Near Real-time Estimation of the Seismic Source Parameters in a Dense Network of Recording Stations

Through the use of sparse representation theory and compressive sensing a methodology for the simultaneous estimation of the seismic source parameters (origin time, hypocenter location and seismic moment tensor) is presented. The methodology provides continuous, automatic estimations and is suitable for real-time monitoring systems. Furthermore, by working in a compressed domain, the limits of real-time applicability with respect to other existing methodologies are extended to consider larger subsurface monitored volumes with a larger number of recording stations. A numerical example in a dense network of recording stations emphasizes the advantages of the compressive sensing approach by reducing the memory requirements to handle the inversion variables from 52 Terabytes to 32 Megabytes with a consequent reduction in processing time from an estimated value of above 90 days to an average of around 42 seconds. The impact of the compression process in the resolvability and accuracy of the estimated source parameters is also discussed.