Real-Time Ambient Noise Subsurface Imaging in Distributed Sensor Networks

Ambient Noise Seismic Imaging (ANSI) is a recently developed geophysical methodology to image the shallow subsurface structures of earth using ambient/environment noise as the source. Integrating ANSI computing within distributed sensor networks will enable real-time continuous monitoring of subsurface dynamics for sustainability studies. However, the research challenges associated with this innovative approach are significant. Traditional data collection using sensor networks imposes practical difficulty for real-time applications, because of the sheer amount of data and large-dense sensor arrays versus limited network bandwidth. This paper is the first to investigate how to utilize the computing capabilities of sensor nodes to perform the computation of ANSI under resource constraints. We explored two distributed approaches (aggregation and consensus) for computing ambient noise eikonal tomography and obtaining phase velocity maps. We performed experiments using CORE emulator to obtain phase velocity maps on real data from USArray Transportable Array. Results demonstrate that our approaches can illuminate phase velocities under network constraints. We also show that the proposed aggregation and consensus algorithms not only balance the computation load but also achieve low communication cost and high data loss tolerance.