Bundesanstalt für Materialforschung und -prüfung

International Symposium (NDT-CE 2003)

Non-Destructive Testing in Civil Engineering 2003
Start > Contributions >Lectures > Radar 2: Print


A. P. Annan, Sensors & Software Inc., 1040 Stacey Court, Mississauga, ON L4W 2X8, Canada


Ground penetrating radar (GPR) is regularly applied to examine building walls, floors, roads,railways and runways. Many applications have been pioneering proof-of-concept tests to define the benefits and limitations of GPR but these seldom use technology optimized for the specific use by non-destructive evaluation (NDE) practitioners.

Modern instrumentation and much enhanced computer power enable creation of GPR systems specifically designed for infrastructure inspection and evaluation. Data presentation is becoming more quantitative, integrated positioning information is enabling informative subsurface image visualization, and initial steps toward the quantitative extraction of physical property information are occurring.

This paper illustrates the design and development process using new products with example applications.

GPR, ground penetrating radar, NDE, NDT, concrete, pavement, reinforcing, conduits.


GPR has seen use as a non-destructive evaluation (NDE) method for a considerable period of time (Clemena, 1991 and Bungey et al, 1991). The GPR method, which gives high resolution sounding capability into lossy dielectrics, is ideally suited to NDE of man-made infrastructure.

Over the years, GPR has been used to study buildings and bridges (Annan et al, 2002, Heiler et al,1995 and Lorenzo et al, 2000,), roads and runways (Davis et al, 1994, Saarenketo & Scuillion, 2000, Jack & Jackson, 1999, and Benedetto & Benedetto, 2001, 2002), railways, pipelines and tunnel liners (Manacorda et al, 2002, Huston et al, 2000, Gobel, 1994, Gerrards et al, 2002) and miscellaneous other structures such as sewers and pipes. In all cases, the structures are made up of materials such as concrete, brick, stone, asphalt, soil, and wood which are relatively transparent to electromagnetic radiowave signals. Electrical properties of these materials are being evaluated more often with increased GPR useage (Robert, 1998, Sararenketo, 1998, Matthews et al, 1997).

The NDE objectives for using GPR are

  • quality assurance to confirm the structure has been built in accord with the design specification (i.e. Will improper construction lead to failure?);
  • monitoring to assure the structure's component materials have not degraded with time (i.e. Is remedial work needed to avoid imminent structural failure?); and
  • developing as-built drawings where none exist (i.e. Provide the basis for designing and implementing structure modifications).

In most cases, an engineering decision must be made based on the NDE evaluation. Furthermore, this decision must be based on an accepted quantifiable engineering criteria.

The major limitation of GPR is that it does not yield information in a form readily used in engineering decision-making. While some simple standards have been evolved for road and bridge deck applications (ASTM, 1987, 1997), the GPR field is not yet mature. Continuing research, experimental development and practical experience are needed to enable GPR observations to be routinely transformed into engineering decision-making data.

As designers, developers and vendors of GPR instrumentation we find that its widespread utilization is impeded by these obstacles. In the following we use our recent instrument developments to illustrate a philosophy which enables us to address these impediments.

GPR Principles

In developing GPR applications, it is important to understand exactly what GPR does and measures. In technical terms, a GPR system generates electromagnetic signals and detects the electromagnetic field interaction with the surrounding material (Davis & Annan, 1989).

Fig 1: GPR can be used in either reflection/scattering or transillumination mode. To date, the reflection/scattering mode is the most common.

GPR is generally used in two ways, namely, reflection/scattering mode or transillumination mode as depicted in Figure 1. In all cases, a measurement of an electromagnetic field component,normally an electric field, is carried out. The electric field component intensity is converted to a voltage signal which is recorded versus time (or in some cases frequency). This signal constitutes the raw GPR information.

The recorded signal is determined by the transfer function of the measurement electronics as well as the surrounding media. The measurement objective in the reflection/scattering mode is to detect changes in the surrounding environment. Note the key word here is "changes" or "differences". Translation of this measurement information into an understanding of the surroundings is the "art" of interpretation. Successful applications reach routine use when the "art" is transformed into a systematic repeatable operation.

For the two modes of deployment shown in Figure 1, examples of the resulting basic GPR information are depicted in Figure 2 and 3. For the reflection/scattering method, displaying the raw data often creates an image which resembles the structure that is present. The transillumination method requires more analysis and image reconstruction (similar to X-ray tomographic imaging) before it can be viewed.

Fig 2: Example of GPR reflection/scattering measurement over two road tunnels. GPR data show signal amplitude as a gray intensity versus horizontal position and signal travel time (or depth).
Fig 3: Example of a transillumination measurement between two boreholes. GPR data amplitude is shown as a gray intensity versus receiver position and signal travel time. Variations in travel time and signal amplitude indicate structure changes between the transmitter and receiver. Such measurements are also made through walls and pillars . Tomographic image reconstruction techniques are used to analyze the structure's interior.

The recorded GPR signal depends on changes within the surrounding media of the GPR wave properties: the radiowave phase velocity, v, attenuation, a and electromagnetic impedance, Z.

The spatial scale of measurement is dictated by the instrument bandwidth, B, (i.e. the range of frequencies contained in the measurement process). Without delving into detail, resolution can be divided into two parts for GPR, namely, radial and lateral resolution (Annan, 2002).

Fig 4: GPR data have an inherent resolution length dependent on material velocity and system bandwidth. Radial and lateral resolution lengths depicted here are different. Resolution is a key aspect of understanding GPR performance and applicability.

while distinct separation in the lateral direction requires


Distinct separation of targets in the radial (depth) direction requires

while distinct separation in the lateral direction requires


Bandwidth and velocity dictate radial resolution while lateral resolution depends on bandwidth,velocity, and distance to the target. These expressions are specifically for reflection mode operation; similar results apply to transillumination mode operation.

The recorded GPR signal which has an inherent bandwidth limit and hence spatial resolution limit, yields phase velocity, attenuation and electromagnetic impedance data that are resolved with this spatial resolution. The physical properties of the material, namely, the dielectric permittivity, e , the magnetic permeability, m , and the electrical conductivity, s , are related to (v,a , Z) and can be inferred with a similar spatial resolution.

GPR contains substantial information about physical properties and wave properties. These data however, seldom directly address the engineering needs of how much material, how thick, how strong, and how damaged.

Application Development

To effectively use a GPR for an NDE application, development work, testing and review of basic principles is required. The first critical question is "Is GPR useful?". GPR is often not useful for, nor the best solution to, many problems. Fortunately, there are many situations where GPR signals do contain quantitative information useful for developing data needed for engineering decision making. A systematic process must be followed in developing successful applications.Key elements of any process involve the following.

  • There must be a fundamentally provable relationship between the GPR observable attributes, velocity, attenuation and impedance, and the sought after engineering parameter. Further, the reliability and uniqueness of the relationship must be known and supported by independent research. An excellent example is the relationship between radiowave velocity and volumetric water content in soils (Topp et al, 1980); after many tests the empirical relationship has lead to a routine technique for measuring water content (Dane & Topp, 2003).
  • If the relationship is imperfect or non-unique, the analysis and procedures must enable a user to only apply the method when conditions are suitable.
  • Sufficient testing and a body of experience must be built up such that users can draw on a pool of experience to assess questionable or fallacious outcomes.
  • Standardized, calibrated and reliable instrumentation and data recording must be used at all times.
  • Systematic, repeatable and methodical measurement procedures must be followed.
  • Standardized data analysis for extraction of quantitative information accompanied by error estimates and reliability indicates must be used.

Historically, GPR was frequently used as a last resort, with the engineering decision-maker expecting a black-and-white answer while the GPR service provider was still determining whether the information desired could be obtained from GPR measurements. Expectations were vastly different between the two communities and were definitely incongruent with success.

Development of an NDE engineering application requires investment. Research teams combining engineering staff with a clear understanding of the problem must work with GPR designers and application specialists. Unfortunately, one or more components in this mix are commonly missing and economic factors often lead to research work being treated as routine inspection.Furthermore, many GPR users possess a limited understanding of GPR fundamentals, again leading to false expectations.

NDE Applications of GPR

We use two examples to illustrate how the preceding thought process has helped to develop solutions which meet user expectations. The first is the use of GPR for layer thickness determination for asphalt or concrete roads. The second is the use of GPR imaging to clear an area of concrete prior to cutting and/or coring thus avoiding embedded structural members and conduits.

Layer Thickness
Layer thickness is a topic which frequently arises in all GPR applications. NDE examples include asphalt thickness on roads, concrete thickness on roads, floors and walls, ballast depth beneath railways, and liner thickness in tunnels, and pipes and sewers. ASTM publication (1987) provides a methodology for determining asphalt/concrete layer thickness.

Following our logical application development strategy, can we safely assume GPR can always measure layer thickness? Lots of GPR records demonstrate that stratified horizons are detectable but this does not mean GPR will work reliably in all cases.

Figure 5 shows a basic logic flowchart for decision-making. A number of variations can be added but we keep the steps minimal to illustrate the process.

Fig 5: Flow chart showing some basic steps involved in applying GPR to a layer thickness application. Cut outs at critical decision points are important to achieve reliable outcomes.

Layer thickness determination requires detection of a response associated with the top and bottom of a layer. Further, some character of the signal must be proportional to layer thickness (for GPR this is usually travel time).

The key question is can GPR detect the top and bottom of a layer? Material knowledge clearly indicates GPR will not detect layer boundaries in all situations. To be detectable, there must be an electromagnetic impedance contrast between the layer and the media above and below.Typical impedances are shown for some common situations in Figure 6. Air/concrete and concrete/metal boundaries show large impedance contrasts while concrete/rock, asphalt/concrete and concrete/base coarse soils can be such that there is little or no impedance contrast.Systematic field tests may be needed to answer the detectability questions; this step must be included in a work plan unless prior site data is available.

Fig 6: GPR reflections at boundaries require a change in electromagnetic impedance between the materials. The ranges of impedance for construction materials overlap indicate some boundaries will not give rise to a detectable reflected signal.. Fig 7: GPR signals are absorbed in soils, rocks and concrete. In some instances, horizons and features become undetectable if attenuation is too high and the feature is too deep.

A detectable layer bottom also depends on the GPR signal attenuation within that layer being sufficiently low. There are many common examples where detection of layer bottom is difficult or impossible. Fresh concrete is opaque to GPR while cured concrete is moderately transparent.Detection of the bottom of a freshly poured concrete slab is virtually impossible. In North America, it is common in winter to spread salt on roads to thaw snow and ice. The salt dissolves creating saline water on, in, and under roads which strongly attenuates GPR signals. Figure 7 illustrates the concept of signal absorption limiting exploration depth.

Depth resolution is determined by GPR measurement bandwidth. While bandwidth can be increased indefinitely, most materials contain water. Increasing bandwidth means using higher GPR frequencies. The presence of small amounts of water cause strong signal absorption above 1000 MHz (Figure 8). For most practical situations, GPR bandwidth greater than 2000 to 3000 MHz is usually not possible.

Fig 8: Most construction materials contain residual moisture. The presence of water causes high frequency signal (above 1000 MHz) to be strongly attenuated. As a result, GPR bandwidth is limited to about 2000 MHz in many natural and construction materials.

There are two possible approaches to determining layer thickness which yield a quantity proportional to thickness. One is the transit time through the layer; as mentioned earlier, the other is the reflection amplitude if the layer is very thin (Annan et al, 1988, Brewster & Annan,1993). With transit time techniques, thickness resolution is determined by bandwidth (see equation 1). Transit time techniques yield absolute and relative thickness estimates while thin layer analysis normally yields a relative thickness estimate.

An absolute thickness estimate is derived from GPR observations only and requires measuring travel time and velocity of propagation without reference to physical control information (such as coring). The errors in both travel time and velocity dictate the reliability of absolute thickness estimates. Absolute layer thickness accuracies of better than ±5% are very difficult to achieve.

Relative layer thickness is achieved with much higher resolution by detecting subtle travel time (or amplitude) differences between observation points. Resolution is dictated by signal-to-noise and recording sensitivity. Relative layer thickness can be transformed into absolute layer thickness by using independent control data.

Conflicting claims about GPR thickness resolution are common. Absolute asphalt thickness accuracy on roads is typically 10 to 20 mm using a 1000 to 2000 MHz bandwidth GPR.Practitioners of road surveys regularly claim 2 to 4 mm accuracy but implicit in this claim is a correlation with coring information. Understanding the basis of the thickness determination removes confusion and establishes correct end-user expectations.

Figure 9 shows the Noggin Roadmap system developed specifically for asphalt, concrete and base course thickness evaluation on roads and runways. Methodology for operation and data analysis follows that described in the preceding.

Fig 9: The Noggin RoadMap profiling system acquiring asphalt thickness data on a city street. Sensor proximity to the road surfaces enhances signal quality. Controlled operation with integrated GPS positioning and odometer driven sampling are key components of the system.
Fig 10: Example of asphalt thickness determinations using automated picking software. Comparative core thickness determinations are shown. Coring was conducted several years earlier. The GPR profile attempted to intersect coring locations but some lateral misalignment occurred. Coherency gives a measure of the reliability of the thickness estimate based on reflection strength and consistency.

An example of extracted information for an urban asphalt covered street is shown in Figure 10.The weak and variable nature of the reflection from the asphalt/base coarse material interface is evident. The estimated thickness and quality of the estimate are shown for automated analysis.Many other details are beyond the scope of this discussion. Suffice it to say, using a logical process results in realistic expectations. Integrating the logical process into the GPR system leads to more reliable and repeatable results.

Concrete Imaging

GPR has emerged recently as an alternate technique to X-raying concrete structures prior to cutting and coring. Historically, X-ray technology created a photographic image of the concrete interior but it required access to both sides of the structure and involved the use of hazardous X-ray sources. GPR is rapidly gaining favor as it can be effectively used by non-GPR specialists, it is quick and easy to use on site, and is non-hazardous.

There are a number of key assumptions which underpin this application.

  • Embedded elements in concrete represent an impedance contrast to the concrete.
  • Absolute measurements of object geometry are less critical than knowing location.
  • The prime objective is the lateral location of positions free of embedded elements to allow drilling or cutting through a slab.
  • Depth to elements has not been available previously so even moderately accurate depth estimates are useful.

These assumptions are compatible with GPR measurements. Extreme accuracies and sensitivities are not required to achieve the main objective. Once the main objective is achieved, other aspects such as the embedded element's depth and geometry can he estimated.?????(Don't know intended meaning.) Understanding the resolution limits is key to establishing realistic user expectations.

After exploring the application and doing numerous tests, the flow chart for concrete imaging shown in Figure 11 was developed to guide system design.

Fig 11: Flow chart showing the basic logic steps for concrete imaging used in Conquest concrete imaging system development.

In developing the application, detectability of embedded features was not in question from an impedance contrast perspective. Air, voids, gas conduits, wires, cables, metal pipes, and reinforcing and tensioning cables all have high impedance contrasts with cured concrete. Airfilled voids and non-metallic gas conduits with small dimensions generate weak return signals posing a potential sensitivity problem. Another limit is created by signal absorption. Concrete attenuates GPR signals limiting exploration depth to 400 to 600 mm.

To image concrete properly, estimation of radiowave velocity in concrete at each location must be part of the data analysis process. In addition, careful repeatable sensor positioning is critical.

All of these factors formed part of the system design. The Conquest system shown in Figure 12 illustrates how the combination of procedure, data acquisition, and signal processing make a system which can be used routinely with reliable outcomes.

Fig 12: Conquest concrete imaging system showing basic features of operation. The integrated procedure steps the operator through the data acquisition and velocity calibration sequence. Plan map images at a series of depths in the concrete are generated on the spot.
Fig 13: Example data image showing reinforcing mesh and an embedded conduit in a concrete slab on pan.

An example of a conduit embedded in concrete beneath a two-directional reinforcing steel is shown in Figure 13. The measurement bandwidth is 1000 MHz. The absolute depth and lateral resolution lengths are about 20 to 30 mm. Relative depths and lateral positioning are typically 3 to 6 mm.

The success of this GPR system has been truly amazing. While GPR has been used to see bars and conduits in concrete for more than 20 years, the advent of an instrument with integrated systematic imaging has transformed potential into practice. The GPR technique, even with some inherent limitations, is rapidly replacing X-ray techniques for the majority of cutting and coring pre-planning requirements.

Summary & Conclusions

The use of GPR in infrastructure NDE is increasing. While application potential has been extensively defined by past experience over the last 20 years, systems must be tailored to specific application requirements to truly make the technique effective.

The system development examples presented here provide the insight into the future of GPR for NDE. Customized systems exploiting computer technology to aid the user along a systematic decision path will become widespread. Problems not amenable to such treatment will remain fringe activities and other techniques must be used. For the subset of tractable applications,unique GPR systems will provide standardized techniques with predictable quantifiable outcomes.


  1. Annan, A.P., Cosway, S.W., De Souza, T., 2002, Applications of GPR to Map Concrete to Delineate Embedded Structural Elements & Defects, Ninth International Conference on Ground Penetrating Radar, Koppenjan, S.K., Lee, H. (eds.), Proceeding of SPIE, Vo. 4758, pp.359-364.
  2. Annan, A/P., Davis, and Gendzwill, D., 1988, Radar Sounding in Potash Mines:Saskatchewan, Canada, Geophysics, Vol. 53, pp. 1556-1564.
  3. Benedetto, A., Benedetto, F., 2002, GPR Experimental Evaluation of Subgrade Soil Characteristics for Rehabilitation of Roads, Ninth International Conference on Ground Penetrating Radar, Koppenjan, S.K., Lee, H. (eds.), Proceeding of SPIE, Vo. 4758, pp. 708-714.
  4. Benedetto, A., Benedetto, F., 2001, About the possibility to automate the GPR signal processing for road pavement monitoring, Proc. SIIV Conference, Verona, Italy.
  5. Benedetto, A., De Blasiis, M.R., 2001, Road Pavement Diagnosis, Quarry and Construction,Vol. 6, pp. 93-111.
  6. Brewster, M.L. and Annan, A.P., 1994, Ground-penetrating radar monitoring of a controlled DNAPL release: 200 MHz radar, Geophysics, Vol. 59, pp. 1211-1221.
  7. Bungey, J.H., Millard, S.G. and Shaw, M.R., 1991. The use of sub-surface radar for structural assessment of insitu concrete, A.C.I., SP128/31, Vol. 2, pp. 497-514.
  8. Bungey, J.H., Shaw, M.R., Millard, S.G., Soutos, M.N., and Austin, B.A., 1997. The influence of concrete composition upon radar test results, Insight, Vol. 39, pp. 474-478.
  9. Clemena, G.G., 1991. Short pulse radar methods, Chapter 11 in Handbook on Non-Destructive Testing of Concrete, Malhotra, V.M. and Carino, N.J. (eds.), CRC Press, Boston, pp. 253-274.
  10. Dane, J. H. and Topp, G. C. (co-eds.),2002, Methods of Soil Analysis, Part 4 Physical Methods, Soil Science Society of America, Inc.
  11. Davis, J.L. and Annan, A.P., 1989, Ground Penetrating Radar for High-Resolution Mapping of Soil and Rock Stratigraphy, Geophysical Prospecting, Vol. 37, pp. 531-551.
  12. Davis, J.L., Rossiter, J.R., Mesher, D.E., Dawley, C.B., 1994, Quantitative measurement of pavement structures using radar, Proceedings of fifth International Conferences on GPR,Kitchener, Ontario, Canada.
  13. Geraads, S., Charachon, B., Loeffler, O., Omnes, G., 2002, Applying a wavenumber notch filter to remove interferences caused by railway sleepers from a GPR section, Ninth International Conference on Ground Penetrating Radar, Koppenjan, S.K., Lee, H. (eds.),Proceedings of SPIE, Vol. 4758, pp. 715-718.
  14. Gobel, C., Hellmann, R., Petzhold, H., 1994, Georadar model and in-situe investigations for inspection of railways tracks, Proceedings of Ground Penetrating Radar Conference, Kitchener, Canada, June 12-16, 1994.
  15. Hugenschmidt, J., 1998, Railway track inspection using GPR - Some examples from Switzerland, Proceedings of Ground Penetrating Radar Conference, Lawrence, Kansas (USA),May 27-30, 1998.
  16. Heiler, M., McNiel, S., Garret, J., 1995, Ground Penetration Radar for Highway and Bridge Deck Condition Assessment and Inventory, SPIE, Vol. 2456/195
  17. Huston, D., Pelczarski, N., Esser, B., 2000, Damage detection in roadways with Ground Penetrating Radar, Proc. Eighth International Conference on GPR, Gold Coasts, Australia.
  18. Jack, R., Jackson, P., 1999, Imaging attributes of railway track formation and ballast using ground probing radar, NDT&E International Vol. 32, pp 457-462.
  19. Lorenzo, H., Hernandez, M.C., and Cuellar, V., 2000, In Situ Radar Remote Sensing Applied to the Detection of Degradation in a concrete floor, Eight International Conference on Ground Penetrating Radar, Noon, D.A., Stickley, G.F., and Longstaff, D. (eds.), SPIE, Vol. 4084 (2000).
  20. Matthews, S.L., Goodier, A., Massey, S., 1997. Permittivity Measurements and Analytical Dielectric Modelling of Plain Structural Concrete, Proceedings of the 7th International Conference on Ground Penetrating Radar, Kansas, USA.
  21. Manacorda, G., Morandi, D., Sarri, A., 2002, A customized GPR system for railroad tracks verification, Ninth International Conference on Ground Penetrating Radar, Koppenjan, S.K.,Lee, H. (eds.), Proceedings of SPIE, Vol. 4758, pp. 719-722.
  22. Robert, A., 1998, Dielectric permittivity of concrete between 50 MHz and 1 GHz and GPR measurements for building materials evaluation, Journal of Applied Geophysics, Vol. 40, pp. 89-94.
  23. Saarenketo, T., 1998, Electrical properties of water in clay and silty soils, Journal of Applied Geophysics, Vol. 40, pp. 73-88.
  24. Saarenketo, T., Scuillion, T., 2000, Road evaluation with ground penetrating radar, Journal of Applied Geophysics, Vol. 43, pp. 119-138.
  25. Shaw, W.R., Molyneaux, T.C.K., Millard, S.G., and Bungey, J.H., 1997. Automated Analysis of GPR Scans on Concrete Structures, Proceedings of 7th International Conference on Ground Penetrating Radar, Kansas, USA.
  26. Topp, G.C., Davis, J.L., and Annan, A.P., 1980, Electromagnetic Determination of Soil Water Content: Measurements in Coaxial Transmission Lines, Water Resources Research, Vol. 16, No. 3, pp. 574-582.
  27. ASTM (1987): Standard Test Method for Determining the Thickness of Bound Pavement Layers Using Short-Pulse Radar, ASTM Designation D 4748-87.
  28. ASTM (1997): Standard Test Method for Evaluating Asphalt-Covered Concrete Bridge Decks Using Ground Penetrating Radar, ASTM Designation D 6087-97 (Re-approved 2001).
STARTPublisher: DGfZPPrograming: NDT.net