Improved background and clutter reduction for pipe detection under pavement using Ground Penetrating Radar (GPR)
Introduction
Subsurface drainage features are routinely incorporated in the design of pavement systems as they are believed to increase pavement service life provided that they are installed correctly and maintained. When poorly maintained, the existence of the drains can lead to roadway freeze-thaw damage or differential settlement, as obstructed or otherwise damaged conduits can serve as concentrated sources of water, ill-controlled paths for erosion, or undesirable air voids prone to local displacement. Maintenance of these drains is therefore essential.
However, effective maintenance is often hampered by the challenge of locating the drains, which may have been installed years prior to the current pavement surface, and thus may be undocumented on as-built drawings, and may have outlets that are visually obscured due to silt build-up and overgrowth of vegetative cover. To this day, the exercise of locating drainage conduits is routinely performed through manual search operations that entail physically walking up and down the slopes of roadway embankments in an effort to find signs of drainage outlets. Given that there are literally tens of thousands of drains associated with a nation’s highways, a more rapid method to explore the subsurface beneath pavements has obvious value.
There are several different methods that can be used to detect subsurface features like these drains below pavement, such as Metal Detectors, Electronic Marker Systems (EMS), Electromagnetic Terrain Conductivity, Acoustic Emission, Resistivity, Ground Penetrating Radar (GPR), Microgravitational Techniques, and Seismic Reflection/Refraction methods (Jeong et al., 2003). Among these, several researchers have demonstrated the potential to locate plastic and clay conduit in agricultural soils (Allred et al., 2004; Szuch et al., 2006), and assess water presence in/around buried conduits (Hunaidi et al., 2010) using ground penetrating radar (GPR). Other researchers have successfully utilized GPR in the analysis of materials such as wood, concrete, and asphalt (Rmeili and Scullion, 1997; Halabe, 1997; Maierhofer, 2003; Devaru et al., 2008). Early detection and evaluation of pipe leaking was also monitored by GPR via microwave tomographic inversion (Crocco et al., 2009; Cataldo et al., 2014). Thus, due to the size, geometry, and varying materials employed in most pavement and pavement drainage systems, and the desire to scan extensive lengths of highway with relative speed and ease, GPR tends to offer the greatest potential to facilitate drainage feature detection in subsurface settings and is likely the most commonly employed method used to complement manual drainage conduit detection in practice.
However, successful use of the method is highly dependent upon subsurface conditions, the presence or lack of water in the studied system, the nature of the subsurface target being sought, and its depth. Further, effective use of GPR still requires expert data interpretation and traditionally relies upon human visual observations or code driven pattern recognition algorithms that seek the 2-D hyperbolic returns indicative of a buried conduit (Olhoeft, 2000). Unfortunately, these procedures can still miss many of the older drains beneath pavements (as illustrated below) as they may be partially or fully filled with sediment and/or may be fabricated from clay or other earthen materials, yielding a return signal that is convolved with significant background noise. Given the safety, economic, and public perception consequences of missing a deteriorated or obstructed drain that could ultimately cause significant road damage, few departments of transportation are willing to forego manual exploration for drains even when they do employ current GPR techniques.
With this in mind, this paper builds on a comprehensive study that highlighted potential avenues to improve GPR detection of sub-pavement drains (Sinfield and Bai, 2013) and puts forward a novel background noise and clutter reduction method for GPR to enhance target signals in what amounts to a constructed environment that tends to have more consistent subsurface properties than one might encounter in a general setting. Within this technique, two major algorithms are employed. Algorithm 1 is the core of this method, and plays the role of reducing background noise and clutter. Algorithm 2 is supplementary, and helps eliminate anomalous discontinuous returns generated by the equipment itself, which could otherwise lead to false detection indications in the output of Algorithm 1. Instead of traditional 2-D GPR images, the result of the proposed algorithms is a 1-D plot along the survey line, highlighting a set of “points of interest” that could indicate buried drain locations. Details of the method and its application to the analysis of GPR surveys conducted on stretches of highway in Indiana in the United States are presented below.
Section snippets
Overview of background noise and clutter reduction methodology
One advantage of GPR is the potential to detect non-metallic targets. However, the strength of the reflection signals from non-metallic targets is significantly weaker than that obtained from metallic targets of equivalent dimension and position. Further, useful information about the target may be obscured by the background signal, typically termed clutter and noise in radar theory. Background clutter and noise normally includes three components: the breakthrough signals directly from
Pre-processing and selection of vertical data analysis interval
GPR data is normally obtained in the form of a B-Scan radar image (Fig. 2), and can be interpreted as a matrix. Each column of the matrix is a digitized single trace of a scan, which is a so called A-Scan in Radar theory. Fig. 2, illustrates a typical GPR reflection signal from a scan of a sub-pavement drain performed along a survey line orthogonal to a buried pipe. The hyperbola shape located approximately halfway across the upper 1/3rd of the image indicates the position of the buried
Overview of Algorithm 1
The method presented here is a moving average background subtraction approach as mentioned above. In this approach, at any given point of signal analysis along a survey line, a point termed the “check pointâ€, an averaged A-scan background signature, is subtracted from the A-scan at that point to reveal the signal of interest. The “average” background signature is developed by examining data (a series of A-scans) in a region of finite length, w, along the survey line that is located a fixed
Overview of Algorithm 2
GPR images are often plagued by anomalous discontinuous scan traces, as shown in Fig. 4. Such anomalous traces could provide false high peaks in the SNRdB plots mentioned above, which might generate an inaccurate detection result for the entire survey line. While causes of these anomalies vary, Liu et al. (2018) assessed the stability of GPR systems, and highlighted that their performance can degrade with use and aging, and is particularly vulnerable to antenna vibration among other variables,
Overview
Based on the methodology introduced previously, two field surveys are analyzed in the following sections.
The first field test was performed on a section of highway US-231 in central Indiana, in the United States, near the intersection of US-231 and INDIANA-25 involving 4 PVC outlet pipes. The average depth of the target pipes was about 0.6 m–0.9 m (2–3 feet), and the pipes had a diameter of 8 inches (~0.2 m) (Pipe No. 1) or 4 inches (~0.1 m) (Pipe No. 2-4). The GPR system used in this test was
Conclusions
This paper introduced a novel background and clutter reduction method to enhance buried pipe detection signals in constructed environments. Generally, there are two major algorithms included. Algorithm 1 is focused on reducing the effects of background noise and clutter. Algorithm 2 is focused on eliminating anomalous signals received by the GPR equipment itself. The results obtained through the above 2 algorithms are combined together in order to get an optimal output for any given GPR
Declaration of Competing Interest
None
Acknowledgements
This work was supported in part by the Joint Transportation Research Program administered by the Indiana Department of Transportation (INDOT) and Purdue University. The authors would also like to thank Dr. Dwayne Harris and other INDOT staff for assisting with data collection. The contents of this paper reflect the views of the authors, who are responsible for the facts and the accuracy of the data presented herein, and do not necessarily reflect the official views or policies of the sponsoring
References (33)
- et al.
Location of buried plastic pipes using multi-agent support based on gpr images
J. Appl. Geophys.
(2011) - et al.
Effects of gpr antenna configuration on subpavement drain detection based on the frequency-shift phenomenon
J. Appl. Geophys.
(2017) - et al.
Time domain reflectometry, ground penetrating radar and electrical resistivity tomography: a comparative analysis of alternative approaches for leak detection in underground pipes
NDT & E Int.
(2014) - et al.
Gpr-signal improvement using a synthetic emitter array
J. Appl. Geophys.
(2011) - et al.
Early-stage leaking pipes gpr monitoring via microwave tomographic inversion
J. Appl. Geophys.
(2009) - et al.
Seasonal variations measured by tdr and gpr on an anthropogenic sandy soil and the implications for utility detection
J. Appl. Geophys.
(2017) - et al.
A novel method to remove gpr background noise based on the similarity of non-neighboring regions
J. Appl. Geophys.
(2017) Maximizing the information return from ground penetrating radar
J. Appl. Geophys.
(2000)- et al.
Background matrix subtraction (bms): a novel background removal algorithm for gpr data
J. Appl. Geophys.
(2014) - et al.
The analysis of ground penetrating radar signal based on generalized s transform with parameters optimization
J. Appl. Geophys.
(2017)
Clutter removal for landmine using different signal processing techniques
Clutter reduction and detection of landmine objects in ground penetrating radar data using singular value decomposition (SVD)
Detection of buried agricultural drainage pipe with geophysical methods
Appl. Eng. Agric.
Clutter reduction and object detection in surface penetrating radar
Detection of shallowly buried objects using impulse radar
IEEE Trans. Geosci. Remote Sens.
An underground obstacle detection and mapping system
Cited by (21)
GPR detection localization of underground structures based on deep learning and reverse time migration
2024, NDT and E International3D reconstruction in underground utilities
2023, Automation in ConstructionEnd-to-end deep learning model for underground utilities localization using GPR
2023, Automation in ConstructionUses of Green's function for enhancing the image resolution of Ground Penetrating Radar (GPR) data
2022, Journal of Applied GeophysicsCitation Excerpt :Horizontal noise removal: The background noise is mainly the type of horizontal noise. The horizontal noise has been observed in all traces, and it can be easily identified in the radargram (Bai and Sinfield, 2020). This process has increased signal to noise ratio of the radargram.
Improving the accuracy of pavement structural quality assessment by correcting numerical hypothetical model of modulus Back-Calculation through GPR
2022, Construction and Building MaterialsCitation Excerpt :Shan Zhao realized the accurate evaluation of the position and size of the internal drainage pipeline of the road through image reconstruction[9]. Hao Bai introduced a background and clutter reduction method to enhance the accuracy of buried pipe detection[10]. Hai Liu calibrated the antenna array by considering the antenna phase center and the antenna offset.