Earthquake Monitoring with the MyShake Global Smartphone Seismic Network

Asaf Inbal, Qingkai Kong, Richard M. Allen, & William H. Savran

Submitted August 14, 2017, SCEC Contribution #7570, 2017 SCEC Annual Meeting Poster #074

Smartphone arrays have the potential for significantly improving seismic monitoring in sparsely instrumented urban areas. This approach benefits from the dense spatial coverage of users, as well as from communication and computational capabilities built into smartphones, which facilitate big seismic data transfer and analysis. Advantages in data acquisition with smartphones trade-off with factors such as the low-quality sensors installed in phones, high noise levels, and strong network heterogeneity, all of which limit effective seismic monitoring.

Here we utilize network and array-processing schemes to asses event detectability with the MyShake global smartphone network. We examine the benefits of using this network in either triggered or continuous modes of operation. A global database of ground motions measured on stationary phones triggered by M2-6 events is used to establish detection probabilities. We find that the probability of detecting an M=3 event with a single phone located <10 km from the epicenter exceeds 70%. Due to the sensor’s self-noise, smaller magnitude events at short epicentral distances are very difficult to detect. To assess detectability with dense smartphone arrays, we employ array back-projection techniques on broadband simulated data. In this class of methods, the array is used as a spatial filter that suppresses signals emitted from shallow noise sources. Filtered traces are stacked to further enhance seismic signals from deep sources. We benchmark our technique against traditional location algorithms using recordings from California, a region with large MyShake user database. We find that locations derived from back-projection images of M~3 events recorded by >20 nearby phones closely match the regional catalog locations. We examine how location uncertainties vary with user distribution and noise levels. To this end, we have developed an empirical noise model for the metropolitan Los-Angeles (LA) area. We find that densities larger than 100 stationary phones/km2 are required to accurately locate M~2 events in the LA basin. Given the projected MyShake user distribution, that condition may be met within the next few years.


Citation
Inbal, A., Kong, Q., Allen, R. M., & Savran, W. H. (2017, 08). Earthquake Monitoring with the MyShake Global Smartphone Seismic Network. Poster Presentation at 2017 SCEC Annual Meeting.


Related Projects & Working Groups
Seismology