Posted by Adesh Pandey

Low-frequency Ambient Noise Tomography (LFANT) is a seismic imaging technique that relies on seismic interferometry. It uses the cross-correlation functions (CCFs) of ambient noise signals from natural sources like wind, waves, and ocean currents, collected by arrays of receivers, such as ocean-bottom nodes (OBNs), to construct virtual shot gathers (VSGs). OBNs enable long-term marine seismic recordings capturing sub-2.0 Hz ambient seismic energy. Interferometric processing on continuous OBN recordings in the Gulf of Mexico (GOM) reveals coherent sub-2.0 Hz surface-wave energy in VSGs (Figure 1, Girard et al., 2023). These VSGs depict surface-wave propagation influenced by salt bodies, identifying salt body boundaries without prior knowledge of subsurface geology (Figure 2, Girard et al., 2023).

Figure 2. Snapshots extracted from one RR VSG at different lags, overlain on a velocity model (with salt represented in pink) for reference only (from Girard et al., 2023).

LFANT involves two fundamental stages: forward modeling of CCFs and inversion. In the forward modeling stage, it computes CCFs using an implicit approach, given the impracticality of explicitly simulating seismograms for all noise sources. Green’s functions (GFs) are used to calculate CCFs based on the noise source distribution, accounting for unknown source locations. The method approximates the source Power Spectral Density (PSD) to represent the source energy distribution, considering the spatial correlation length of noise sources (Eq-1).

This forward simulation produces both time and frequency-domain CCFs. The subsequent adjoint-state inversion minimizes the misfit between observed and simulated CCFs to extract subsurface velocity structures. Through an objective function optimization process, the method updates the source signature, PSD, and geologic structure (S-wave velocity, P-wave velocity, and mass density) to match observed CCFs. Misfit components are analyzed, refined, and connected using techniques like deconvolution and sensitivity kernels, resulting in a powerful subsurface characterization approach. Currently, my research is focusing on modelling the CCF’s for offshore scenario and subsequently inverting the GOM’s ambient dataset.

The usefulness of LFANT lies in its capability to map subsurface velocity variations, providing critical insights into geological features. Despite challenges of generating and recovering low-frequency information with conventional marine seismic data, interferometric methods can effectively identify valuable low-frequency wavefield energy. This technique can effectively complement traditional seismic imaging methods by providing a cost-effective means to access deep-seated geological information, especially in challenging offshore exploration scenarios. It can also have important implications for low-frequency elastic model building with multi-component ambient seismic data for active-source seismic imaging and inversion.