Imaging the Northern Los Angeles Basins with Autocorrelations

Caifeng Zou, Robert W. Clayton

TL;DR

Researchers at Caltech have developed a refined method to map the "basement" of sedimentary basins north of Los Angeles using the seismic "hum" of the Earth and distant earthquakes. By applying a "common mode removal" technique to autocorrelations of seismic data, they produced high-resolution cross-sections that reveal how these basins might channel and amplify energy from the San Andreas Fault.

What Problem Does This Paper Address?

When a major earthquake hits the San Andreas Fault (SAF), the energy doesn't just radiate outward uniformly. Sedimentary basins—deep "bowls" filled with soft soil and rock—act like acoustic resonators. They trap seismic waves, causing them to bounce around and amplify, leading to much stronger and longer-lasting shaking than on solid bedrock.

The San Gabriel, Chino, and San Bernardino basins are critical "waveguides" that could funnel earthquake energy directly into the Los Angeles metropolitan area. However, our current 3D models of these basins are often under-constrained. If we don't know exactly how deep or wide these basins are, our earthquake hazard simulations (like ShakeOut or CyberShake) might be significantly underestimating the risk to urban infrastructure. This paper seeks a low-cost, high-resolution way to "see" the bottom of these basins.

Methods

The authors utilize seismic autocorrelation, a technique that essentially turns every single seismometer into its own source-receiver pair.

  1. The Principle: When a seismic wave (from ambient noise or a distant earthquake) hits the bottom of a basin, it reflects back up. By correlating a signal with itself (autocorrelation), researchers can identify these "echoes" to determine the two-way travel time to the basin's floor.
  2. Common Mode Removal: The biggest challenge in autocorrelation is the "source time function"—the messy signature of the original wave that obscures the subtle reflections. The authors used a "common mode" approach: they averaged the autocorrelations across an entire line of sensors to estimate the source signature and then subtracted it from each individual sensor. This "cleans" the data, leaving behind the specific reflections of the subsurface structure.
  3. Horizontal vs. Vertical: Interestingly, the study found that horizontal components (Radial and Transverse) produced much clearer images than the standard vertical component, likely due to the higher signal-to-noise ratio of shear waves in these environments.

Datasets

The study utilized data from the Basin Amplification Seismic Investigation (BASIN) project:

  • Nodal Arrays: 10 temporary lines consisting of 758 "nodes" (compact, three-component geophones).
  • Spacing: Sensors were placed 250–300 meters apart, providing much denser coverage than traditional permanent seismic stations.
  • Signal Sources:
    • Ambient Noise: Roughly one month of continuous background vibrations.
  • Teleseismic Events: Distant, large earthquakes ($M_w \geq 6.0$) from places like Chile, Argentina, and Papua New Guinea.

Key Insights and Ideas

  • One Earthquake is Enough: One of the most surprising findings was that a single well-chosen distant earthquake could produce a subsurface image nearly as good as a full month of ambient noise data. This suggests that rapid imaging could be done with very short deployments if a large quake occurs elsewhere in the world.
  • Complementary Strengths: While traditional "Receiver Functions" (another seismic imaging method) are great at seeing deep structures like the Moho (the crust-mantle boundary), they often struggle with shallow basin details. Autocorrelations proved to be a powerful independent tool for mapping the sediment-to-rock interface.
  • Hidden Features: The method successfully identified the Red Hill Fault (a thrust fault in the Chino Basin) and a "slab-like" feature in the San Gabriel Basin that hadn't been clearly captured by previous comprehensive models.

Key Results

  • Validation: The autocorrelation cross-sections showed an "impressive agreement" with existing basin depth models derived from gravity and borehole data.
  • Basin Geometry: The images clearly delineated the deep San Gabriel Basin and the relatively shallower Chino Basin.
  • Resolution: The 250m sensor spacing allowed the researchers to see lateral variations in the basin floor that are often smoothed over in regional models. For example, they captured the steep deepening of the basement across the Raymond Fault.

Limitations and Future Work

  • The "Flatness" Problem: Because the "common mode removal" technique subtracts features that are the same across all stations, it inadvertently filters out perfectly horizontal geological layers. The authors note that this method is best suited for basins with sloping sides or irregular bottoms.
  • Deep Imaging: The method struggled to image the Moho (30+ km deep). The authors suspect the "multiples" (repeated echoes) from the bottom of the shallow basins are so strong that they mask the much deeper reflections from the base of the crust.
  • Future Directions: The authors suggest that integrating these high-resolution basin shapes into future 3D velocity models will significantly improve the accuracy of ground-shaking predictions for Los Angeles.

Summary updated

Abstract

Abstract We show reflectivity cross sections for the San Gabriel, Chino, and San Bernardino basins north of Los Angeles (LA), California, determined from autocorrelations of ambient noise and teleseismic earthquake waves. These basins are thought to channel the seismic energy from earthquakes on the San Andreas fault to LA, and a more accurate model of their depth is important for hazard mitigation. We use the causal side of the autocorrelation function (ACF) to determine the zero-offset reflection response. To minimize the smoothing effect of the source time function, we remove the common mode from the autocorrelation to reveal the zero-offset reflection response. We apply this to 10 temporary nodal lines consisting of a total of 758 geophones with an intraline spacing of 250–300 m. We also show that the ACF from teleseismic events can provide illumination on the subsurface that is consistent with ambient noise. Both autocorrelation results compare favorably to receiver functions.
BibTeX
@inproceedings{zou2024imaging,
  author = {Caifeng Zou and Robert W. Clayton},
  title = {{Imaging the Northern Los Angeles Basins with Autocorrelations}},
  booktitle = {Seismological},
  year = {2024},
  doi = {10.1785/0220240140},
}
Figure