Crustal Structure Imaging of Ghana from Single-Station Ambient Noise Autocorrelations and Earthquake Arrival Time Inversion

This study utilizes single-station ambient noise autocorrelation and joint earthquake arrival time inversion to generate a high-resolution crustal velocity model and seismicity catalog for southern Ghana, successfully imaging the Paleozoic basement beneath the Voltaian Basin and demonstrating the efficacy of passive seismic methods for crustal characterization in sparsely instrumented intraplate regions.

The Gist

Imagine the Earth's crust in Ghana as a giant, multi-layered cake. For a long time, geologists have been trying to figure out exactly how thick each layer is, what ingredients are inside, and where the "icing" (sediments) meets the "cake" (ancient basement rock). The problem? They only had a few "forks" (seismic stations) to poke the cake, and the cake is so big that the forks couldn't reach the bottom or see the fine details.

This paper is like a team of scientists using a new, super-smart trick to take a high-resolution X-ray of Ghana's underground without digging a single hole.

Here is the story of how they did it, broken down into simple parts:

1. The Problem: The "Blind" Cake

Ghana sits on a very old, stable part of the Earth's crust called a "craton." It's rich in resources, but we didn't know exactly what was buried deep underground. Previous maps were blurry. They knew the crust was about 40–45 km thick, but they couldn't see the specific layers inside, like where the hard rock starts or where the ancient faults (cracks in the crust) are hiding.

2. The New Trick: Listening to the Earth's Hum

Instead of using explosions or heavy trucks to create sound waves (which is expensive and invasive), the scientists used Ambient Noise Autocorrelation.

• The Analogy: Imagine you are in a quiet room, but there is a constant, low hum from the refrigerator, the wind outside, and people talking. Usually, you ignore this noise. But these scientists realized that this "background hum" (seismic noise) is actually bouncing off the layers of rock underground, just like an echo in a canyon.
• The Method: They took data from six seismic stations (microphones) in Ghana and one extra one in neighboring Côte d'Ivoire. They listened to this "hum" for two years. By using a special math trick called Phase Cross-Correlation, they filtered out the random noise and kept only the "echoes" that were bouncing back from deep underground.
• The Result: They turned these echoes into a "sonar map" of the crust, showing them the boundaries between different rock layers.

3. The Missing Piece: The Speed Limit Map

Knowing the time it takes for an echo to return is useful, but to know the depth, you need to know how fast the sound travels through the rock.

• The Problem: The old "speed limit maps" (velocity models) for Ghana were outdated and inaccurate. If you guess the speed wrong, your depth calculation is wrong.
• The Solution: The team looked at 740 small earthquakes that happened in the region. They used a "grid-search" (trying millions of combinations) to find the perfect speed model that explained how those earthquake waves traveled.
• The Upgrade: This new map showed that the top layer of the crust is actually much slower (softer) than we thought, and the deep layers have specific speed changes that tell us about the rock types.

4. What They Found: The Hidden Layers

With their new "echo map" and the accurate "speed map," they could finally see the layers clearly:

• The "Cake" Layers: They found a distinct boundary about 13–15 km deep. This looks like a giant, ancient scar or "unconformity" from billions of years ago, preserved deep underground. It's like finding a layer of old frosting that got buried under new cake.
• The Rock Switch: They found a boundary around 26–30 km deep where the rock changes from "felsic" (lighter, granite-like) to "mafic" (heavier, basalt-like). This matches what we expected from other studies, confirming their method works.
• The Bottom of the Crust: They saw the transition to the mantle (the layer below the crust) at about 41–45 km, which is the "bottom of the cake."

5. The Earthquake Detective Work

The team also used their new, accurate speed map to re-locate 740 earthquakes.

• The Clusters: They found that earthquakes aren't random; they cluster in specific zones, mostly between 8 km and 18 km deep.
• The "Brittle-Ductile" Boundary: The fact that earthquakes stop around 18 km is a big clue. It suggests that below this depth, the rock is too hot and soft (ductile) to snap and cause an earthquake. It's like the difference between snapping a cold stick of chalk (brittle) and bending warm taffy (ductile).
• The Silent Fault: They looked for activity along a famous fault called the "Coastal Boundary Fault." Surprisingly, it was silent. No earthquakes were detected there during their study. This suggests the fault might be "asleep" or locked, which is good news for earthquake risk in that specific area.
• The Connection: They confirmed that stress from the Mid-Atlantic Ridge (far out in the ocean) is traveling through the crust to cause these earthquakes in Ghana.

6. Why This Matters

This study is a game-changer for two reasons:

1. It's Cheap and Smart: It proves you don't need expensive explosions or dense networks of sensors to see deep underground. You just need to listen to the Earth's natural hum and do the math right.
2. Safety and Resources: By knowing exactly where the layers are and where earthquakes happen (and where they don't), we can better predict earthquake risks and look for resources like gold or oil more effectively.

In a nutshell: The scientists turned the Earth's constant background noise into a giant ultrasound machine, giving us the clearest picture yet of what lies beneath the feet of people in Ghana. They found ancient rock layers, mapped the "breaking point" of the crust, and discovered that some famous faults are currently taking a nap.

Technical Summary

Here is a detailed technical summary of the manuscript "Crustal Structure Imaging of Ghana from Single-Station Ambient Noise Autocorrelations and Earthquake Arrival Time Inversion", submitted to the Journal of Geophysical Research: Solid Earth.

1. Problem Statement

The crustal architecture of southern Ghana, located within the West African Craton (WAC), remains poorly resolved despite its tectonic significance and resource potential. Existing geological and geophysical studies rely on sparse seismic networks, borehole data, and offshore controlled-source surveys, which provide only broad constraints.

• Limitations of Previous Work:
  • Sparse Networks: Traditional seismological methods (e.g., receiver functions, Rayleigh wave inversion) using sparse broadband networks yield low-resolution, one-dimensional (1D) models that lack the ability to define sediment-basement interfaces or intra-crustal stratigraphy accurately.
  • Offshore Bias: High-resolution Multi-Channel Seismic (MCS) surveys are largely restricted to offshore areas and suffer from depth attenuation, failing to image deep crustal structures onshore.
  • Velocity Model Uncertainty: Existing 1D velocity models for the region are often based on uniform layer approximations or limited earthquake datasets, leading to inaccuracies when converting time-domain data to depth.
• Objective: To image the crustal structure of southern Ghana with high resolution, specifically targeting the sediment-basement interface and intra-crustal layers, and to update the regional seismicity catalog using improved velocity constraints.

2. Methodology

The study employs a dual-methodological approach combining passive seismic interferometry and local earthquake tomography.

A. Single-Station Ambient Noise Autocorrelation (SSANA)

• Data Source: Continuous vertical-component waveform data from six broadband stations of the Ghana Digital Seismic Network (GHDSN) (2012–2014) and one additional station in Côte d'Ivoire (GTSN).
• Processing Workflow:
  1. Pre-processing: Data segmented into 1-day chunks, then 1-hour windows. Instrument response removal, detrending, and band-pass filtering were applied. Two frequency bands were used: 3–13 Hz for shallow structures and 1–6 Hz for deep structures.
  2. Phase Cross-Correlation (PCC): Instead of standard amplitude-based correlation, the study utilized PCC to measure instantaneous phase similarity. This suppresses high-amplitude transients and enhances the stability of coherent reflections without needing amplitude normalization.
  3. Phase-Weighted Stacking (PWS): Hourly PCC traces were stacked using PWS (with an exponent $p=3$) to enhance signal-to-noise ratio by weighting coherent arrivals.
• Output: Zero-offset P-wave reflection responses (Two-Way Travel Time - TWT) representing acoustic impedance contrasts beneath each station.

B. 1D Velocity Model Estimation & Earthquake Relocation

• Data Compilation: An enhanced earthquake catalog was created by combining phase picks from GHDSN and the Côte d'Ivoire station (DBIC). Deep learning (DeepScan) was used for phase detection, and the PyOcto framework was used for phase association.
• Joint Inversion: A grid-search algorithm (HYP program) was used to simultaneously invert for hypocenter parameters and a 1D P-wave velocity model.
  • Model Structure: A 6-layer model with 10 parameters (layer depths, velocities, and $V_p/V_s$ ratio).
  • Optimization: A two-step search strategy (coarse then fine grid) minimized the RMS residual between observed and calculated arrival times for 3,734 phases from 453 events.
• Depth Conversion: The derived 1D velocity model was used to convert the SSANA TWT sections into depth sections using the integral relationship $\tau(z) = 2 \int_0^z \frac{1}{V_p(\zeta)} d\zeta$.
• Seismicity Relocation: Earthquakes were relocated using the NonLinLoc algorithm (probabilistic grid search) with the new velocity model to generate a high-resolution seismicity map.

3. Key Contributions

1. Novel Velocity Model: Development of a regionally constrained 1D velocity model for southern Ghana that reveals significantly lower near-surface velocities (4.90 km/s at 0–1 km) and distinct mid-crustal gradients compared to previous models (e.g., Custódio et al., 2022).
2. Passive Imaging in Data-Sparse Regions: Demonstration that SSANA, enhanced by PCC and PWS, can resolve detailed crustal stratigraphy (sediment-basement, mid-crustal interfaces) in intraplate regions with sparse station coverage.
3. Validation Framework: Successful validation of passive noise sections against active-source MCS profiles and borehole logs (Well 16.1), proving the method's ability to image lithological boundaries (e.g., Carboniferous-Devonian contact) often missed by active sources.
4. Enhanced Seismic Catalog: Creation of a revised catalog with 741 events (a 1.61x increase over previous studies) by integrating the Côte d'Ivoire station, which improved azimuthal coverage and reduced location uncertainties.

4. Key Results

A. Crustal Structure Imaging

• Sediment-Basement Interface: Clear imaging of the sediment-basement boundary beneath the Voltaian Basin and coastal basins.
• Mid-Crustal Unconformity: A laterally coherent, high-amplitude reflector identified at 13–15 km depth beneath the Birimian terrane (stations MRON, KUKU, WEIJ). This is interpreted as a Paleoproterozoic unconformity preserved beneath the Pan-African Dahomeyan belt.
• Felsic-Mafic Transition: Strong reflections observed at depths of 26–30 km (varying by station), corresponding to the transition from felsic/intermediate upper crust to mafic lower crust. This aligns with independent shear-wave velocity models.
• Moho Proximity: A deep interface zone at 32–35 km is identified as an intra-lower crustal feature or Moho-proximal transition, distinct from the actual Moho (estimated at 41–45 km).

B. Seismicity and Tectonics

• Seismogenic Depth: Seismicity is concentrated between 8–18 km depth. The upper limit of the brittle-ductile transition is estimated at ~18 km, correlating with a sharp reflector in the depth-converted noise sections.
• Fault Clusters: Eight distinct seismic clusters (A–H) were identified, aligning with the Akwapim Fault Zone (AkFZ) and Ashanti Fault Zone (AsFZ).
• Coastal Boundary Fault (CBF) Quiescence: Despite improved detection capabilities, the Coastal Boundary Fault showed no significant seismicity during the 2012–2014 period, suggesting it is currently inactive or locked.
• Tectonic Linkage: A clear structural link was identified between the inland AkFZ and the offshore Romanche Fracture Zone (RFZ), supporting the hypothesis that stress from the Mid-Atlantic Ridge is transferred through the RFZ to drive intraplate seismicity in southern Ghana.

5. Significance

• Resource Exploration: The high-resolution imaging of sediment thickness and basement configuration provides critical data for hydrocarbon and mineral exploration in the Voltaian and coastal basins.
• Seismic Hazard Assessment: The refined velocity model and relocated seismicity catalog offer a more accurate basis for probabilistic seismic hazard analysis (PSHA) in Ghana, moving beyond generic global models.
• Methodological Advancement: The study validates the efficacy of passive seismic interferometry (SSANA) combined with deep learning phase picking for imaging complex intraplate crustal structures where active sources are unavailable.
• Tectonic Understanding: The results clarify the structural continuity between the West African Craton and the Dahomeyan belt, and confirm the role of inherited lithospheric structures in controlling modern deformation in West Africa.

In conclusion, this study successfully bridges the gap between sparse seismic data and high-resolution crustal imaging, providing a new structural framework for southern Ghana that integrates passive noise imaging, local earthquake tomography, and geological validation.