Influence of non-uniform noise source distribution on ambient noise imaging: Insights from 2D numerical simulations

Ambient noise imaging reconstructs the Green functions of subsurface structure from cross-correlation functions between seismic station pairs to image the Earth's interior. This method differs from traditional seismic approaches in that it does not rely on earthquakes or artificial sources, offering key advantages such as cost-effectiveness and the capability for continuous monitoring. As a result, it has become an essential method in studies of subsurface velocity structures, anisotropy, attenuation, and temporal velocity variations. However, it is usually assumed that noise sources are uniformly distributed, which overlooks the spatial heterogeneity and temporal variation of noise sources in real field conditions. Such idealized assumptions may introduce substantial errors into the retrieval of Green functions and adversely affect subsequent seismic imaging. Here, we employ 2D seismic wavefield simulations to evaluate the impact of different noise source distributions on Green functions reconstruction. Our results show that source heterogeneity leads to travel-time discrepancies and waveform distortions in cross-correlation functions, ultimately resulting in biased velocity structure imaging and inaccurate velocity change monitoring. Building on these results, we further explore potential correction strategies through model tests.