Search

Ground-penetrating radar (GPR) is becoming a mainstream tool in planetary exploration, and one of the few in-situ planetary geophysical methods. There are currently three missions (Perseverance, Tianwen-1, Chang'E-4) with GPR-equipped rovers, and two future missions (Chang'E-7, ExoMars) that will include GPR in their scientific payload. The large number of GPR data, combined with the novel setup of the measurements, creates the need for new data processing and interpretation techniques to address the unique challenges of in-situ planetary radar. The current paper proposes an interpretation pipeline that starts with a novel stochastic hyperbola fitting that estimates the probability kernel density of the bulk permittivity at different depths. Subsequently, the bulk permittivity distribution is transformed via a novel probabilistic inversion to a 1-dimensional (1D) permittivity profile. The inverted 1D permittivity profile is then used as an input to a bespoke reverse-time migration (RTM) using the finite-difference time-domain (FDTD) method. RTM using FDTD does not assume a clinical homogeneous half-space; instead, it accounts for the expected layered structure of the investigated medium. Lastly, the migrated radargram is clustered in order to identify subsurface targets and distinguish them from the background medium. Each of the processing steps has never been reported in planetary radar; and together act as a complete processing toolbox tuned for planetary science. The suggested interpretation pipeline is validated numerically in a 1D case study with a complex layered structure and multiple subsurface targets. The proposed processing scheme is then applied to the GPR data from the Chang'E-4 mission at the Von Karman crater, revealing a previously unseen layered structure and a complex distribution of rocks/boulders.

Internal decays in trees can rapidly escalate into a full decomposition of the inner structural layer, i.e., the “heartwood” layer, due to the action of aggressive diseases and fungal infections. This process leads to the formation of big cavities and hollows, which remain surrounded by the sapwood layer only. Estimating the thickness of the sapwood layer with a high degree of accuracy is therefore crucial for a correct assessment of the structural integrity of hollow trees, as well as an extremely challenging task. In this context, ground-penetrating radar (GPR) has proven effective in providing details of the internal structure of trees. Nevertheless, the existing GPR processing methods still offer limited information on their internal configuration. This study investigates the effectiveness of GPR enhanced by a microwave tomography inversion approach in the assessment of hollow trees. To this aim, a living hollow tree was investigated by performing a set of pseudo-circular scans along the bark perimeter with a hand-held common-offset GPR system. The tree was then felled, and sections were cut for testing purposes. A dedicated data processing framework was developed and tested through numerical simulations of hollow tree sections. The internal structure of the real trunk was therefore reconstructed via a tomographic imaging approach and the outcomes were quantitatively analysed by way of comparison with the real sections’ main geometric features. The tomographic approach has proven very accurate in locating the sapwood-cavity interface as well as in the evaluation of the sapwood layer thickness, with a centimetre prediction accuracy.

Hyperbola fitting is a mainstream interpretation technique used in ground-penetrating radar (GPR) due to its simplicity and relatively low computational requirements. Conventional hyperbola fitting is based on the assumption that the investigated medium is a homogeneous half-space, and that the target is an ideal reflector with zero radius. However, the zero-radius assumption can be easily removed by formulating the problem in a more generalized way that considers targets with arbitrary size. Such approaches were recently investigated in the literature, suggesting that hyperbola fitting can be used not only for estimating the velocity of the medium, but also for estimating the radius of subsurface cylinders, a very challenging problem with no conclusive solution to this day. In this letter, through a series of synthetic and laboratory experiments, we demonstrate that for practical GPR survey, hyperbola fitting is not suitable for simultaneously estimating both the velocity of the medium and the size of the target, due to its inherent nonuniqueness, making the results unreliable and sensitive to noise.

In hydrocarbon drilling, mud filtrate penetrates permeable formations and alters the pore fluid characteristics in the immediate vicinity of the borehole. Typically, the prevailing in situ pore fluids are displaced by the invading mud filtrate, which leads to gradually changing distributions of the fluid and electrical properties. Understanding this invasion process is crucial for the interpretation of logging data and associated reservoir evaluations. Conventional logging methods tend to be inadequate for this purpose as their resolution is too low. We find that invasion depth can be determined from borehole radar data using an optimized antenna configuration and time-lapse measurements. A series of parametric sensitivity analyses provide information about the effects of variations of the rock and fluid properties on the identification and extraction of borehole radar signals reflected from the invasion front. Our results suggest that by embedding the radar antennas in cavities filled with an absorbing dielectric material, it is possible to minimize the interference arising from the metal components of the logging tool. In the simulated reservoir scenario, a time-lapse measurement mode with a time interval of at least 6 h can reliably extract the radar signals reflected from the invasion front, and the proposed borehole radar has a lateral detection range from 0.15 to 1 m. A comprehensive range of parametric sensitivity analyses indicates that the signals reflected from the invasion front are principally influenced by oil viscosity, porosity, and mud and formation water salinity, as well as by molecular diffusion coefficient and cementation exponent. These properties and parameters should be carefully explored and assessed when applying borehole radar to evaluate mud invasion information in a reservoir environment.

Introduction: The Perseverance rover landed at Jezero crater in 2021, close to the delta of an ancient river (1). This crater once hosted a paleolake (2). Previous studies of this system have concluded that the current fan-delta units were formed during the last phase of the Jezero fluvial activity, which postdates the formation of the olivine-rich unit and predates the formation of the current floor unit. From crater counts, the formation of the delta unit is assigned to an Early or Late Hesperian age (3.5 + 0.1/-0.3 Gy) (3). However, little is known about the processes that happened before and the plausible fluvial and igneous history that determined the evolution of the fresh impact crater. A comparative analysis with craters of similar size suggests that Jezero crater has experienced approximately 1 km of infilling compared to the morphologically fresh crater(4,5). The crater floor depth d can be calculated with the power-law scaling d = 0.372D0.375 + 0.072D0.62 (6) , which is valid for craters wider than 7 km. For Jezero D=49 km and the crater floor base should be at a depth of 2,404 m. The purpose of this work is to investigate the plausible existence of preserved radar reflectors (7–12) that may indicate the geometry of crater filling layers. Recent research by other authors (7) have confirmed the usefulness of this approach to detect preserved layers in other craters. Here we shall apply a similar methodology to investigate it on Jezero crater. Materials and methods: SHARAD is a sounding radar experiment aboard the MRO mission designed to reveal the shallow structures of the Martian subsurface.(9). We visualize SHARAD radargrams using JMARS software. When a potential echo is detected as a bright feature, these signals have to be compared with the simulated cluttergrams, that are produced using the current surface topography, (MOLA-HRSC with 200m/pix. and MOLA with 463m/pix) in order to discriminate ghost signals. The protocol followed to investigate the plausible existence of buried reflectors on MARS using SHARAD data is the following one: Interpret visually the radargrams looking for subsurface reflectors in JMARS. Download the geometry data for each radargram under investigation and crop only the area of study, to reduce the computational time. Build the clutter simulation using the sets of topography data mentioned above. The data were downloaded from astrogeology. The DTM was selected only for the area of interest to reduce the overall computational resources. The clutter simulator estimates the reflected echoes from the topography based on the directivity pattern of the antenna, the distance from the source, and the orientation of the slope with respect to the angle of incidence (13) Compare the off-nadir reflections with the apparent bright reflectors. If a true reflector is detected, the wave travel times can be transformed to depth using Δh=c Δt/√ε where ε is the assumed permittivity of the materials above the reflector. In order to test our methodology we validate it in a previously described case, in a crater near Mangala Fossa region (7). A confirmed reflector, which is distinguished in the radargram underneath the subsurface (Fig1-b) is not seen in the cluttergram (Fig1-a) confirming it as a true buried reflector. Any other bright feature that appears in both images is an artifact produced by the radar echo of the surface. Next we apply this methodology to Jezero crater. We observe in Figure 2-b an example of a potential reflector. To confirm or discard the nature of this reflector we compare with the cluttergrams at different resolution scales, in Figure 2-a and 2-c. The reflectors from SHARAD data are observed on the radagrams at both resolution scales and therefore they are interpreted as artifacts produced by surface topographic echoes. Similar examples are shown in Figures 3 and 4. Conclusions: Our preliminary analysis, can discard the existence of true recognizable reflectors at Jezero. Assuming ε= 6 (14) and a crater filling depth of about 1 km, a crater-base reflector should have given an echo at a single travel time less or equal than 8165 ns. This distance is covered in our investigation and therefore this analysis suggests that SHARAD measurements can neither detect any reflector within the crater filling nor distinguish the crater floor. This may be due to the presence of strong off nadir reflections produced by the surface topography that may overwrite true reflectors. Moreover, gradual variations between layers can reduce their reflection coefficients making them transparent to radar. Future studies will investigate the existence of subsurface reflectors in environments with higher signal to clutter ratio, such as paleolake environments with smoother surfaces. Acknowledgments: We are grateful to the PDS, SHARAD and HRSC for the materials and JMARS and QGIS for the software. The work of F. Mansilla is supported by the grant PRE2020-09170 founded by MCIN/AEI/10.13039/501100011033. Spanish State Research Agency (AEI) Project No. MDM‐2017‐0737 Unidad de Excelencia “María de Maeztu”‐ Centro de Astrobiología (INTA‐CSIC) and the Spanish Ministry of Science and Innovation (PID2019‐104205GB‐C21). References: N. Mangold et al., Science. 374, 711–717 (2021). C. I. Fassett, J. W. Head, Geophys. Res. Lett.32, 14, (2005) N. Mangold et al., Astrobiology. 20, 994–1013 (2020). S. C. Schon, et al., Planet. Space Sci. 67, 28–45 (2012). B. D. Boatwright, J. W. Head, Planet. Sci. J. 2, 52 (2021). L. Pan, et al., Nat. Commun. 10, 4257 (2019). R. R. Bharti, et al., Icarus. 371, 114713 (2022). I. Ganesh, L. M. Carter, I. B. Smith, J. Volcanol. Geotherm. Res. 390, 106748 (2020). R. Seu et al., J. Geophys. Res., 112, E5 (2007). M. N. Simon, et al., J. Geophys. Res. Planets. 119, 2291–2299 (2014). T. R. Watters et al., Science. 318, 1125–1128 (2007). S. Xiong et al., Earth Space Sci. 8 (2021). P. Choudhary, et al., IEEE Geosci. Remote Sens. Lett. 13, 1285–1289 (2016). S. Eide, et al., IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 14, 2484–2493 (2021).

Chang’E-4 was the first mission to land a human object on the far side of the Moon. The landing site was at the Von Kármán (VK) crater at the South-Pole Aitken (SPA) basin, one of biggest craters in the solar system. SPA is believed that was created by a huge impact that penetrated the lunar crust and uplifted mantle materials. Evidence of these materials are expected to be found by the Yutu-2, the rover of the mission that is still active to this day, having covered more than 1km on the lunar’s surface. Yutu-2 is equipped with a stereo camera, visible/near-infrared imaging spectrometer, alpha particle x-ray spectrometer and Ground-Penetrating Radar (GPR). In-situ GPR is a powerful geophysical methodology with a uniquely wide range of applications to civil engineering, archaeology and geophysics. In planetary science, it was first used in 2013 during the Chang’E-3 mission. Since then, GPR has become a very popular instrument in planetary missions, and has been included in the scientific payload of Chang’E-4, E-5, Tianwen-1, and Perseverance. It is also planned to be used in the future missions Chang’E-7 (2024) and ExoMars (September 2022). Yutu-2 rover is equipped with three different GPR systems. One low frequency and two high frequency antennas. Unfortunately, due to interferences between the antenna and the metallic parts of the rover, the low frequency data have very low signal to clutter ratio making the interpretation of these data unreliable. On the other hand, the signal from the high frequency antennas is very clear, probably due the lack of ilmenite in the area, which results to low electromagnetic losses (compared to the Chang’E-3 landing site). This resulted to good quality radagrams that provided new insights into the structure and composition of the top ejecta layers at the VK crater. In the current paper, we introduce a complete processing scheme, tuned for high frequency lunar penetrating radar. The first step of the proposed framework is an advanced hyperbola fitting (AHF) capable of inferring previously unseen layers due to their smooth boundaries. Subsequently, the reconstructed layered structure is used in a Reverse-Time Migration (RTM) coupled with Finite-Differences Time-Domain (FDTD) method. Via this approach, the radagram is focused subject to a 1D model, avoiding homogeneity constrains that often deviate from reality. Lastly, an un-supervised thresholding is applied to cluster the migrated image into two categories i.e. A) the background host medium and B) rocks/boulders. The suggested scheme is applied to the high frequency data collected by the Yutu-2 rover at the first 100 meters of the mission. A layered structure is inferred at the top 12 meters, similar to the results presented in [1]. Moreover, using the proposed RTM, an abundance of rocks/boulders was revealed. The distribution of the rocks/boulders correlates with the permittivity/density profile, indicating the reliability of the proposed scheme. References[1] Giannakis, I., Zhou, F., Warren, C., & Giannopoulos, A. (2021). Inferring the shallow layered structure at the Chang’E-4 landing site: A novel interpretation approach using lunar penetrating radar. Geophysical Research Letters, 48, e2021GL092866

Ash dieback, acute oak decline (AOD), and Xylella Fastidiosa are emerging infectious diseases (EIDs) that have spread rapidly in European forests during the last decade. Quarantine measurements have mostly failed to repress the outbreaks and millions of trees have already been infected. Identifying infected trees in a nondestructive manner is of high importance for monitoring, managing, and preventing EIDs. The aim of this article is to examine the capabilities of ground penetrating radar (GPR) on evaluating the internal structure of tree trunks and detecting tree decay associated with EIDs. Traditionally used processing schemes tuned for GPR line acquisitions are modified accordingly to be compatible with the new measurement configurations. In particular, a detection framework is presented based on a modified Kirchhoff and a reverse-time migration. Both of the aforementioned methodologies are compatible with measurements taken along closed irregular curves assuming a homogeneous permittivity distribution. To that extent, prior to migration, a novel focal criterion is used that estimates the bulk permittivity of the host medium from the measured B-scans. The suggested detection scheme is successfully tested on both numerical and laboratory measurements, indicating that GPR has the potential to become a coherent and practical tool for detecting tree decay associated with EIDs.

Full-waveform inversion (FWI) is considered one of the most promising interpretation tools for hydrogeological applications using ground-penetrating radar. However, FWI has had limited practical uptake for several reasons: large computational requirements, an inability to reconstruct loss mechanisms of soil, and the need for a good initial starting model. We aim to address these issues via a novel FWI subject to a fractally correlated distribution of water. Initially, the dispersive properties of the soil are expressed as a function of the water fraction using a semiempirical model. This approach means that the permittivity, conductivity, and relaxation mechanisms are all correlated, and therefore, sensitivity problems between the permittivity and loss mechanisms no longer affect the performance of FWI. Subsequently, the distribution of the water fraction is constrained to follow a fractal geometry. Fractal-correlated noise is then compressed using a principal component analysis (PCA) in order to further reduce the number of the system's unknowns and accelerate FWI. PCA reduces the volume and dimensions of the optimization space, and thus, initialization is no longer necessary. Finally, a novel measurement configuration is suggested that uses superposition with all the individual measurements in order to reduce the number of forward models that need to be executed for every iteration of FWI. These enhancements substantially reduce the computational requirements of FWI and therefore eliminate the need for high-performance computers and time-consuming algorithms. The proposed scheme has been successfully tested with several numerical case-studies, which indicates the potential of this approach to become a commercially appealing interpretation tool for hydrogeology.

The simulation, or forward modeling, of Ground Penetrating Radar (GPR) is becoming a more frequently used approach to facilitate interpretation of complex real GPR data, and as an essential component of full-waveform inversion (FWI). However, general full-wave 3D electromagnetic (EM) solvers, such as ones based on the Finite-Difference Time-Domain (FDTD) method, are still computationally demanding for simulating realistic GPR problems. We have developed a novel near real-time, forward modeling approach for GPR that is based on a machine learning (ML) architecture. The ML framework uses an innovative training method which combines a predictive principal component analysis technique, a detailed model of the GPR transducer, and a large dataset of modeled GPR responses from our FDTD simulation software. The ML-based forward solver is parameterized for a specific GPR application, but the framework can be applied to many different classes of GPR problems.To demonstrate the novelty and computational efficiency of our ML-based GPR forward solver, we used it to carry out FWI for a common infrastructure assessment application – determining the location and diameter of reinforcement bars in concrete. We tested our FWI with synthetic and real data, and found a good level of accuracy in determining the rebar location, size, and surrounding material properties from both datasets. The combination of the near real-time computation, which is orders of magnitude less than what is achievable by traditional full-wave 3D EM solvers, and the accuracy of our ML-based forward model is a significant step towards commercially-viable applications of FWI of GPR.

The Finite-Difference Time-Domain (FDTD) method is a popular numerical modelling technique in computational electromagnetics. The volumetric nature of the FDTD technique means simulations often require extensive computational resources (both processing time and memory). The simulation of Ground Penetrating Radar (GPR) is one such challenge, where the GPR transducer, subsurface/structure, and targets must all be included in the model, and must all be adequately discretised. Additionally, forward simulations of GPR can necessitate hundreds of models with different geometries (A-scans) to be executed. This is exacerbated by an order of magnitude when solving the inverse GPR problem or when using forward models to train machine learning algorithms. We have developed one of the first open source GPU-accelerated FDTD solvers specifically focused on modelling GPR. We designed optimal kernels for GPU execution using NVIDIA’s CUDA framework. Our GPU solver achieved performance throughputs of up to 1194 Mcells/s and 3405 Mcells/s on NVIDIA Kepler and Pascal architectures, respectively. This is up to 30 times faster than the parallelised (OpenMP) CPU solver can achieve on a commonly-used desktop CPU (Intel Core i7-4790K). We found the cost–performance benefit of the NVIDIA GeForce-series Pascal-based GPUs – targeted towards the gaming market – to be especially notable, potentially allowing many individuals to benefit from this work using commodity workstations. We also note that the equivalent Tesla-series P100 GPU – targeted towards data-centre usage – demonstrates significant overall performance advantages due to its use of high-bandwidth memory. The performance benefits of our GPU-accelerated solver were demonstrated in a GPR environment by running a large-scale, realistic (including dispersive media, rough surface topography, and detailed antenna model) simulation of a buried anti-personnel landmine scenario. New version program summary Program Title: gprMax Program Files doi: http://dx.doi.org/10.17632/kjjm4z87nj.1 Licensing provisions: GPLv3 Programming language: Python, Cython, CUDA Journal reference of previous version: Comput. Phys. Comm., 209 (2016), 163–170 Does the new version supersede the previous version?: Yes Reasons for the new version: Performance improvements due to implementation of CUDA-based GPU engine Summary of revisions: A FDTD solver has been written in CUDA for execution on NVIDIA GPUs. This is in addition to the existing FDTD solver which has been parallelised using Cython/OpenMP for running on CPUs. Nature of problem: Classical electrodynamics Solution method: Finite-Difference Time-Domain (FDTD)

Full-waveform inversion (FWI) using ground-penetrating radar (GPR) is gaining momentum as a powerful hydrogeological tool for inferring the hydraulic properties of soils between boreholes [1]. Nonetheless, the large computational requirements of FWI make it often unattainable with limited practical uptake [2]. In addition, the inability to accurate reconstruct the loss mechanisms and the need for a good initial model, further reduce the applicability of FWI [1], [2].In order to overcome the aforementioned limitations, we suggest a novel framework that substantially reduces the optimization space of FWI which consequently reduces the overall computational requirements [2]. This methodology assumes that the water fraction of the investigated medium follows a fractal distribution [3]. Based on that, using a principal components analysis on 3000 randomly generated fractals, we build an orthonormal basis that is fine-tuned for fractal correlated noise. Furthermore, it is proven [2], that fractal correlated noise is highly compressible and can be sufficiently represented with just 30-40 principal components. This reduces the optimization space since now FWI needs to fine-tune just these 30-40 parameters instead of every cell of the investigated medium [2].The involved fractals describe the distribution of the water fraction that is subsequently transformed to dielectric properties via a semi-empirical formula that relates readily available soil properties to the frequency depended complex electric permittivity [4], [5]. Via this approach, we overcome the need for a simultaneous FWI for both permittivity and conductivity [6]. This further reduces the optimization space and overcomes pitfalls associated with reconstructing loss mechanisms [2].References[1] Klotzsche, A., Vereecken, H., & Kruk van der J., (2019), Review of Crosshole Ground-Penetrating Radar Full-Waveform Inversion of Experimental Data: Recent Developments, challenges, and Pitfalls, Geophysics, vol. 84, pp. H13-H28.[2] Giannakis, I, Giannopoulos, A., Warren, C. & Sofroniou, A., (2021), Fractal-Constrained Crosshole/Borehole-to-Surface Full Waveform Inversion for Hydrogeological Applications Using Ground-Penetrating Radar, IEEE Transactions on Geoscience and Remote Sensing, Early Access.[3] Turcotte, L. (1992), Fractal and Chaos in Geology and Geophysics, Cambrige, UK: The Press Syndicate of the University of Cambridge.[4] Peplinski, N. R., Ulaby, F. T., & Dobson, M. C., (1995), Dielectric Properties of Soils in the 0.3-1.3 GHz Range, IEEE Transactions on Geoscience and Remote Sensing, vol. 33, no. 3, pp. 803-807.[5] Giannakis, I., Realistic Numerical Modelling of Ground Penetrating Radar for Landmine Detection, (2016), PhD Thesis Submitted at The University of Edinburgh.[6] Meles, G. A., Kruk, van der J., Grennhalgh, S. A., Ernst, J. R., Maurer, H & Green, A. G., (2010), A New Vector Waveform Inversion Algorithm for Simultaneous Updating of Conductivity and Permittivity Parameters from Combination Cross/Borehole-to-Surface GPR Data, IEEE Transactions on Geoscience and Remote Sensing, vol. 48, no. 9, pp. 3391-3407.

Lack of well-labelled and coherent training data is the main reason why machine learning (ML) and data-driven interpretations are not established in the field of Ground-Penetrating Radar (GPR). Non-representative and limited datasets lead to non-reliable ML-schemes that overfit, and are unable to compete with traditional deterministic approaches. To that extent, numerical data can potentially complement insufficient measured datasets and overcome this lack of data, even in the presence of large feature spaces.Using synthetic data in ML is not new and it has been extensively applied to computer vision. Applying numerical data in ML requires a numerical framework capable of generating synthetic but nonetheless realistic datasets. Regarding GPR, such a framework is possible using gprMax, an open source electromagnetic solver, fine-tuned for GPR applications [1], [2], [3]. gprMax is fully parallelised and can be run using multiple CPU’s and GPU’s. In addition, it has a flexible scriptable format that makes it easy to generate big data in a trivial manner. Stochastic geometries, realistic soils, vegetation, targets [3] and models of commercial antennas [4], [5] are some of the features that can be easily incorporated in the training data.The capability of gprMax to generate realistic numerical datasets is demonstrated in [6], [7]. The investigated problem is assessing the depth and the diameter of rebars in reinforced concrete. Estimating the diameter of rebars using GPR is particularly challenging with no conclusive solution. Using a synthetic training set, generated using gprMax, we managed to effectively train ML-schemes capable of estimating the diameter of rebar in an accurate and efficient manner [6], [7]. The aforementioned case studies support the premise that gprMax has the potential to provide realistic training data to applications where well-labelled data are not available, such as landmine detection, non-destructive testing and planetary sciences.References[1] Warren, C., Giannopoulos, A. & Giannakis, I., (2016). gprMax: Open Source software to simulate electromagnetic wave propagation for Ground Penetrating Radar, Computer Physics Communications, 209, 163-170.[2] Warren, C., Giannopoulos, A., Gray, A., Giannakis, I., Patterson, A., Wetter, L. & Hamrah, A., (2018). A CUDA-based GPU engine for gprMax: Open source FDTD, electromagnetic simulation software. Computer Physics Communications, 237, 208-218.[3] Giannakis, I., Giannopoulos, A. & Warren, C. (2016). A realistic FDTD numerical modeling framework of Ground Penetrating Radar for landmine detection. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 9(1), 37-51.[4] Giannakis, I., Giannopoulos, A. & Warren, C., (2018). Realistic FDTD GPR antenna models optimized using a novel linear/non-linear full waveform inversion. IEEE Transactions on Geoscience and Remote Sensing, 207(3), 1768-1778.[5] Warren, C., Giannopoulos, A. (2011). Creating finite-difference time-domain models of commercial ground-penetrating radar antennas using Taguchi’s optimization method. Geophysics, 76(2), G37-G47[6] Giannakis, I., Giannopoulos, A. & Warren, C. (2021). A Machine Learning Scheme for Estimating the Diameter of Reinforcing Bars Using Ground Penetrating Radar. IEEE Geoscience and Remote Sensing Letters.[7] Giannakis, I., Giannopoulos, A., & Warren, C. (2019). A machine learning-based fast-forward solver for ground penetrating radar with application to full-waveform inversion. IEEE Transactions on Geoscience and Remote Sensing. 57(7), 4417-4426.

Ground penetrating radar (GPR) is a well-established tool for detecting and locating reinforcing bars (rebars) in concrete structures. However, using GPR to quantify the diameter of rebars is a challenging problem that current processing approaches fail to tackle. To that extent, we have developed a novel machine learning framework that can estimate the diameter of the investigated rebar within the resolution range of the employed antenna. The suggested approach combines neural networks and a random forest regression and has been trained entirely using synthetic data. Although the training process relied only on numerical training sets, nonetheless, the suggested scheme is successfully evaluated with real data indicating the generalization capabilities of the resulting regression. The only required input of the proposed technique is a single A-scan, avoiding laborious measurement configurations and multisensor approaches. In addition, the results are provided in real time and making this method practical and commercially appealing.

This paper proposes a nondestructive evaluation method based on deep learning using combined ground-penetrating radar (GPR) and electromagnetic induction (EMI) data for autonomic and accurate estimation of the cover thickness and diameter of reinforcement bars. A real-time object detection algorithm—You Only Look Once–version 3 (YOLO v3)—is adopted to automatically identify the reinforcement bar reflected signals from radargrams, with which the range of the cover thickness is roughly predicted. Subsequently, EMI data, accompanied with the cover thickness range, are imported to a one-dimensional convolutional neural network (1D CNN), pretrained by calibrated EMI and GPR data, to simultaneously estimate the cover thickness and reinforcement bar diameter. Testing with the on-site GPR data shows that YOLO v3 is superior to Single Shot Multibox Detector method in GPR hyperbolic signal identification. Testing of 1D CNN with the EMI and GPR data collected in an in-house sand pit experiment shows that the estimation accuracy of the cover thickness and reinforcement bar diameter is, respectively, 96.8% and 90.3% with a permissible error of 1 mm. Further, an experiment with concrete specimens demonstrates that among the 22 estimated values (including the reinforcement bar diameter and cover thickness), there are 17 values accurately estimated, while the inaccurately estimated values have an error up to 2 mm. The experimental results show that the proposed method can autonomically evaluate the reinforcement bar diameter and cover thickness with a high accuracy.

The preservation of natural assets is nowadays an essential commitment. In this regard, root systems are endangered by fungal diseases which can undermine the health and stability of trees. Within this framework, ground penetrating radar (GPR) is emerging as a reliable non-destructive method for root investigation. A coherent GPR-based root-detection framework is presented in this paper. The proposed methodology is a multi-stage data analysis system that is applied to semi-circular measurements collected around the investigated tree. In the first step, the raw data are processed by applying several standard and advanced signal processing techniques in order to reduce noise-related information. In the second stage, the presence of any discontinuity element within the survey area is investigated by analysing the signal reflectivity. Then, a tracking algorithm aimed at identifying patterns compatible with tree roots is implemented. Finally, the mass density of roots is estimated by means of continuous functions in order to achieve a more realistic representation of the root paths and to identify their length in a continuous and more realistic domain. The method was validated in a case study in London (UK), where the root system of a real tree was surveyed using GPR and a soil test pit was excavated for validation purposes. Results support the feasibility of the data processing framework implemented in this study.

In this paper, a coherent framework for estimating the density and the distribution of roots using ground penetrating radar is presented. The proposed methodology is a multi-stage data processing scheme that is applied in semi-circular measurements collected concentrically around the investigated tree. The adopted processing methodology consists of three distinct and sequential steps. In the first step, the raw B-scans are subject to time-zero correction, zero-offset removal, time-varying gain and the singular value decomposition (SVD) filter. The SVD filter is used in order to effectively eliminate multiples and ringing noise from the B-Scans and increase the overall signal-to-clutter ratio. The second step consists of a tracking algorithm that aims at identifying patterns that resemble tree roots. In the last step, a continuous function is fitted to each root in order to effectively interpolate between points and subsequently estimate the density of the roots. This paper concludes with a case study on an urban tree at the Gunnersbury Park, London, United Kingdom. The top soil around the tree was excavated to 40 cm below the surface approximately in order to expose the tree root architecture. Then the exposed tree root system was used to compare the survey results for validity purposes and ultimately support the viability of the proposed data processing methods adopted in this investigation.

Street trees are broadly acknowledged to be an essential asset of cities. Nonetheless, the growth of tree roots can lead to critical damage, such as the uplifting or cracking of road pavements and curbs, with severe implications on the infrastructure’s safety. For this reason, the assessment and control of tree root systems’ development have become crucial task in forestry and urban management. Within this framework, Ground Penetrating Radar (GPR) is widely recognised as an effective non-destructive testing (NDT) method for the monitoring and assessment of road infrastructures. This study aims to demonstrate the capability of GPR in mapping the root system’s architecture of street trees. To this end, a GPR system equipped with a 700 MHz central frequency antenna was employed to investigate the area around a street tree, located in the vicinity of a flexible pavement structure. A novel processing framework based on the analysis of the signal reflectivity was implemented, in order to automatically identify the reflections from the pavement layers and apply dedicated advanced signal processing techniques. A multistage data processing methodology was employed to map the tree root system architecture. Finally, information on the mass density of roots at different depths was also provided. Results have proven the potential of the proposed methodology to achieve an automatic detection and mapping of roots under road pavements.

In this paper, we have presented a methodology combining ground penetrating radar (GPR) and a low-cost GPS receiver for three-dimensional detection of tree roots. This research aims to provide an effective and affordable testing tool to assess the root system of a number of trees. For this purpose, a low-cost GPS receiver was used, which recorded the approximate position of each GPR track, collected with a 500 MHz RAMAC shielded antenna. A dedicated post-processing methodology based on the precise position of the satellite data, satellite clock offsets data, and a local reference Global Navigation Satellite System (GNSS) Earth Observation Network System (GEONET) Station close to the survey site was developed. Firstly, the positioning information of local GEONET stations was used to filter out the errors caused by satellite position error, satellite clock offset, and ionosphere. In addition, the advanced Kalman filter was designed to minimise receiver offset and the multipath error, in order to obtain a high precision position of each GPR track. Kirchhoff migration considering near-field effect was used to identify the three-dimensional distribution of the root. In a later stage, a novel processing scheme was used to detect and clearly map the coarse roots of the investigated tree. A successful case study is proposed, which supports the following premise: the current scheme is an affordable and accurate mapping method of the root system architecture.

Non-destructive testing (NDT) for health monitoring of trees is a suitable candidate for detecting signs of early decay [1]. Recent developments [2,3,4] have highlighted that ground-penetrating radar (GPR) has the potential to provide with a robust and accurate detection tool with minimum computational and operational requirements in the field. In particular, a processing framework is suggested in [2] that can effectively remove ringing noise and unwanted clutter. Subsequently, an arc length parameterisation is employed in order to utilise a wheel-measurement device to accurately position the measured traces. Lastly, two migration schemes; Kirchhoff and reverse-time migration, are successfully applied on numerical and laboratory data in [3]. In the current paper, the detection scheme described in [2,3] using reverse-time migration is tested in two case studies that involve diseased urban trees within the greater London area, UK (Kensington and Gunnersbury park). Both of the trees were cut down after the completion of the measurements and furthermore cut into several slices to get direct information with regards to their internal structure. The processing scheme described in [3,4] managed to adequately detect the internal decay present in both trees. The aforementioned case studies provide coherent evidences to support the premise that GPR is capable of detecting decay in diseased trunks and therefore has the potential to become an accurate and efficient diagnostic tool against emerging infectious diseases of trees. Acknowledgements The authors would like to express their sincere thanks and gratitude to the following trusts, charities, organizations and individuals for their generosity in supporting this project: Lord Faringdon Charitable Trust, The Schroder Foundation, Cazenove Charitable Trust, Ernest Cook, Sir Henry Keswick, Ian Bond, P. F. Charitable Trust, Prospect Investment Manage- ment Limited, The Adrian Swire Charitable Trust, The John Swire 1989 Charitable Trust, The Sackler Trust, The Tanlaw Foundation and The Wyfold Charitable Trust. This paper is dedicated to the memory of Jonathon West, a friend, a colleague, a forester, a conservationist and an environmentalist who died following an accident in the woodland that he loved. References[1] P. Niemz, D. Mannesm, ”Non-destructive testing of wood and wood-based materials,” J. Cult. Heritage, vol. 13, pp. S26-S34, 2012.[2] I. Giannakis, F. Tosti, L. Lantini and A. M. Alani, "Health Monitoring of Tree Trunks Using Ground Penetrating Radar," IEEE Transactions on Geoscience and Remote Sensing, vol. 57, no. 10, pp. 8317-8326, 2019.[3] I. Giannakis, F. Tosti, L. Lantini and A. M. Alani, "Diagnosing Emerging Infectious Diseases of Trees Using Ground Penetrating Radar," IEEE Transactions on Geoscience and Remote Sensing, Early Access, doi: 10.1109/TGRS.2019.2944070 [4] A. M. Alani, F. Soldovieri, I. Catapano, I. Giannakos, G. Gennarelli, L. Lantini, G. Ludeno and F. Tosti, “The Use of Ground Penetrating Radar and Microwave Tomography for the Detection of Decay and Cavities in Tree Trunks,” Remote Sensing, vol. 11, no. 18, 2019.

Street trees are widely recognised to be an essential asset for the urban environment, as they bring several environmental, social and economic benefits [1]. However, the conflicting coexistence of tree root systems with the built environment, and especially with road infrastructures, is often cause of extensive damage, such as the uplifting and cracking of sidewalks and curbs, which could seriously compromise the safety of pedestrians, cyclists and drivers.In this context, Ground Penetrating Radar (GPR) has long been proven to be an effective non-destructive testing (NDT) method for the evaluation and monitoring of road pavements. The effectiveness of this tool lies not only in its ease of use and cost-effectiveness, but also in the proven reliability of the results provided. Besides, recent studies have explored the capability of GPR in detecting and mapping tree roots [2]. Algorithms for the reconstruction of the tree root systems have been developed, and the spatial variations of root mass density have been also investigated [3].The aim of this study is, therefore, to investigate the GPR potential in mapping the architecture of root systems in street trees. In particular, this research aims to improve upon the existing methods for detection of roots, focusing on the identification of the road pavement layers. In this way, different advanced signal processing techniques can be applied at specific sections, in order to remove reflections from the pavement layers without affecting root detection. This allows, therefore, to reduce false alarms when investigating trees with root systems developing underneath road pavements.In this regard, data from trees of different species have been acquired and processed, using different antenna systems and survey methodologies, in an effort to investigate the impact of these parameters on the GPR overall performance. AcknowledgementsThe authors would like to express their sincere thanks and gratitude to the following trusts, charities, organisations and individuals for their generosity in supporting this project: Lord Faringdon Charitable Trust, The Schroder Foundation, Cazenove Charitable Trust, Ernest Cook Trust, Sir Henry Keswick, Ian Bond, P. F. Charitable Trust, Prospect Investment Management Limited, The Adrian Swire Charitable Trust, The John Swire 1989 Charitable Trust, The Sackler Trust, The Tanlaw Foundation, and The Wyfold Charitable Trust. This paper is dedicated to the memory of our colleague and friend Jonathan West, one of the original supporters of this research project. References[1] J. Mullaney, T. Lucke, S. J. Trueman, 2015. “A review of benefits and challenges in growing street trees in paved urban environments,” Landscape and Urban Planning, 134, 157-166.[2] A. M. Alani, L. Lantini, 2019. “Recent advances in tree root mapping and assessment using non-destructive testing methods: a focus on ground penetrating radar,” Surveys in Geophysics, 1-42.[3] L. Lantini, F. Tosti, Giannakis, I., Egyir, D., A. Benedetto, A. M. Alani, 2019. “A Novel Processing Framework for Tree Root Mapping and Density Estimation using Ground Penetrating Radar,” In 10th International Workshop on Advanced Ground Penetrating Radar, EAGE.

In oil drilling, mud filtrate penetrates into porous formations and alters the compositions and properties of the pore fluids. This disturbs the logging signals and brings errors to reservoir evaluation. Drilling and logging engineers therefore deem mud invasion as undesired and attempt to eliminate its adverse effects. However, the mud-contaminated formation carries valuable information, notably with regard to its hydraulic properties. Typically, the invasion depth critically depends on the formation porosity and permeability. Therefore, if adequately characterized, mud invasion effects could be used for reservoir evaluation. To pursue this objective, we have applied borehole radar to measure mud invasion depth considering its high radial spatial resolution compared with conventional logging tools, which then allows us to estimate the reservoir permeability based on the acquired invasion depth. We investigate the feasibility of this strategy numerically through coupled electromagnetic and fluid modeling in an oil-bearing layer drilled using freshwater-based mud. Time-lapse logging is simulated to extract the signals reflected from the invasion front, and a dual-offset downhole antenna mode enables time-to-depth conversion to determine the invasion depth. Based on drilling, coring, and logging data, a quantitative interpretation chart is established, mapping the porosity, permeability, and initial water saturation into the invasion depth. The estimated permeability is in a good agreement with the actual formation permeability. Our results therefore suggest that borehole radar has significant potential to estimate permeability through mud invasion effects.

Modern socioeconomic factors such as global timber trade and international travelling have contributed to the rapid increase of Emerging Infectious Diseases (EIDs) of trees with devastating effects to the European forests and woodlands. To that extent, numerous non-destructive methodologies have been suggested as diagnostic tools in order to effectively monitor and maintain potential outbreaks. Ground-penetrating radar (GPR) is an appealing method for tree monitoring as it provides with a trivially deployable and efficient detection tool, suitable for large-scale forestry applications. Nonetheless, traditional GPR approaches are tuned for surface measurements and they are not compatible with the unique measurement configurations associated with forestry applications. Within that context, we present a novel-processing framework, which is capable of addressing features with irregular measurements on closed surfaces. A positioning method is described that exploits a wheel-measuring device in order to accurately associate each A-Scan with its corresponding coordinates. In addition, a processing pipeline is presented that aims at eliminating the ringing noise due to the layered nature of the trees. Lastly, a reverse-time migration is applied to the processed B-Scan in order to effectively map the reflectors present within the trunk. The suggested scheme is successfully tested in both numerical and real-field experiments, indicating the validity of the current approach.

The danger related to the structural stability of hollowed trees is a matter of wide discussion among the scientific community. Hollow cores in trees can extend to more than 50% of the total diameter [1] and, while the presence of a hollow tree might appear dramatic in terms of public safety, it is not always a cause of concern. It is known that hollow trees can form in many years or even decades [2] and, although the heartwood is effectively dead, the tree can continue to form sapwood on the exterior of the trunk to create a cylinder. However, robustness and structural support provided by this cylinder to the trunk and canopy above depend on the ratio of healthy to diseased tissue.In this context, Ground Penetrating Radar (GPR) has proven to be an effective non-invasive tool, capable of generating information about the inner structure of tree trunks in terms of existence, location, and geometry of defects [3], [4]. Nevertheless, it had been observed that the currently available and known GPR-related processing and data interpretation methods and tools are able to provide only limited information on the tree inner structure.In this study, we present a microwave tomographic approach for improved GPR data processing with the aim of detecting and characterising the geometry of hollowed trees. Tests were performed at Gunnesbury Park, London, UK. In particular, a number of 15 circular measurements were collected around the tree using the Aladdin 2 GHz hand-held antenna system manufactured by IDS GeoRadar (Part of Hexagon), covering a height of 140 cm. The tree was eventually felled and three sections were cut for validation purposes.Results presented in this abstract are part of a major research project that the authors have undertaken for the last three years. AcknowledgementsThe authors would like to express their sincere thanks and gratitude to the following trusts, charities, organisations and individuals for their generosity in supporting this project: Lord Faringdon Charitable Trust, The Schroder Foundation, Cazenove Charitable Trust, Ernest Cook Trust, Sir Henry Keswick, Ian Bond, P.F. Charitable Trust, Prospect Investment Management Limited, The Adrian Swire Charitable Trust, The John Swire 1989 Charitable Trust, The Sackler Trust, The Tanlaw Foundation, and The Wyfold Charitable Trust. This paper is dedicated to the memory of our colleague and friend Jonathan West, one of the original supporters of this research project. References[1] Braithwaite, R.W. (1985). The Kakadu fauna survey: an ecological survey of Kakudu National Park. Canberra, Australia: Australian Parks and Wildlife Service.[2] Ruxton, G.D. (2014). Why are so many trees hollow? Biology Letters, 10 (11).[3] Giannakis, I., Tosti, F., Lantini, L., Alani, A.M. (2019). Diagnosing Emerging Infectious Diseases of Trees Using Ground Penetrating Radar, IEEE Transactions on Geoscience and Remote Sensing. doi: 10.1109/TGRS.2019.2944070[4] Alani, A.M., Soldovieri, F., Catapano, I., Giannakis, I., Gennarelli, G., Lantini, L., Ludeno, G., Tosti, F. (2019). The Use of Ground Penetrating Radar and Microwave Tomography for the Detection of Decay and Cavities in Tree Trunks. Remote Sens., 11, 2073.

This study presents recent advances achieved in the use of ground penetrating radar (GPR) for the health monitoring of tree root systems. The main objectives of the research were to provide an effective and high-resolution mapping of the root systems belonging to different species of matured trees, as well as to investigate areas of roots interconnection. To this purpose, a data processing methodology based on three main stages was developed. A pre-processing algorithm was first proposed to remove noise-related information from the raw data and to enhance deep reflections from attenuated targets. Afterwards, an algorithm for identification of targets (i.e. the vertices of the reflection hyperbolas) and their automated tracking in a three-dimensional environment was developed. A third stage was focused on estimating tree root density with emphasis on the interconnection area. To test the feasibility of the proposed methodology, the soil around two different tree species (i.e. maple and ash trees) was investigated using a ground-coupled GPR system equipped with a 700 MHz central frequency antenna. The method has proven to identify distinctive features of both trees, in terms of shallow (i.e. within the first 25 cm from the soil surface) and deep (i.e. deeper than 25 cm from the soil surface) root systems. In addition, results have allowed to assess how different root systems interact with each other.

Aggressive fungal and insect attacks have reached an alarming level, threatening a variety of tree species, such as ash and oak trees, in the United Kingdom and beyond. In this context, Ground Penetrating Radar (GPR) has proven to be an effective non-invasive tool, capable of generating information about the inner structure of tree trunks in terms of existence, location, and geometry of defects. Nevertheless, it had been observed that the currently available and known GPR-related processing and data interpretation methods and tools are able to provide only limited information regarding the existence of defects and anomalies within the tree inner structure. In this study, we present a microwave tomographic approach for improved GPR data processing with the aim of detecting and characterising the geometry of decay and cavities in trees. The microwave tomographic approach is able to pinpoint explicitly the position of the measurement points on the tree surface and thus to consider the actual geometry of the sections beyond the classical (circular) ones. The robustness of the microwave tomographic approach with respect to noise and data uncertainty is tackled by exploiting a regularised scheme in the inversion process based on the Truncated Singular Value Decomposition (TSVD). A demonstration of the potential of the microwave tomography approach is provided for both simulated data and measurements collected in controlled conditions. First, the performance analysis was carried out by processing simulated data achieved by means of a Finite-Difference Time-Domain (FDTD) in three scenarios characterised by different geometric trunk shapes, internal trunk configurations and target dimensions. Finally, the method was validated on a real trunk by proving the viability of the proposed approach in identifying the position of cavities and decay in tree trunks.

Finite-Difference Time-Domain (FDTD) modelling of Ground Penetrating Radar (GPR) is becoming regularly used in model-based interpretation methods like full waveform inversion (FWI), and machine learning schemes using synthetic training data. Oversimplifications in such forward models can compromise the accuracy and realism with which real GPR responses can be simulated, and this degrades the overall performance of the aforementioned interpretation techniques. Therefore, a forward model must be able to accurately simulate every part of the GPR problem that can affect the resulting scattered field. A key element is the antenna system and excitation waveform, so the model must contain a complete description of the antenna including the excitation source and waveform, the geometry, and the dielectric properties of materials in the antenna. The challenge is that some of these parameters are not known or easily measured, especially for commercial GPR antennas that are used in practice. We present a novel hybrid linear/non-linear FWI approach which can be used, with only knowledge of the basic antenna geometry, to simultaneously optimise the dielectric properties and excitation waveform of the antenna, and minimise the error between real and synthetic data. The accuracy and stability of our proposed methodology is demonstrated by successfully modelling a Geophysical Survey Systems (GSSI) Inc. 1.5~GHz commercial antenna. Our framework allows accurate models of GPR antennas to be developed without requiring detailed knowledge of every component in the antenna. This is significant because it allows commercial GPR antennas, regularly used in GPR surveys, to be more readily simulated.

Ground penetrating radar (GPR) is traditionally applied to smooth surfaces in which the assumption of half-space is an adequate approximation that does not deviate much from reality. Nonetheless, using GPR for internal structure characterization of tree trunks requires measurements on an irregularly shaped closed curve. A typical hyperbola fitting has no physical meaning in this new context since the reflection patterns are strongly associated with the shape of the tree trunk. Instead of a clinical hyperbola, the reflections give rise to complex-shaped patterns that are difficult to be analyzed even in the absence of clutter. In this paper, a novel processing scheme is described which can interpret complex reflection patterns assuming a circular target subject to any arbitrary shaped surface. The proposed methodology can be applied using commercial hand-held antennas in real time, avoiding computationally costly tomographic approaches that require the usage of custom-made bespoke antenna arrays. The validity of the current approach is illustrated both with numerical and real experiments.

Summary In the phase of oil drilling, mud filtrate penetrates into porous formations and alters the pore fluid properties. This complicates well logging exploration, and inevitably gives rise to shift in reservoir estimation. Logging engineers deem mud invasion a harm and attempt to eliminate its impact on logging data exploration. However, from our point of view, the mud-contaminated parts of the formation do also carry some valuable information, notably with regard to the key hydraulic properties. Therefore, if adequately characterized, mud invasion effects, in turn, could be utilized for reservoir estimation. Typically, the invasion depth critically depends on the formation porosity and permeability. To achieve this objective, we propose to use borehole radar to determine the mud invasion depth considering a high spatial resolution of ground-penetrating radar (GPR) compared with the conventional logging tools. We implement numerical investigations on the feasibility of this approach by coupling electromagnetic (EM) modelling with fluid flow modelling in an oil-bearing formation disturbed by mud invasion effects. The simulations imply that a time-lapse radar logging is able to extract EM reflection signals from mud invasion front, and the invasion depth and EM velocity can be obtained by a downhole antenna displacement of one source and two receivers. We find that there exists a positive correlation between the estimated invasion depth and permeability curves, and a negative correlation between the estimated velocity and porosity curves. We suggest that borehole radar has potential to estimate permeability and porosity of oil reservoirs, wherein the mud invasion effect is positively utilized. The study demonstrates a potential method of oil reservoir estimation and a novel application of GPR in oil fields.

Estimating the permittivity of heterogeneous mixtures based on the permittivity of their components is of high importance with many applications in ground penetrating radar (GPR) and in electrodynamics-based sensing in general. Complex Refractive Index Model (CRIM) is the most mainstream approach for estimating the bulk permittivity of heterogeneous materials and has been widely applied for GPR applications. The popularity of CRIM is primarily based on its simplicity while its accuracy has never been rigorously tested. In the current study, an optimised shape factor is derived that is fine-tuned for modelling the dielectric properties of concrete. The bulk permittivity of concrete is expressed with respect to its components i.e., aggregate particles, cement particles, air-voids and volumetric water fraction. Different combinations of the above materials are accurately modelled using the Finite-Difference Time-Domain (FDTD) method. The numerically estimated bulk permittivity is then used to fine-tune the shape factor of the CRIM model. Then, using laboratory measurements it is shown that the revised CRIM model over-performs the default shape factor and provides with more accurate estimations of the bulk permittivity of concrete.

A regression model is developed in order to estimate in real time the signal-to-clutter ratio (SCR) for landmine detection using ground-penetrating radar. Artificial neural networks are employed in order to express SCR with respect to the soil's properties, the depth of the target, and the central frequency of the pulse. The SCR is synthetically evaluated for a wide range of diverse and controlled scenarios using the finite-difference time-domain method. Fractals are used to describe the geometry of the soil's heterogeneities as well as the roughness of the surface. The dispersive dielectric properties of the soil are expressed with respect to traditionally used soil parameters, namely, sand fraction, clay fraction, water fraction, bulk density, and particle density. Through this approach, a coherent and uniformly distributed training set is created. The overall performance of the resulting nonlinear function is evaluated using scenarios which are not included in the training process. The calculated and the predicted SCR are in good agreement, indicating the validity and the generalization capabilities of the suggested framework.

A three-dimensional (3-D) finite-difference time-domain (FDTD) algorithm is used in order to simulate ground penetrating radar (GPR) for landmine detection. Two bowtie GPR transducers are chosen for the simulations and two widely employed antipersonnel (AP) landmines, namely PMA-1 and PMN are used. The validity of the modeled antennas and landmines is tested through a comparison between numerical and laboratory measurements. The modeled AP landmines are buried in a realistically simulated soil. The geometrical characteristics of soil’s inhomogeneity are modeled using fractal correlated noise, which gives rise to Gaussian semivariograms often encountered in the field. Fractals are also employed in order to simulate the roughness of the soil’s surface. A frequency-dependent complex electrical permittivity model is used for the dielectric properties of the soil, which relates both the velocity and the attenuation of the electromagnetic waves with the soil’s bulk density, sand particles density, clay fraction, sand fraction, and volumetric water fraction. Debye functions are employed to simulate this complex electrical permittivity. Background features like vegetation and water puddles are also included in the models and it is shown that they can affect the performance of GPR at frequencies used for landmine detection (0.5–3 GHz). It is envisaged that this modeling framework would be useful as a testbed for developing novel GPR signal processing and interpretations procedures and some preliminary results from using it in such a way are presented.

The ability to produce, store and analyse large amounts of well-labeled data as well as recent advancements on supervised training, led machine learning to gain a renewed popularity. In the present paper, the applicability of machine learning to simulate ground penetrating radar (GPR) for high frequency applications is examined. A well-labelled and equally distributed training set is generated synthetically using the finite-difference time-domain (FDTD) method. Special care was taken in order to model the antennas and the soils with sufficient accuracy. Through a stochastic parameterisation, each model is expressed using only seven parameters (i.e. the fractal dimension of water fraction, the height of the antenna and so on). Based on these parameters and the synthetically generated training set, a machine learning framework is trained to predict the resulting A-Scan in real-time. Thus, overcoming the time-consuming calculations required for an equivalent FDTD simulation.

Forward modelling of Ground Penetrating Radar (GPR) is often used to facilitate interpretation of complex GPR data and as a key ingredient of full waveform inversion (FWI) processes. As general 3D full-wave electromagnetic solvers are computationally very demanding routine application of advanced GPR modelling is not popular. A novel concept for creating a fast GPR forward model based on machine learning (ML) concepts is presented. This ML-based model is trained using a dataset obtained from a realistic 3D Finite-Difference Time-Domain (FDTD) gprMax model. The fast model is trained for a specific GPR application that can be easily parametrised and have a somewhat constrained variability. However, the training uses GPR A-Scans obtained from very realistic forward models that include all complex scattering effects and antenna coupling mechanisms. To demonstrate the efficiency of the approach an application, using real GPR data, of the fast forward solver within a FWI process, using a global optimiser requiring a great number of forward model calculations, is presented producing very promising results.