Understanding GPR Data Processing: From Raw Scans to Subsurface Insights

Ground Penetrating Radar (GPR) is a powerful non-invasive technique used in geology, archaeology, engineering, and environmental studies to explore the shallow subsurface. However, raw GPR data is often noisy, misaligned, or difficult to interpret directly. That’s where data processing comes in as an important step to improve signal quality, decrease clutter, and extract meaningful geophysical information.

In this post, we will walk through key GPR data processing tools and concepts as typically observed in professional software such as EKKO_Project.

Raw GPR Data

Signal Attribute Analysis

These techniques transform the GPR signal into various attributes to highlight features:

• Envelope: Displays the amplitude "envelope" of the signal using Hilbert transform. Useful for highlighting object boundaries regardless of polarity. What is the Envelope in GPR?

  In Ground Penetrating Radar (GPR), each trace (i.e., each radar signal recorded over time) is an oscillating waveform composed of positive and negative amplitudes. These oscillations reflect how the electromagnetic wave is reflected or refracted by different subsurface materials.

  The Envelope is a mathematical curve that outlines the "outer boundary" of these oscillations. It shows how the signal’s overall amplitude changes over time regardless of whether it’s positive or negative.

  How is the Envelope Computed?

  The envelope is calculated using the Hilbert Transform, a technique from signal processing that generates the "analytic signal" of a real-valued trace. Here's what happens:

  1. Original Trace: This is your raw GPR waveform with both positive and negative peaks.

  2. Hilbert Transform: A version of the trace is computed that is phase-shifted by 90° (quadrature signal).

  3. Analytic Signal: Combine the original trace (real part) and the Hilbert transform (imaginary part).

  4. Envelope: Take the magnitude of this complex signal:

     Envelope(t)=Real(t)2+Imag(t)2Envelope(t)=Real(t)2+Imag(t)2​

     This gives a smooth, positive-only curve that tracks the amplitude strength over time.

  Why is the Envelope Useful?

  • Polarity-Independent: Since it doesn't care whether the signal is positive or negative, it focuses solely on how strong the reflection is.

  • Edge Detection: Subsurface interfaces (e.g., layer boundaries, voids, pipes) often produce strong reflections. The envelope makes these easier to see.

  • Improved Visualization: It simplifies radargrams by removing confusing wave oscillations and retaining just the energy “shape.”

  • Supports Object Delineation: Especially useful in archaeological or utility surveys where target outlines (like walls, tanks, or cables) matter more than signal polarity.
    

• Instantaneous Frequency: Shows how frequency changes over time. Helps identify materials with varying dielectric properties.

• Instantaneous Phase: Tracks phase shifts over time; useful for detecting thin layers or buried interfaces.

Signal Attribute Analysis

Migration Techniques

Migration repositions reflected energy to its correct spatial location, correcting for diffraction effects:

• FK Migration: Uses frequency–wavenumber (Fourier) domain techniques to collapse hyperbolas into points, especially effective for layered media.

• Kirchhoff Migration: A time-domain method using summation along diffraction hyperbolas. More flexible with irregular geometries.

FK & Kirchhoff Migration

Background Removal and Filtering

These methods reduce noise and enhance useful reflections:

• Background Average Subtraction: Removes the average trace from the data, ideal for eliminating horizontal banding.

• Background Subtraction: A general term that includes methods to remove consistent, non-reflective noise patterns.

• Horizontal Filter: Applies filtering across traces to reduce lateral noise.

• Spatial Median Filter: Uses a median value in a local neighbourhood to remove spikes or impulsive noise while preserving edges.

Background Removal & Filtering

Frequency and Amplitude Filters

Control signal strength and frequency content:

• Bandpass Filter: Allows only a specific frequency range (e.g., 50–300 MHz), improving signal clarity.

• Highpass / Lowpass Filters: Remove low or high frequencies respectively to suppress DC shifts or high-frequency noise.

• Convolution: Applies a kernel to the data, often used for custom filtering or sharpening.

• Dewow: Removes low-frequency “wow” caused by instrument drift, usually using a running average subtraction.

• DynaT: A dynamic time filter that adjusts filtering based on signal strength across time.

• Time Median / Vertical Filter: Reduces vertical noise (over time within a trace) using median filtering.

• DC Removal

Frequency & Amplitude Filters

Data Editing & Trace Manipulation

Adjust traces spatially or temporally:

• Edit First Break / Repick First Break: Adjust the first arrival time of a reflection, often needed for accurate velocity modeling or time-zero correction.

Gain

• AGC (Automatic Gain Control):AGC dynamically adjusts the gain along each trace to normalize amplitudes, making weak reflections more visible. It’s particularly useful when signal strength decays rapidly with depth, though it can sometimes distort true amplitude relationships.

• Constant Gain:This method applies a fixed amplification across all samples in the trace. It’s simple and preserves relative amplitude differences but may not adequately enhance deeper reflections if there's significant signal decay.

• SEC2 (Spreading and Exponential Compensation – Type 2):SEC2 applies a time-dependent gain to counteract geometric spreading and signal attenuation. It enhances deeper signals using a function typically proportional to time squared or exponential time functions. SEC2 is a refined version of gain correction, offering more nuanced control than AGC or constant gain.

Gain

• Antenna separation - The physical distance between the transmitting and receiving antennas in a GPR system. This parameter plays a crucial role in determining the depth of investigation, resolution, and signal quality. In bistatic systems (where transmitter and receiver are separate), adjusting the antenna separation allows you to:

  • Control investigation depth: Larger separation typically enables deeper signal penetration because the energy path travels farther into the ground.

  • Influence resolution: Smaller separations are better for resolving near-surface features, while larger separations may smooth out small anomalies but improve sensitivity to deeper layers.

  • Optimize signal strength: Improper spacing can lead to destructive interference, poor coupling with the ground, or insufficient signal return, so it's often adjusted during data acquisition or processing.

  In processing software, adjusting for antenna separation ensures proper time-zero calibration and accurate depth conversions, particularly when velocity models are involved.

• Crop Data – Horizontal/Vertical: Trims data on the trace (horizontal) or depth/time (vertical) axis.

• Pad Data – Horizontal/Vertical: Adds zero-padding to align or extend data dimensions.

• Reposition Traces / Reposition Using GPS: Corrects the trace position in space using manual or GPS data.

• Reverse Line: Flips the order of traces, useful when data was collected in reverse.

Analysis & Signal Enhancements

Tools to inspect and modify signal characteristics:

• Amplitude Spectra: Shows frequency distribution of the signal to diagnose filter needs.

• Average Trace: A mean trace used for comparison or background removal.

• CMP/WARR Analysis: Velocity analysis through Common Midpoint (CMP) or Wide-Angle Reflection and Refraction (WARR) data.

• Declip: Recovers clipped signals from saturated traces.

• Flip Polarity: Inverts trace polarity; necessary when reflection conventions are inconsistent.

• Insert Traces: Adds missing traces, usually for interpolation or regular spacing.

• Mute: Zeros part of the trace (e.g., early-time noise).

• Normal Moveout (NMO) Correction: Adjusts CMP data to account for offset-related travel time differences.

• Rectify: Converts all signal values to positive or absolute values, emphasizing amplitude regardless of direction.

Gain Functions

Control how amplitude changes with time:

• AGC (Automatic Gain Control): Normalizes amplitude over a moving window; enhances deep reflections.

• Constant Gain: Applies a uniform gain factor across the trace.

• SEC2: A time-dependent exponential gain (e.g., SEC2(Auto)) that compensates for signal attenuation with depth.

Acquisition Parameters

• Antenna Separation: Important for velocity calculations and resolution; affects how shallow or deep the radar can see.

Conclusion

GPR data processing is essential to make raw data interpretable. Each processing step—whether it’s a filter, gain adjustment, or migration algorithm—serves a specific role in revealing subsurface structures more clearly and accurately. By combining these techniques strategically, you can significantly enhance your radargrams, time slices, and 3D volume interpretations.

Whether you're analyzing lakebed sediments, detecting utilities, or studying glacier melt layers, understanding and applying these tools is key to producing reliable GPR interpretations.