Nonlinear Moment Tensor Inversion of the Envelopes and Cumulative Energy of Regional Seismic Waveforms

Linear moment tensor inversion of seismograms recorded at regional distances of a moderate magnitude earthquake is a routine process to determine seismic moment, focal mechanism and depth of an event. Here, we are working on approaching the same problem quite differently. We are testing a nonlinear moment-tensor inversion scheme to invert processed seismic waveforms. Instead of trying to find the best match between data and the synthetic, as done in the linear moment tensor-inversion, we want to fit either the envelopes or the cumulative energy curves of the seismograms, both of which require a nonlinear inversion scheme. Envelope and cumulative energy processing smooths out the waveforms and makes them less complicated as compared to the original data without frequency filtering. The proposed scheme is first tested with moderate magnitude regional earthquakes in a frequency band of 0.05-0.1 Hz to show that it can extract the desired source information. Then we plan to start including higher frequency content in the seismograms so that P-wave and S-wave as well as surface-wave energy is incorporated. We are testing a "cut-and-paste" approach to match sections of the envelopes such as the P-wave window, S-wave window, surface-wave window or even seismogram sections with multiple phases. Likewise, we plan to test the same approach using cumulative energy curve of the seismograms. Our goal is to apply this scheme to smaller magnitude events which are rich in high frequency content but lacking in observable longer period surface-wave energy. This work currently is being tested using data from the TA stations in northeastern North America in the epicentral distance of 70-400 km.