021). The signal-to-noise ratio (SNR) or the correlation to the reference can be used to evaluate the quality of each NCF.…”

Optimal Stacking of Noise Cross-Correlation Functions

“…In the presence of strong noise, it could be effective to stack only a subset of the NCF ensemble that exceeds a quality threshold and to ignore the low-quality NCFs. This concept, called selective stacking, has been implemented in previous studies (e.g., Olivier et al 2015;Thangraj and Pulliam 2021). The signal-to-noise ratio (SNR) or the correlation to the reference can be used to evaluate the quality of each NCF.…”