← Back to Sessions

Effects of the Distribution of Ambient Noise Sources in Subsurface Models Inverted From Noise Correlations

Description: 

Cross-correlations of seismic ambient noise are often used to image subsurface structure. Although it is possible to account for the distribution of noise sources and treat noise correlations as self-consistent observations, most studies currently assume that noise sources are uniformly distributed and interpret noise correlations as empirical Green's functions. However, this assumption is not always correct, as noise sources are often localized and unevenly distributed. In this work, we investigate how the treatment of the noise source distribution changes subsurface models obtained from noise correlations. Our main focus is to study how inverted models change if a realistic heterogeneous noise source distribution is either incorrectly assumed to be uniform or properly taken into account. Furthermore, we explore the consequences of ignoring distant noise sources in a regional tomography, which may be necessary to avoid excessive computational requirements. To reach these objectives, we conduct a series of 2-D synthetic inversions for local subsurface structure, and thereby imitate a local-scale experiment exploiting ocean noise. The synthetic dataset consists of noise correlations computed with a large-scale noise source distribution and a laterally heterogeneous Earth structure model. We invert this dataset using three approaches, each dealing with the noise source distribution differently. Additionally, we repeat the experiments using a second synthetic dataset generated with a different noise source distribution to investigate how the estimated subsurface varies artificially due to the Green's function approximation. Our results demonstrate that the Green's function approximation introduces errors in the inverted models with a magnitude that depends on the distribution of noise sources. Since the location of noise sources changes over time, this suggests that model errors are also time-dependent. In contrast, source-related errors are avoided if the noise source distribution is accounted for. However, all noise sources, including those located away from the area of interest, must be properly considered.

Session: 3D Wavefield Simulations: From Seismic Imaging to Ground Motion Modelling [Poster Session]

Type: Poster

Date: 5/2/2024

Presentation Time: 08:00 AM (local time)

Presenting Author: Eduardo

Student Presenter: Yes

Invited Presentation: 

Authors