SANS: Publicly available daily seismic ambient noise source maps on a regional to global scale

To improve methods in full-waveform ambient noise tomography and monitoring it is important to have knowledge of the spatio-temporal variations of the noise source distribution. Without this knowledge, an uneven distribution of sources may bias observations, and a changing source distribution may be falsely interpreted as subsurface velocity changes. By combining two methods to locate noise sources and decreasing the computational cost, we are able to invert for the global noise source distribution of the secondary microseisms on a daily basis. Additionally, we present a web framework where the Seismic Ambient Noise Source (SANS) maps are made available to the public. 

Many different methods to locate ambient noise sources have been developed. Bowden et al. (2021) show how a more data-driven Matched-Field Processing (MFP) approach and a more rigorous finite-frequency sensitivity kernel method can be derived from one another.  Igel at al. (2021) implement spatially variable grids and pre-computed wavefields to make the finite-frequency inversion more efficient. This has made daily inversions on a regional to global scale feasible for secondary microseismic noise sources in a frequency range from 0.1 to 0.2 Hz. Since the inversion approach allows for prior information to be implemented, we use the more efficient MFP method to create an initial model and steer the inversion in the right direction. 

In collaboration with the Swiss National Supercomputing Centre (CSCS) we are able to run this workflow on a daily basis. The resulting noise source maps are subsequently made available to the public through our web framework SANS. A user can look through all iterations of the inversions, download all model and inversion files, and implement them in their own methods. Additionally, code is provided to help the user create plots and simplify the implementation in other studies. We are looking for collaboration with ambient noise tomography studies to investigate how the implementation of noise source maps could potentially improve the resulting structure models.