Ground Penetrating Radar Data Research Articles

Modeling and advanced Ground Penetrating Radar techniques for evaluating pipe leakage and pavement distress

This study provides a comprehensive analysis of Ground Penetrating Radar (GPR) data processing techniques for detecting water leakage in urban pavements. By examining GPR images characterized by continuous and discontinuous high-amplitude reflections, the research identifies clear indicators of water saturation caused by leaking underground pipes. Computer modeling of GPR profiles enables effective comparisons with real data and helps in calibrating key GPR attributes such as energy, amplitude variance, and RMS amplitude. These attributes allow for the precise delineation of leakage zones in both synthetic and real GPR datasets collected in the Amazon region of Brazil. The Reflector Tilt Method (RTM) significantly enhances GPR data analysis, especially for detecting subsurface water leaks. RTM utilizes changes in the relative dielectric constant and electromagnetic wave velocity induced by water saturation in unconsolidated, horizontally stratified sand layers. By analyzing the tilt of reflections in GPR profiles, RTM differentiates between dry and saturated zones, providing insights into the geometry and extent of water infiltration. Notably, this study demonstrates how increased subsurface water saturation causes underlying reflectors to tilt. Using the tilt angle, we propose a novel technique to quantify the increase in the dielectric constant due to water saturation. This technique could also be adapted to detect leaks involving other fluids, as long as there is a strong contrast between the dielectric constant and the background. The integration of RTM improves the creation of accurate three-dimensional models of affected areas, enhancing leak detection and aiding in understanding water movement dynamics within the subsurface. The study underscores the importance of combining GPR attributes with RTM to generate detailed 3D models of leakage zones, which are essential for estimating the volume of infiltrated areas. Observations indicate significant water saturation near leaking pipes at shallow depths, with saturation levels decreasing rapidly as depth increases. By identifying 2D geophysical signatures of water leaks in public distribution networks, the research highlights the effectiveness of non-invasive GPR methods in mitigating risks such as soil erosion and surface collapse. This approach demonstrates the utility of GPR attributes and RTM as essential tools for assessing urban pavement integrity, enabling early detection of leaks and precise mapping of affected zones. The findings contribute significantly to the efficient and sustainable management of urban infrastructure, enhancing maintenance strategies and reducing potential hazards associated with undetected fluid leaks.

Geophysical methods reveal a subsurface historic wastewater tunnel exposed by a sinkhole: a case study in Bandung City, Indonesia

Bandung City, the capital of West Java Province in Indonesia, has changed its physical face rapidly due to the many constructions of office buildings, hotels, and other facilities until now. To avoid the instability of existing buildings and for building construction in the future, knowledge of the subsurface conditions in the city must be enhanced. A sudden soil collapse in the yard of the PAG Building occurred in early 2018 and resulted in a sinkhole with a diameter of 5 m, which revealed a historic wastewater tunnel structure in the subsurface. Non-invasive, low-cost, and time-effective geophysical methods to solve those problems are proposed. Ground-penetrating radar (GPR) and Electrical Resistivity Tomography (ERT), widely applied to identify artificial objects underground, came to be used for these purposes. A research to know the suitability of both methods for application began with numerical simulation, followed by field measurement on 4 GPR and ERT lines and analysis of each method. The results show that GPR data, through both numerical and field data, could identify and locate the tunnel in a radargram due to its hyperbola shape, whereas ERT data provided the property contrast between the tunnel and its host. The results suggest that GPR and ERT techniques are effective for revealing water tunnels in the study area with a maximum depth of 1.8 m and a 65% reduction in size. These findings can be used as a guide to using both methods to reveal the water tunnel network in the central part of Bandung City for sustainable urban planning.

Temperate glacier survey on the Yulong Snow Mountain using low-frequency ground-penetrating radar data with fast coherent enhancement

Temperate glacier survey on the Yulong Snow Mountain using low-frequency ground-penetrating radar data with fast coherent enhancement

Multi-objective waveform inversion of multi-offset surface ground-penetrating radar data

Multi-objective waveform inversion of multi-offset surface ground-penetrating radar data

De Geer moraine internal architecture based on sedimentological and geophysical investigations and implications for ice‐marginal reconstructions

De Geer moraines (DGMs) may act as valuable ice margin indicators; however, to date, their variable mode of formation has presented challenges for this utility. Morphometric investigations provide useful insights into formation processes, which can be developed using sedimentological and geophysical methods. Here we present sedimentological and ground penetrating radar (GPR) data of DGMs located in southwest Finland. Individual lithofacies were identified and interpreted using sediment architectural elements. These were correlated with neighbouring GPR radargrams and extrapolated across the wider study area. Generally, internal architecture presents a multi‐phase structure with lower units representing subglacial traction till and ice margin infill deposits, truncated by a larger prominent push unit, which is then successively deformed via the overriding of active ice. Significantly, there are notable differences between proximal and distal structures, with proximal sides characterized by silts, clays, and diamicton with laminae, stratification and thrust planes, and distal sides characterized by poorly consolidated diamicton and proglacial water current reworkings. Internal architecture of both prominent and intermediate ridges is very similar, reflecting similar formation processes, however, slight differences also reflect inter‐seasonal variations. Based on our findings, we present an integrated conceptual model for the genesis of DGMs whereby inter‐seasonal ridge forming processes occur within a sub‐aqueous ice‐marginal environment. Our model highlights that DGMs can be subcategorized as: (i) sediment deposition at an unstable margin during summer calving, and/or (ii) sediment pushing at a stabilized margin during a winter re‐advance. We do not find evidence of crevasse filling as a mechanism for DGM formation. We propose a landform assemblage classification whereby ‘De Geer terrain’ is used to describe series of parallel ridges arranged in a typical washboard‐like configuration. This classification identifies all DGMs derived within a sub‐aqueous ice‐marginal environment, whilst also capturing the equifinal characteristics between individual landforms.

Automatic Target Detection Using Ground Penetrating Radar (GPR) Data Based on A Machine Learning Approach: A Survey

Ground Penetrating Radar (GPR) is a valuable tool for subsurface exploration across diverse fields such as archaeology, defence, and civil engineering. Traditionally, GPR data has been interpreted manually, requiring extensive expertise and being time-consuming. Recently, machine- learning approaches have emerged as effective solutions for automating target detection, significantly reducing time and human dependency while enhancing accuracy. This review provides a comprehensive overview of the current advancements in automatic target detection using GPR data, focusing on machine learning-based methods. We summarize feature extraction techniques, popular machine learning algorithms, and challenges within this domain. This paper also identifies future directions and potential improvements for robust, accurate, and efficient GPR-based automatic target detection systems. Key Words: Machine Learning, Ground Penetrating Radar, Object Detection, Signal Processing.

Field Scale Soil Moisture Estimation with Ground Penetrating Radar and Sentinel 1 Data

This study examines the combined use of ground penetrating radar (GPR) and Sentinel-1 synthetic aperture radar (SAR) data for estimating soil moisture in a 25-decare field in Sivas, Türkiye. Soil moisture, vital for sustainable agriculture and ecosystem management, was assessed using in situ measurements, SAR backscatter analysis, and GPR-derived dielectric constants. A novel empirical model adapted from the classical soil moisture index (SSM) was developed for Sentinel-1, while GPR data were processed using the reflected wave method for estimating moisture at 0–10 cm depth. GPR demonstrated a stronger correlation within situ measurements (R2 = 74%) than Sentinel-1 (R2 = 32%), reflecting its ability to detect localized moisture variations. Sentinel-1 provided broader trends, revealing its utility for large-scale analysis. Combining these techniques overcame individual limitations, offering detailed spatial insights and actionable data for precision agriculture and water management. This integrated approach highlights the complementary strengths of GPR and SAR, enabling accurate soil moisture mapping in heterogeneous conditions. The findings emphasize the value of multi-technique methods for addressing challenges in sustainable resource management, improving irrigation strategies, and mitigating climate impacts.

Enhancing full-waveform inversion of field GPR data: A source-independent approach with dynamic reference selection via SE-Wave-U-Net

Ground-penetrating radar (GPR) is a widely used nondestructive detection tool. Full-waveform inversion (FWI) offers a reliable algorithm for accurately imaging GPR data. Despite its potential, FWI’s effectiveness is often compromised by unknown excitation sources, which can lead to computational failures, diminished imaging accuracy, or both. Addressing this challenge, this paper introduces a convolution-type objective function designed to mitigate the detrimental effects of these unknown excitation sources on FWI imaging. Critical to the success of this function is the precise selection of reference traces. Further refining the capability for signal separation, we have integrated squeeze-and-excitation (SE) modules into the traditional Wave-U-Net architecture, culminating in the development of the SE-Wave-U-Net framework. This framework is adept at dynamically extracting accurate reference traces from field GPR data, which are essential for accurate FWI imaging. The efficacy, accuracy, and practical applicability of our methods are validated through extensive analysis of numerical, laboratory experiments, and field GPR data. Our algorithm not only establishes a high-precision computational framework for FWI but also enhances the reliability of GPR imaging in complex environments.

Full-parametric and joint inversion of multimode surface wave data for identifying glacial ice thickness and freezing extent in subglacial sediments via the hunger games search algorithm

Determining glacier ice thickness and the extent of freezing in subglacial sediments are crucial in glaciological studies. Noninvasive geophysical methods, such as multichannel analysis of surface waves, are typically used for these tasks. In this study, we introduce a novel metaheuristic called hunger games search (HGS), which simulates the hunger-driven instincts and behavioral decisions of animals for the full-parametric inversion of seismic surface waves. We apply HGS to determine layer thicknesses, densities, S-wave velocities, and primary wave velocities of different layers through the joint inversion of multimode Rayleigh wave dispersion curves (RWDCs). This marks the first study using HGS for the inversion of dispersion data. Sensitivity studies of model parameters prior to the inversion indicate the necessity of postinversion uncertainty evaluations to mitigate the effects of varying sensitivity levels. In addition, a parameter tuning study is carried out to maximize the performance of HGS. Compared with some swarm intelligence-based optimizers (particle swarm optimization, cuckoo search, gray wolf optimization, sparrow search optimization, and whale optimization algorithm), HGS outperforms in the inversion of synthetic multimode RWDCs with 10% uncertainties with respect to a simulated glacier structure. In real data applications, a data set acquired at Midtdalsbreen, an outlet of the Norwegian Hardangerjøkulen ice cap, is inverted using the tailored HGS metaheuristic. The results obtained are consistent with previous geophysical studies. Furthermore, our analysis reveals that the performance of HGS when dealing with real applications is not highly sensitive to the selection of layers within the range of five to eight. The accuracy of HGS is undoubtedly contaminated by a larger model space; however, additional depth information derived from colocated ground-penetrating radar data can be directly integrated into HGS to obtain satisfactory results again.

A Method to Detect Concealed Damage in Concrete Tunnels Using a Radar Feature Vector and Bayesian Analysis of Ground-Penetrating Radar Data

Many machine learning (ML)-based detection methods for interpreting ground-penetrating radar (GPR) data of concrete tunnels require extensive labeled damage-state data for model training, limiting their practical use in concealed damage detection of in-service tunnels. This study presents a probabilistic, data-driven method for GPR-based damage detection, which exempts the requirement in the training process of supervised ML models. The approach involves extracting a radar feature vector (RFV), building a Bayesian baseline model with healthy data, and quantifying damage severity with the Bayes factor. The RFV is a complex vector obtained by radargram data fusion. Bayesian regression is applied to build a model for the relationship between real and imaginary parts of the RFV. The Bayes factor is employed for defect identification and severity assessment, by quantifying the difference between the RFV built with new observations and the baseline RFV predicted by the baseline model with new input. The probability of damage is calculated to reflect the influence of uncertainties on the detection result. The effectiveness of the proposed method is validated through simulated data with random noise and physical model tests. This method facilitates GPR-based hidden damage detection of in-service tunnels when lacking labeled damage-state data in the model training process.

Optimising Ground Penetrating Radar data interpretation: A hybrid approach with AI-assisted Kalman Filter and Wavelet Transform for detecting and locating buried utilities

Ground Penetrating Radar (GPR) is widely used for detecting buried utilities, but data interpretation remains challenging due to noise and clutter. Although various methods exist for processing GPR data, the Kalman Filter (KF) has been underutilised despite its strength as an estimator. Traditional KF-based algorithms in GPR studies often rely on chi-squared hypothesis testing, which requires expert-defined thresholds and can lead to biased or uncertain outcomes. This paper introduces a novel KF-based framework that addresses these limitations. The framework employs Kalman Filters for noise reduction, with an optimisation algorithm based on a genetic algorithm to fine-tune KF input parameters. A Normalised Innovation Squared (NIS) parameter is used to generate an NIS signal function for identifying anomalies. Additionally, discrete wavelet transforms are applied to the NIS signal function for anomaly detection and localisation, using varying decomposition levels and vanishing moments. Results demonstrate a proportional relationship between wavelet decomposition levels, selected wavelets, and the detection rates of true and false positives. Statistical analysis using receiver operating characteristic curves shows that the optimal detection rate for all tested wavelets occurs at decomposition levels 5 and 6. This framework enhances GPR data interpretation with minimal user interaction, representing a step forward toward autonomy in GPR data processing and interpretation.

Relationships between lobate debris aprons and lineated valley fill on Mars: Evidence for an extensive Amazonian valley glacial landsystem in Mamers Valles

We examine the characteristics and relationships of Lineated Valley Fill (LVF) and Lobate Debris Aprons (LDA) in Mamers Valles on Mars, a ∼950 km-long fretted valley at the dichotomy boundary. The relationships and distinctions between these glacial landforms are established by detailed analysis of LDA/LVF morphology, topography, and related features are assessed to understand their origin and modification. We document the transition from unconfined LDA to compressed and folded LVF and vice versa, implying that LDA and LVF are intimately related in morphology and mode of origin. Linear LDA dominate Mamers Valles, originating from alcoves, theater-like remnant crater rims, and tributary valleys, while circumferential LDA are arrayed around isolated mesas. Narrow valley areas display the convergence of lobes originating from either side, forming parallel linear ridges that deform into complex folds and become LVF, typically in a local and regional downvalley direction. In contrast, when LVF flows out of a topographically confined area, the material forms a piedmont-like LDA. Thus, local topography is the primary factor in determining whether a deposit will appear LVF-like, LDA-like, or have characteristics of both. Superimposed crater morphology and ground-penetrating radar data suggest the current presence of subsurface ice protected by ∼15–20 m of sublimation lag deposits, with minimal deformation and flow since superposed crater formation. Regional integration leads to the interpretation that the LDA-LVF exposures and ice entry points into the fretted valleys represent the waning stages of a more widespread regional Amazonian plateau glacial landsystem that occupied fretted terrain valleys formed earlier in the Late Noachian-Early Hesperian.

Research on Airborne Ground-Penetrating Radar Imaging Technology in Complex Terrain

The integration of ground-penetrating radar (GPR) with unmanned aerial vehicles (UAVs) enables efficient non-contact detection, performing exceptionally well in complex terrains and extreme environments. However, challenges in data processing and interpretation remain significant obstacles to fully utilizing this technology. To mitigate the effects of numerical dispersion, this paper develops a high-order finite-difference time-domain (FDTD(2,4)) three-dimensional code suitable for airborne GPR numerical simulations. The simulation results are compared with traditional FDTD methods, validating the accuracy of the proposed approach. Additionally, a Kirchhoff migration algorithm that considers the influence of the air layer is developed for airborne GPR. Different processing strategies are applied to flat and undulating terrain models, significantly improving the identification of shallowly buried targets. Particularly under undulating terrain conditions, the energy ratio method is introduced, effectively suppressing the interference of surface reflections caused by terrain variations. This innovative approach offers a new technical pathway for efficient GPR data processing in complex terrains. The study provides new insights and methods for the practical application of airborne GPR.

Full-Waveform Inversion of Two-Parameter Ground-Penetrating Radar Based on Quadratic Wasserstein Distance

Full-waveform inversion (FWI) is one of the most promising techniques in current ground-penetrating radar (GPR) inversion methods. The least-squares method is usually used, minimizing the mismatch between the observed signal and the simulated signal. However, the cycle-skipping problem has become an urgent focus of this method because of the nonlinearity of the inversion problem. To mitigate the issue of local minima, the optimal transport problem has been introduced into full-waveform inversion in this study. The Wasserstein distance derived from the optimal transport problem is defined as the mismatch function in the FWI objective function, replacing the L2 norm. In this study, the Wasserstein distance is computed by using entropy regularization and the Sinkhorn algorithm to reduce computational complexity and improve efficiency. Additionally, this study presents the objective function for dual-parameter full-waveform inversion of ground-penetrating radar, with the Wasserstein distance as the mismatch function. By normalizing with the Softplus function, the electromagnetic wave signals are adjusted to meet the non-negativity and mass conservation assumptions of the Wasserstein distance, and the convexity of the method has been proven. A multi-scale frequency-domain Wasserstein distance full-waveform inversion method based on the Softplus normalization approach is proposed, enabling the simultaneous inversion of relative permittivity and conductivity from ground-penetrating radar data. Numerical simulation cases demonstrate that this method has low initial model dependency and low noise sensitivity, allowing for high-precision inversion of relative permittivity and conductivity. The inversion results show that it, in particular, significantly improves the accuracy of conductivity inversion.

Tunnel Quality Inspection Based on GPR and Machine Learning

Abstract The development of non-destructive testing, particularly ground penetrating radar (GPR) technology, has gained popularity in underground detection due to its high resolution and non-destructive capabilities. However, the lack of intuitive imaging methods in GPR data analysis leads to poor interpretability, requiring skilled interpreters. To address this, the article suggests a machine learning-based algorithm for tunnel defect detection using GPR data. The YOLOv5 model network is used to train GPR data from actual road and railway tunnel inspections, and a CBAM attention mechanism is added to the original model to achieve high-precision radar scanning image defect recognition and localization. Compared with the original recognition model, the optimized recognition algorithm significantly improves recognition accuracy and greatly improves the efficiency of defect recognition compared to manual interpretation.

Full-waveform Inversion of Noisy GPR Data in Frequency Domain Based on Quasi-Newtonian Algorithm

Abstract Ground penetrating radar (GPR) is a non-destructive detection technology. Full-waveform inversion (FWI) of GPR data promise for the quantitative characterization of the earth’s shallow subsurface parameters, which is useful for the interpretation and subsurface mapping. In this paper, we present a full-waveform inversion target imaging framework for GPR. In the method, the singular value decomposition (SVD) and least squares (LS) methods are used to denoise the measured scattering field. Secondly, the contrast source inversion (CSI) method is improved by introducing the quasi-Newtonian algorithm. Theoretical analysis and practice show that the proposed method can improve the robustness and stability of the inversion, and finally achieve a result with lower contrast error.

Enhanced Scoring Instance Segmentation Network for Subsurface Targets Recognition from GPR B-Scans

Abstract Ground Penetrating Radar (GPR) is a non-destructive technique used for detecting subsurface structure and space. However, the interpretation of field GPR data is quite labour-intensive and time-consuming. This paper develops a deep learning-based approach for the detection and segmentation of subsurface targets based on GPR data. Since existing deep learning methods do not always accurately reflect the true quality of segmentation masks and exhibit locality when handling details in complex scenes, the proposed method employs the Mask Scoring R-CNN (MS R-CNN) framework as its primary framework and makes additional modifications to its backbone and neck architecture. Experimental results on GPR bridge dataset show notable improvements in loss metrics, segmentation accuracy, and bounding box precision. These enhancements have effectively reduced overall losses and enhanced the performance of Average Precision index of the enhanced MS R-CNN, demonstrating the effectiveness of our approach in the accurate identification of underground objects under complex environments.

Adaptive Variational Mode Decomposition Denoising Scheme for Ground Penetrating Radar Data

Abstract Addressing the impact of noise on target detection performance in Ground Penetrating Radar (GPR), an adaptive Variational Mode Decomposition (VMD) denoising scheme based on the Whale Optimization Algorithm (WOA) is proposed. An improved adaptive VMD method, based on WOA, is employed in this study for modal decomposition of GPR data, with target modes being retained and noise modes removed to obtain denoised GPR data. It is demonstrated by numerical simulation results that the SNR is increased by 20.846 dB through this scheme, surpassing common denoising methods and achieving the best results in terms of RMSE and MSE. Numerical simulations substantiate the scheme’s effectiveness in denoising GPR data.

Spatial variability of hydraulic parameters of a cropped soil using horizontal crosshole ground penetrating radar

AbstractSoil hydraulic parameters (SHP) play a crucial role controlling the spatiotemporal distribution of water in the soil–plant continuum and thus affect water availability for crops. To provide reliable information on the SHP at different scales, measurement techniques with a good spatial resolution and low labor costs are required. In this study, we used crosshole ground penetrating radar (GPR)‐derived soil water contents (SWCs) measured along horizontal rhizotubes under a controlled experimental test site cropped with winter wheat to estimate the unimodal and dual‐porosity soil hydraulic characteristics with different soil layer setups. Therefore, sequential inversion of the GPR‐derived SWCs was performed using the hydrological model HYDRUS‐1D, whereby the SWC data were either averaged prior inversion or used in a spatially distributed way. To analyze if the time‐lapse gathered GPR data contain enough information to estimate the SHP, additional synthetic studies were performed increasing the data resolution to daily GPR measurements. The results showed that the time‐lapse data contained enough information to estimate the SHP accurately. Additionally, spatially distributed soil hydraulic characteristics differed from the one estimated based on averaged SWCs derived from spatially distributed GPR data. Finally, we derived spatially resolved SHP, which can be used for 3D process rhizosphere processes and root–soil interaction modeling.

Deviation Aware Mask for Localizing Ground Penetrating Radar Data Alignment

Abstract This paper proposes a data align pipeline for LGPR (Localization Ground Penetrating Radar). LGPR achieves robot localization by matching prior underground data map, therefore high-precision maps are essential. Utilizing GPR in a time acquisition mode for robot localization enables the technology to generate maps with resolutions far superior to those produced by camera or lidar mapping methods. However, it also introduces several challenges. On one hand, the process of map generation heavily relies on velocity control of the moving platform, and non-uniform motion in the localization scenario can result in feature deformation of GPR images. On the other hand, in the absence of relative a priori position conditions, there exists a certain degree of spatial dimension offset between the multiple data acquisitions from array GPR systems, rendering existing time-series deformation correction schemes ineffective. To address these challenges, by designing a deviation-aware distance, we aim to enhance existing time-series regularization approaches without relying on additional sensor position information, thus achieving synchronized alignment of multiple rounds of ground-penetrating radar data for the generation of high-quality underground map data used to evaluate LGPR performance. In this work, we compare existing synchronization strategies of LGPR with those of other sensors and alignment schemes, and validate the effectiveness of our proposed approach.