Strain-based forward modeling and inversion of seismic moment tensors using distributed acoustic sensing (DAS) observations

This study used a waveform inversion of distributed acoustic sensing (DAS) data, acquired in two horizontal monitoring wells, to estimate the moment tensor (MT) induced microearthquakes. An analytical forward model was developed simulate far-field tangential strain generated by an MT source homogeneous and anisotropic medium, averaged over gauge length along fiber arbitrary orientation. To prepare data for inversion, secondary scattered waves were removed from field observations, using f - k filtering time-windowing. The modeled observed primary arrivals aligned cut-and-paste approach. parameters inverted via least-squares approach, their uncertainties determined through bootstrap analysis. Using simulated with additive noise derived same configuration as method adequately resolved MT. Despite assumption Gaussian noise, which underlies robust presence heavy-tailed data. When applied independent results P-waves, S-waves, both together yielded that consistent between events different wave types. agreement resulting stress illustrated reliability method. small enough make useful geophysical interpretation. variance reduction obtained predicted most probable satisfying, even though polarity P-waves not always correctly reproduced.