3D NM-GPR

Ground Penetrating Radar (GPR)

Ground Penetrating Radar (GPR) is a well-established technique that involves emitting ultrawideband electromagnetic (EM) signals into the ground. The time and strength of returning reflections are measured by the GPR and used to detect and image subsurface features.

Noise-modulation

Noise-Modulated GPR systems emit EM waves into the ground in the form of a coded sequence. Like traditional GPR systems that send brief pulses, the emitted signal is an ultrabroadband time domain signal. Unlike those systems, however, the use of a code adds phase adjustments that spreads the signal over time. Return signals measured by the GPR equipment are then  cross-correlated with the emitted code (or other deconvolution methods are used) to remove these phase adjustments and produce the GPR A-scan. A more detailed overview of the general noise-modulation approach can be found in Ground Penetrating Radar, 2nd Edition (Daniels, Ed., 2004) (link)

GPR equipment using variants of the noise-modulation approach have been used for more than 15 years in Australia and 5 years in South-East Asia.

NM-GPR

The NM-GPR approach used by CodedRADAR is not merely an implementation of noise-modulation techniques using conventional hardware. Rather, it is a complete rethink of the type of signals emitted by GPR systems and the hardware used with the aim of maximising collection efficiency, easing regulatory compliance and achieving superior 3D operational performance.

Limitations of conventional methods

Different types of GPR equipment emit different forms of EM signal. The most commonly-used system types are Impulse GPR, which emit brief pulsed signals, and Stepped-Frequency GPR, which emit narrowband sinusoids. Both types typically use a high-fidelity receiver (such as a 16-bit Analog to Digital Converter, ADC) to measure returning reflections.

These receivers, however, are unable to measure the entire return signal in one operation. So instead, an incremental sampling approach is used.

Impulse GPR systems do this by measuring the magnitude of the return signal for one or more brief instants after each emitted pulse, while adjusting the delay between pulse generation and sampling. Measurements from many pulses at different delays are then combined to create one equivalent A-scan. While simple, this approach is also highly inefficient because the ADC only “listens” to a very small fraction of each return signal.

Step frequency GPR systems, on the other hand, emit a series of narrowband signals at discrete steps over the frequency range of interest. The amplitude and phase of the response is measured at each frequency step, before moving to the next. An Inverse Fourier Transform (IFT) then combines and converts the measurements at all frequency steps into the time-domain, producing the GPR A-scan. While far more efficient in terms of receiver design (i.e. the ADC is continuously “listening” at each step), the hardware is also more complex and difficult to scale. Furthermore, the narrowband signals used are problematic in terms of regulatory compliance.

Performance needs: 2D vs 3D GPR

Most commercial GPRs are 2D Impulse systems. They typically use a pair of antennas (1 transmitting antenna (Tx) + 1 receiving antenna (Rx)) at a fixed spacing (common offset) operating over a given range of frequencies. The data produced is a quasi-vertical slice into the ground along the travelled path. In recent times, various GPR manufacturers have added antennas of different sizes to create “dual” or “triple” frequency GPR’s. These produce one or two additional 2D datasets along the same path with different resolutions and penetration depths, but the basic operation is the same.

3D GPRs, by comparison, measure on many Tx and Rx combinations across an antenna array to produce multiple parallel 2D profiles per pass. The measurements can then be combined to form a 3D data volume.  Some 3D systems collect dual-polarised measurements, where even more data is gathered using a second array of antennas oriented normal to the first. While similar 3D datasets can be gathered using a 2D system and scanning along a grid, the ease of data collection makes using a dedicated 3D GPR system a much faster and more practical solution, particularly for scanning large areas. The much greater number of antenna combinations in 3D systems places far greater demand on hardware collection capabilities. For instance, while 2D GPRs may only collect between 1 and 4 radargrams (B-scans) per pass, 3D systems may collect 30 or more.

Without substantially greater collection capacity, the GPR hardware in a 3D system will not keep up with demand. When this occurs, it becomes necessary to compromise – either by reducing collection speed, number of active GPR channels across the array, measurement spacing in the direction of travel, penetration depth, data quality or a combination of these.

3D NM-GPR: More and better

The NM-GPR approach achieves many benefits of conventional GPR technologies, while avoiding important limitations. Superior receiver efficiency, use of a greater number of dedicated receiving samplers and reduced system complexity makes NM-GPR particularly well suited to 3D use.

A key difference is in the receiver architecture. Instead of using high-fidelity receivers, NM-GPR systems instead use multiple (4 or 8) ultra-fast low-fidelity samplers within the GPR digitiser unit. Each is listening 100% of the time for maximum efficiency, with stacking used to achieve high signal fidelity.

NM-GPR digitiser
(4 Tx + 8 Rx ports)

Due to their relative simplicity, adding more of this type of receiver does not adversely affect overall system complexity or reliability. Furthermore, because coded sequences are inherently ultra-broadband and spread over time, peak power transmissions are lower and there are no narrowband frequency concentrations, for easier regulatory compliance.

These differences provide NM-GPR with many significant advantages over competing technologies, which has enabled production of much faster and more reliable 3D GPR equipment that avoid the usual performance compromises.