IMS location capability improvement with the ambient noise tomography

The Preparatory Commission for the CTBTO routinely process data from the International Monitoring System, IMS – a global network of seismic, hydro-acoustic, and infrasound stations. The data are processed to detect, locate, and screen events that may have characterization parameters similar to those from nuclear explosions. The observation and processing systems are required to be sensitive to low-magnitude events, especially in unusual locations (e.g., aseismic regions). A promising way to improve the system sensitivity is by refining the receiver velocity models underneath IMS stations by incorporating a number of ambient noise processing techniques into the International Data Center (IDC) practice. In particular, this approach should lead to reduction of the arrival time residuals between empirical and observed onset times of seismic waves. The Big Data basis for this approach is using a vast amount of seismic noise data acquired in the IDC for more than 20 years. It would also allow to shed a light on the existence of seismic velocity evolution at least for unstable crustal regions applying a time-lapse ambient noise tomography (ANT) method (4D high resolution passive seismic). A lack of reference models can be partially overcome and examining the models within the seismic array aperture can be performed by the convergence of the spatial seismic correlation methods and the local single station measurements - seismic impedance and the direct Rayleigh ellipticity estimations by the H/V ratio and random decrement techniques

We conducted a case study for ARCES IMS array-station in Northern Norway, which consists of 4 rings of all 3C broadband (120s shallow vault   seismometers. Besides building an averaged uppermost ARCES velocity model, we demonstrate the trial application of the ANT methods for the individual model retrieval at different flanks of spatially distributed sensors comprising seismic arrays as a generalized way to aggregating the block velocity models.  Modified spatial autocorrelation (MSPAC) has been applied for ARCES data both for the whole set of elements as well as for four geographically symmetrical sub-groups relative to the array center. Spatial correlation patterns demonstrate the Bessel function (relative to the ground motion frequency) behavior as predicted by Aki (1957). The cross-correlation analysis of the background noise at ARCES was carried out in the wide frequency range because of the broadband hybrid channel frequency response at each array element.  Revealed models demonstrate considerable difference and thus could be further utilized for improvement of event location and as a station specific correction instrument.

Also, we provide an example with the spiral geometry but smaller aperture seismic array in Norcia intermountain basin, Northern Italy. The model estimation based on MSPAC conducted with the medium range sensors provides the results consistent with the well and gravity study conducted in Italy (2019).

For enhancement of CTBTO OSI aftershock monitoring system, the same approach can be utilized by retrofitting velocity models produced with the noise data collected from the temporarily OSI array. The same method could be also implemented in hydrofracking and induced seismicity monitoring.