Algorithmic Detection of Subsurface Voids: Case Studies in Urban Geophysical Surveys

The Multi-channel Analysis of Surface Waves (MASW) methodology represents a significant advancement in the field of near-surface geophysics, particularly in the characterization of subsurface anomalies. Developed by researchers at the Kansas Geological Survey (KGS) in the late 1990s, this technique utilizes the dispersive properties of Rayleigh waves to produce shear-wave velocity profiles of the earth's shallow subsurface. By analyzing the velocity-frequency relationship, geophysicists can infer the mechanical properties of geological layers and detect localized irregularities such as voids, sinkholes, and anthropogenic structures.

In urban environments, the detection of subsurface voids—whether occurring naturally as karst features or resulting from historical mining and infrastructure decay—is critical for risk mitigation and engineering design. The MASW approach provides a non-invasive means of mapping these hazards by measuring ground-motion signatures via a linear array of geophones. Subsequent data processing involves the extraction of dispersion curves and the application of inversion algorithms to translate wave velocities into depth-dependent lithological models.

At a glance

• Primary Methodology:Multi-channel Analysis of Surface Waves (MASW), utilizing active and passive seismic sources.
• Key Research Origin:Developed by Choon Park, Richard Miller, and Jianghai Xia at the Kansas Geological Survey circa 1998.
• Target Anomalies:Karst voids, abandoned coal mines, buried utilities, and foundation instabilities.
• Signal Type:Vertically polarized Rayleigh waves (surface waves) and, in specific cases, Love waves.
• Primary Output:1D and 2D shear-wave velocity (Vs) profiles used for lithological characterization and engineering assessments.
• Depth Sensitivity:Typically ranges from 1 meter to over 30 meters depending on source energy and array length.

Background

Before the standardization of MASW, the analysis of surface waves was primarily conducted using the Spectral Analysis of Surface Waves (SASW) method. SASW, developed in the 1980s, relied on a two-channel setup that often proved sensitive to signal-to-noise ratio challenges and the interference of higher-order modes. The transition to multi-channel acquisition in the late 1990s allowed for the simultaneous recording of seismic data across 12, 24, or 48 geophones. This redundancy improved the isolation of the fundamental mode Rayleigh wave from body waves and environmental noise, making the technique viable for complex urban settings.

Rayleigh waves are characterized by an elliptical retrograde motion of particles at the surface. In a heterogeneous medium, these waves are dispersive, meaning different frequencies travel at different phase velocities. Higher frequencies generally probe shallower depths, while lower frequencies penetrate deeper into the earth. By accurately measuring this dispersion, researchers can reconstruct the vertical shear-wave velocity structure of the site. Because shear-wave velocity is directly related to the shear modulus of the material, it serves as a proxy for stiffness and density, which are critical for identifying voids.

The MASW Procedural Workflow

The standard MASW workflow is divided into three distinct phases: data acquisition, dispersion analysis, and inversion. During acquisition, a seismic source, such as a sledgehammer or an automated weight drop, generates waves that are captured by a geophone array. In urban surveys, where space is constrained, the spacing of the geophones determines the resolution and the depth of investigation. Closely spaced sensors provide high-resolution images of the shallowest layers, while wider spacing is required for deeper penetration.

Dispersion analysis involves transforming the recorded time-domain data into the frequency-phase velocity domain using techniques such as the Wavefield Transform. This transformation produces a "phase velocity spectrum" or a "dispersion image," where the energy of the Rayleigh waves is concentrated along specific paths known as dispersion curves. Identifying the fundamental mode curve among higher-order modes is a critical step that requires precise algorithmic filtering.

The final phase, inversion, is a mathematical process where an initial earth model is iteratively adjusted until its theoretical dispersion curve matches the observed data. This process is inherently non-linear and requires sophisticated optimization algorithms to ensure the final model accurately reflects the actual subsurface conditions rather than falling into a local mathematical minimum.

Algorithmic Adaptations for Void Detection

Detecting a subsurface void presents a unique challenge because a void represents a significant contrast in elastic moduli compared to the surrounding host rock or soil. For instance, air-filled or water-filled karst voids exhibit zero or very low shear-wave velocities. When a surface wave encounters such an anomaly, the wavefield undergoes scattering, diffraction, and resonance. Standard 1D inversion algorithms, which assume laterally homogeneous layers, often struggle to resolve these localized 3D features.

Karst and Abandoned Mine Case Studies

Case studies in regions prone to karst topography, such as the Midwestern United States and parts of Southern Europe, have demonstrated the efficacy of adaptive MASW algorithms. In these projects, researchers often employ a "rolling spread" acquisition technique, where the entire geophone array is moved progressively along a transect. This creates a series of 1D profiles that are stitched together to form a 2D cross-section. Algorithmic enhancements, such as the inclusion of lateral constraints, help to stabilize the inversion process in areas with abrupt geological transitions.

In the context of abandoned mines, the presence of rubble-filled galleries or partially collapsed rooms creates a complex seismic signature. Studies have shown that voids can cause a noticeable shift in the phase velocity of the dispersion curve. Specifically, the presence of a void typically results in a localized decrease in phase velocity at frequencies whose wavelengths are comparable to the depth of the void. Advanced algorithms now incorporate diffraction imaging and attenuation analysis to refine the boundaries of these hidden hazards.

Active vs. Passive Source Sensitivity

A significant area of research in the MASW discipline is the comparative analysis of active source data and passive microtremor recordings. Active MASW utilizes controlled sources like hammers, providing high-frequency data (typically 5 Hz to 50 Hz) that is ideal for characterizing the top 20 to 30 meters of the subsurface. This is the zone most critical for urban foundation stability and utility detection.

Passive MASW, often referred to as Microtremor Survey Method (MSM) or MAM (Microtremor Analysis Method), utilizes ambient noise—such as traffic, wind, and industrial machinery—as its source. This noise is predominantly composed of low-frequency surface waves (1 Hz to 15 Hz), allowing for much deeper investigations, sometimes exceeding 100 meters. In urban environments, passive MASW is particularly useful because it turns "cultural noise" into a data source rather than a hindrance.

Sensitivity Comparison

Research suggests that while passive methods offer greater depth, they often lack the vertical resolution required to pinpoint small, shallow anomalies. Active sources provide the high-frequency energy necessary to define the roof of a shallow void. However, in heavily congested urban areas where hammer strikes are impractical or the ambient noise floor is too high, passive microtremor analysis becomes the primary tool. The integration of both active and passive data into a single joint inversion has emerged as a strong solution, combining the high resolution of active sources with the depth penetration of passive sources.

"The integration of multi-modal dispersion curves and the combined use of active and passive wavefields are essential for reducing the non-uniqueness inherent in seismic inversion, particularly when imaging complex urban subsurfaces."

Urban Geophysical Survey Challenges

Urban geophysical surveys face several environmental and technical hurdles. Paved surfaces, such as asphalt and concrete, can act as high-velocity wave guides, potentially masking the signal from the underlying soil. This phenomenon, known as the "stiff layer" problem, requires specialized geophone couplings or the use of land streamers—geophones mounted on heavy sleds that can be towed behind a vehicle.

Furthermore, the presence of underground utilities (pipes, cables, and sewers) can introduce artificial anomalies into the seismic data. Decoupling the signature of a concrete sewer pipe from a natural karst void requires high-density data acquisition and the use of migration algorithms similar to those used in reflection seismology. The goal is to move from simple 2D profiling to true 3D volumetric imaging of the subsurface.

Data Interpretation and Uncertainty

A critical aspect of MASW in urban surveys is the communication of uncertainty. Geophysical inversion is a non-unique process; multiple subsurface models can theoretically produce the same observed dispersion curve. To address this, current research focuses on Bayesian inversion techniques, which provide a probability distribution of possible models rather than a single "best fit." This allows engineers to understand the confidence level of a detected void before initiating costly excavation or grouting programs.

The role of the geophysicist remains vital in interpreting the algorithmic output. Understanding the local geological context—such as the expected depth to bedrock or the typical size of karst features in the region—helps in constraining the inversion algorithms and filtering out false positives caused by surface noise or equipment artifacts. As algorithms continue to evolve, the integration of MASW with other geophysical methods, such as Ground Penetrating Radar (GPR) and Electrical Resistivity Tomography (ERT), provides a multi-physics approach to subsurface characterization.