GPR Signal Physics and Soil Geology Effects

A GPR system consists of a transmitter antenna that fires short pulses of electromagnetic energy into the ground, a receiver antenna that listens for echoes, and a control unit that records the return signal as a time series called a trace. The transmitter and receiver are held at a fixed separation and moved together across the ground surface, collecting one trace every few centimetres along the traverse direction. Stacking all the traces side by side produces a two-dimensional section: depth (converted from two-way travel time) on the vertical axis and horizontal distance along the traverse on the horizontal axis.

Reflections occur wherever the dielectric permittivity changes. At a boundary between two materials with different permittivities, some of the pulse energy is reflected back toward the surface (detected by the receiver) and some continues downward (where it may reflect from deeper boundaries). The reflection coefficient at a sharp boundary is determined by the permittivity contrast: a large contrast produces a strong reflection. A grave boundary, where disturbed fill meets undisturbed host soil, is a moderate contrast. A body cavity filled with air against saturated soil is a strong contrast. A plastic coffin lid against damp clay is a weaker contrast that may not be detectable at depth.

A clandestine burial disturbs the soil in ways that alter its dielectric properties. The act of digging inverts the soil profile, bringing deeper subsoil to the surface and mixing topsoil downward. This changes the moisture distribution and the organic content of the fill, creating a zone of different permittivity from the undisturbed material surrounding it. On top of that, a decomposing body releases fluids into the grave fill, increasing its electrical conductivity and altering its permittivity, typically over a period of months to years.

The vertical boundaries of the grave cut are often the most reliable GPR reflectors, particularly in the early stages of burial. They represent a relatively sharp boundary between disturbed fill and undisturbed host material. In a clean GPR profile over a burial, these sidewalls appear as subvertical reflections that converge or bound a zone of internal reflections. The base of the grave, where the permittivity steps from fill to the deeper undisturbed soil, is often detectable as a near-horizontal reflector.

Attenuation is the single most common reason GPR fails to detect a genuine target. The attenuation coefficient of a soil depends primarily on its electrical conductivity, which rises with moisture content, clay content, and dissolved ion concentration. In pure sand with minimal clay at field capacity, attenuation at 250 MHz is roughly 1-2 dB/m, and a target at 2 m depth is detectable. In a saturated illite-rich clay, attenuation can exceed 10-20 dB/m, and the signal is largely gone before it reaches 0.5 m.

Pringle and colleagues (2008) published the most cited field validation of GPR performance across soil types in a forensic context. Using controlled burials at multiple UK sites, Pringle found that sandy and chalk sites consistently allowed detection at depths greater than 1 m, while heavy clay sites produced results that depended heavily on season and burial age. That study is the empirical foundation for soil-type guidance in UK forensic practice; Schultz (2008) extended the evidence base with controlled burial studies in Florida, documenting time-dependent changes in GPR anomaly detectability across burial stages.

GPR antennas are manufactured in discrete frequency bands, each with a characteristic resolution-depth tradeoff. The centre frequency determines the wavelength in the soil: higher frequency means shorter wavelength, better resolution, but faster attenuation. Lower frequency means longer wavelength, coarser resolution, but greater penetration.

• 100 MHz: deep penetration (5-10 m in dry ground), but resolves only features larger than about 0.3 m. Used for deep geological profiling and large-void detection; overkill for most single-body forensic searches.
• 250 MHz: the forensic workhorse. Reaches 2-3 m in sandy soil, resolves to roughly 6 cm in dry conditions. Balances depth and resolution for single-body burials at typical clandestine depths (0.3-1.5 m).
• 500 MHz: good resolution (3-4 cm) but limited to about 1 m depth in dry ground, less in moist soil. Useful for shallow burials or when fine structural detail within a grave is sought.
• 900 MHz-1 GHz: very high resolution, very shallow penetration. Used for detecting fine features in concrete, road surfaces, or archaeological contexts at centimetre depths.

GPR records time, not depth. Converting the two-way travel time of a reflection to a depth requires knowing the velocity of the radar wave in the soil. Velocity depends on permittivity: v = c / sqrt(epsilon_r), where c is the speed of light in vacuum and epsilon_r is the relative permittivity. Dry sand (epsilon_r around 4) gives a velocity of about 0.15 m/ns. Saturated sandy loam (epsilon_r around 20) gives about 0.07 m/ns. Using the wrong velocity produces a depth map that is systematically too shallow or too deep.

1. Hyperbola fitting
   When the survey passes over a point-like object (a stone, a pipe, or a body part), the time section shows a hyperbola. The shape of the hyperbola's wings is controlled by velocity. Software fits a modeled hyperbola to the data, adjusting velocity until the fit minimises residuals. This is the quickest in-field calibration method.
2. Common mid-point (CMP) acquisition
   The transmitter and receiver are separated progressively while remaining centred on a fixed midpoint. The change in arrival time with separation is used to compute a velocity profile with depth. More rigorous than hyperbola fitting but requires additional acquisition time.
3. Known-depth reflector
   Where a target of known depth exists (a buried pipe at 0.8 m confirmed by utility records), the travel time to its reflection calibrates the velocity directly. Useful when both a known and unknown target are present in the same survey.

For court presentation, the velocity used in depth conversion must be documented in the survey report, along with the calibration method. Any uncertainty in velocity propagates directly into depth uncertainty. A velocity known to within 10% gives a depth estimate with 10% uncertainty, which translates to 0.1 m uncertainty at 1 m depth. This is adequate for directing an excavator but should be stated explicitly rather than implied as exact.

Raw GPR data is almost never presented directly to a court or to an investigation team. Standard processing steps improve interpretability and remove artefacts. Each step must be documented so that the processing chain is reproducible and the original raw data remains available for independent review.

• Time-zero correction: shifts all traces so that the ground surface reflection is at a consistent time. Without this, surface topography introduces false depth variations.
• Dewow filtering: removes the low-frequency drift produced by direct coupling between transmitter and receiver, which can mask shallow reflections.
• Gain application: amplifies signal that has weakened with depth, making deep reflections comparable in amplitude to shallow ones. Gain must be applied consistently to avoid artefacts.
• Background subtraction: removes horizontal banding produced by consistent reflections (antenna ringing, surface coupling), which can obscure horizontal subsurface features. Use with care: it also removes genuine horizontal reflectors.
• Migration: collapses hyperbolic reflections to point positions, sharpening the image. Requires an accurate velocity estimate to work correctly. Incorrect migration velocity creates migration smiles (artefacts).

Annotated sections presented to a court should identify anomaly locations, their interpreted depth and dimensions, and the confidence level assigned to each interpretation. The Cheetham (2005) framework for presenting geophysical evidence, developed in the context of UK forensic archaeology, recommends a three-tier classification: confirmed (excavated and verified), inferred (two or more methods agree), and possible (single method, moderate anomaly). Using this or an equivalent scheme protects both the practitioner and the investigation from overstatement.