Rapid finite-frequency microseismic noise source inversion at regional to global scales

Ambient noise cross-correlations can be used as self-consistent observables, opening novel possibilities for investigating ambient noise sources. To optimize the forward-modelling of global ambient noise cross-correlations for any given distribution of noise sources in the microscismic frequency range up to 0.2 Hz, we implement (i) pre-computed wavefields and (ii) spatially variable grids. This enables rapid inversions for microseismic noise sources based on finite-frequency source sensitivity kernels. We use this advancement to perform regional and global gradient-based iterative inversions of the logarithmic energy ratio in the causal and acausal branches of microscismic noise cross-correlations. Synthetic inversions show promising results, with good recovery of the main dominant noise sources of the target model. Data inversions for several consecutive days at the beginning of October 2019 demonstrate the capability of inverting for the spatio-temporal variations of the sources of secondary microseisms in the ocean. This paves the way for daily ambient noise source inversions which could help improve full-waveform ambient noise tomography and subsurface monitoring methods.

Automated summary

By optimizing the forward modeling of global ambient noise cross-correlations and implementing rapid inversions, the main dominant noise sources of the target model can be effectively recovered. Data inversions show promising results for inverting spatio-temporal variations of secondary microseisms in the ocean.