Full-waveform Ground-penetrating Radar Research Articles

Estimation of soil hydraulic properties based on full waveform inversion of ground-penetrating radar data with Grey Wolf optimization algorithm

Abstract In this paper, we propose an inversion scheme to directly and non-invasively determine soil hydraulic properties. The scheme is based on hydrogeophysical inversion full waveform ground penetrating radar (GPR) data with Grey Wolf Optimizer algorithm. This inversion scheme includes a soil hydraulic model, a petrophysical relationship, a ground penetrating radar forward modelling and a Grey Wolf Optimizer algorithm. In order to evaluate its performance, a numerical experiment was carried out and the proposed scheme was applied to the experimental data. The results indicate that the proposed inversion scheme is promising for determining soil hydraulic properties with GPR.

Thickness regression for backfill grouting of shield tunnels based on GPR data and CatBoost & BO-TPE: A full-scale model test study

Ground penetrating radar (GPR) is a vital non-destructive testing (NDT) technology that can be employed for detecting the backfill grouting of shield tunnels. To achieve intelligent analysis of GPR data and overcome the subjectivity of traditional data processing methods, the CatBoost & BO-TPE model was constructed for regressing the grouting thickness based on GPR waveforms. A full-scale model test and corresponding numerical simulations were carried out to collect GPR data at 400 and 900 MHz, with known backfill grouting thickness. The model test helps address the limitation of not knowing the grout body condition in actual field detection. The data were then used to create machine learning datasets. The method of feature selection was proposed based on the analysis of feature importance and the electromagnetic (EM) propagation law in mediums. The research shows that: (1) the CatBoost & BO-TPE model exhibited outstanding performance in both experimental and numerical data, achieving R2 values of 0.9760, 0.8971, 0.8808, and 0.5437 for numerical data and test data at 400 and 900 MHz. It outperformed extreme gradient boosting (XGBoost) and random forest (RF) in terms of performance in the backfill grouting thickness regression; (2) compared with the full-waveform GPR data, the feature selection method proposed in this paper can promote the performance of the model. The selected features within the 5–30 ns of the A-scan can yield the best performance for the model; (3) compared to GPR data at 900 MHz, GPR data at 400 MHz exhibited better performance in the CatBoost & BO-TPE model. This indicates that the results of the machine learning model can provide feedback for the selection of GPR parameters; (4) the application results of the trained CatBoost & BO-TPE model in engineering are in line with the patterns observed through traditional processing methods, yet they demonstrate a more quantitative and objective nature compared to the traditional method.

Joint full-waveform ground-penetrating radar and electrical resistivity inversion applied to field data acquired on the surface

We exploit the different but complementary data sensitivities of ground-penetrating radar (GPR) and electrical resistivity (ER) by applying a multiphysics, multiparameter, simultaneous 2.5D joint inversion without invoking petrophysical relationships. Our method joins full-waveform inversion (FWI) GPR with adjoint derived ER sensitivities on the same computational domain. We incorporate a stable source estimation routine into the FWI-GPR. We apply our method in a controlled alluvial aquifer using only surface-acquired data. The site exhibits a shallow groundwater boundary and unconsolidated heterogeneous alluvial deposits. We compare our recovered parameters to individual FWI-GPR and ER results, and we compare them to log measurements of capacitive conductivity and neutron-derived porosity. Our joint inversion provides a more representative depiction of subsurface structures because it incorporates multiple intrinsic parameters, and it is therefore superior to an interpretation based on log data, FWI-GPR, or ER alone.

Joint inversion of full-waveform ground-penetrating radar and electrical resistivity data: Part 1

We have developed an algorithm for joint inversion of full-waveform ground-penetrating radar (GPR) and electrical resistivity (ER) data. The GPR data are sensitive to electrical permittivity through reflectivity and velocity, and electrical conductivity through reflectivity and attenuation. The ER data are directly sensitive to the electrical conductivity. The two types of data are inherently linked through Maxwell’s equations, and we jointly invert them. Our results show that the two types of data work cooperatively to effectively regularize each other while honoring the physics of the geophysical methods. We first compute sensitivity updates separately for the GPR and ER data using the adjoint method, and then we sum these updates to account for both types of sensitivities. The sensitivities are added with the paradigm of letting both data types always contribute to our inversion in proportion to how well their respective objective functions are being resolved in each iteration. Our algorithm makes no assumptions of the subsurface geometry nor the structural similarities between the parameters with the caveat of needing a good initial model. We find that our joint inversion outperforms the GPR and ER separate inversions, and we determine that GPR effectively supports ER in regions of low conductivity, whereas ER supports GPR in regions with strong attenuation.

Joint inversion of full-waveform ground-penetrating radar and electrical resistivity data — Part 2: Enhancing low frequencies with the envelope transform and cross gradients

Recovering material properties of the subsurface using ground-penetrating radar (GPR) data of finite bandwidth with missing low frequencies and in the presence of strong attenuation is a challenging problem. We have adopted three nonlinear inverse methods for recovering electrical conductivity and permittivity of the subsurface by joining GPR multioffset and electrical resistivity (ER) data acquired at the surface. All of the methods use ER data to constrain the low spatial frequency of the conductivity solution. The first method uses the envelope of the GPR data to exploit low-frequency content in full-waveform inversion and does not assume structural similarities of the material properties. The second method uses cross gradients to manage weak amplitudes in the GPR data by assuming structural similarities between permittivity and conductivity. The third method uses the envelope of the GPR data and the cross gradient of the model parameters. By joining ER and GPR data, exploiting low-frequency content in the GPR data, and assuming structural similarities between the electrical permittivity and conductivity, we are able to recover subsurface parameters in regions where the GPR data have a signal-to-noise ratio close to one.

Underground structure defect detection and reconstruction using crosshole GPR and Bayesian waveform inversion

Abstract Crosshole ground-penetrating radar (GPR) is a widely used measurement technique to help inspect the structural integrity of man-made underground structures, yet the resulting waveform and travel-time data can be difficult, complex and challenging to interpret. Here, we introduce the elements of a Bayesian inversion method for analyzing crosshole GPR data to guide detection of defects (weakness zones) in underground concrete structures. This framework uses as main building blocks the two-dimensional finite-difference time-domain (FDTD) simulator, the discrete cosine transform (DCT) method, and the DREAM (ZS) algorithm to reconstruct the relative permittivity field of an underground concrete structure from full-waveform GPR inversion. The FDTD simulator solves numerically Maxwell's equations in the time and space domain of the crosshole GPR experiment and simulates iteratively the electromagnetic (EM) waveforms. The DCT algorithm transforms the Cartesian parameterization to the frequency domain and reduces drastically the dimensionality of the parameter space by retaining only the low-frequency DCT-coefficients. Markov chain Monte Carlo (MCMC) simulation with the DREAM (ZS) algorithm is used to estimate the posterior distribution of the DCT-coefficients. The usefulness and applicability of the FDTD-DCT-DREAM (ZS) framework is demonstrated on a synthetic test example involving a unit square concrete structure with a small defect. Our results demonstrate that the proposed method successfully detects and locates defects in concrete structures. The inversion results appear rather insensitive to the noise level of the measured GPR waveforms, and the amount of data used (number of receiving antenna positions), as long as a sufficient number of measurements is available. The more DCT-coefficients that are used to characterize the concrete structure, the more accurate the results, yet the larger the posterior uncertainty of the reconstructed permittivity fields.

Quantitative multi-layer electromagnetic induction inversion and full-waveform inversion of crosshole ground penetrating radar data

Due to the recent system developments for the electromagnetic characterization of the subsurface, fast and easy acquisition is made feasible due to the fast measurement speed, easy coupling with GPS systems, and the availability of multi-channel electromagnetic induction (EMI) and ground penetrating radar (GPR) systems. Moreover, the increasing computer power enables the use of accurate forward modeling programs in advanced inversion algorithms where no approximations are used and the full information content of the measured data can be exploited. Here, recent developments of large-scale quantitative EMI inversion and full-waveform GPR inversion are discussed that yield higher resolution of quantitative medium properties compared to conventional approaches. In both cases a detailed forward model is used in the inversion procedure that is based on Maxwell’s equations. The multi-channel EMI data that have different sensing depths for the different source-receiver offset are calibrated using a short electrical resistivity tomography (ERT) calibration line which makes it possible to invert for electrical conductivity changes with depth over large areas. The crosshole GPR full-waveform inversion yields significant higher resolution of the permittivity and conductivity images compared to ray-based inversion results.

Estimation of the near surface soil water content during evaporation using air‐launched ground‐penetrating radar

ABSTRACTEvaporation is an important process in the global water cycle and its variation affects the near surface soil water content, which is crucial for surface hydrology and climate modelling. Soil evaporation rate is often characterized by two distinct phases, namely, the energy limited phase (stage‐I) and the soil hydraulic limited period (stage‐II). In this paper, a laboratory experiment was conducted using a sand box filled with fine sand, which was subject to evaporation for a period of twenty three days. The setup was equipped with a weighting system to record automatically the weight of the sand box with a constant time‐step. Furthermore, time‐lapse air‐launched ground penetrating radar (GPR) measurements were performed to monitor the evaporation process. The GPR model involves a full‐waveform frequency‐domain solution of Maxwell’s equations for wave propagation in three‐dimensional multilayered media. The accuracy of the full‐waveform GPR forward modelling with respect to three different petrophysical models was investigated. Moreover, full‐waveform inversion of the GPR data was used to estimate the quantitative information, such as near surface soil water content. The two stages of evaporation can be clearly observed in the radargram, which indicates qualitatively that enough information is contained in the GPR data. The full‐waveform GPR inversion allows for accurate estimation of the near surface soil water content during extended evaporation phases, when a wide frequency range of GPR (0.8–5.0 GHz) is taken into account. In addition, the results indicate that the CRIM model may constitute a relevant alternative in solving the frequency‐dependency issue for full waveform GPR modelling.

Full-waveform GPR inversion to assess chloride gradients in concrete

The spreading of deicing salt on roads during winter and the consequent migration of chloride ions within concrete exposes structures reinforcement to corrosion. Reinforcement corrosion is problematic because it reduces the structure's load carrying capacity and a long-term consequence can be structural failure. The use of Ground Penetrating Radar (GPR) technology is commonly applied to reinforced concrete structural mapping. This paper provides methodologies for the calculation of the electromagnetic parameters from GPR data accounting for chloride content and its distribution within concrete. The previous work showed that they are linked with water and chloride content. A GPR dataset was carried out with bi-static off-ground GPR antennas during an experiment recreating chloride contamination processes in concrete specimens. Two novel solutions are presented in this paper, their use on the dataset returned the electromagnetic parameters of the concrete specimens with knowledge of their thickness. Firstly, a ray-based approach is provided as a fast computational tool to determine average material properties. Thereafter, a novel full-waveform inversion based on a multilayered forward model is presented. It allows the identification of conductivity gradients present inside the concrete. Finally, the results are compared with additional destructive testing; they present a good correspondence.

Accounting for soil surface roughness in the inversion of ultrawideband off-ground GPR signal for soil moisture retrieval

We combined a full-waveform ground-penetrating radar (GPR) model with a roughness model to retrieve surface soil moisture through signal inversion. The proposed approach was validated under laboratory conditions with measurements performed above a sand layer subjected to seven different water contents and four different surface roughness conditions. The radar measurements were performed in the frequency domain in the range of 1–3 GHz and the roughness amplitude standard deviation was varied from 0 to 1 cm. Two inversion strategies were investigated: (1) Full-waveform inversion using the correct model configuration, and (2) inversion focused on the surface reflection only. The roughness model provided a good description of the frequency-dependent roughness effect. For the full-waveform analysis, accounting for roughness permitted us to simultaneously retrieve water content and roughness amplitude. However, in this approach, information on soil layering was assumed to be known. For the surface reflection analysis, which is applicable under field conditions, accounting for roughness only enabled water content to be reconstructed, but with a root mean square error (RMS) in terms of water content of [Formula: see text] compared to an RMS of [Formula: see text] for an analysis where roughness is neglected. However, this inversion strategy required a priori information on soil surface roughness, estimated, e.g., from laser profiler measurements.

Taming the non-linearity problem in GPR full-waveform inversion for high contrast media

We present a new algorithm for the inversion of full-waveform ground-penetrating radar (GPR) data. It is designed to tame the non-linearity issue that afflicts inverse scattering problems, especially in high contrast media. We first investigate the limitations of current full-waveform time-domain inversion schemes for GPR data and then introduce a much-improved approach based on a combined frequency-time-domain analysis. We show by means of several synthetic tests and theoretical considerations that local minima trapping (common in full bandwidth time-domain inversion) can be avoided by starting the inversion with only the low frequency content of the data. Resolution associated with the high frequencies can then be achieved by progressively expanding to wider bandwidths as the iterations proceed. Although based on a frequency analysis of the data, the new method is entirely implemented by means of a time-domain forward solver, thus combining the benefits of both frequency-domain (low frequency inversion conveys stability and avoids convergence to a local minimum; whereas high frequency inversion conveys resolution) and time-domain methods (simplicity of interpretation and recognition of events; ready availability of FDTD simulation tools).

A New Vector Waveform Inversion Algorithm for Simultaneous Updating of Conductivity and Permittivity Parameters From Combination Crosshole/Borehole-to-Surface GPR Data

We have developed a new full-waveform groundpenetrating radar (GPR) multicomponent inversion scheme for imaging the shallow subsurface using arbitrary recording configurations. It yields significantly higher resolution images than conventional tomographic techniques based on first-arrival times and pulse amplitudes. The inversion is formulated as a nonlinear least squares problem in which the misfit between observed and modeled data is minimized. The full-waveform modeling is implemented by means of a finite-difference time-domain solution of Maxwell's equations. We derive here an iterative gradient method in which the steepest descent direction, used to update iteratively the permittivity and conductivity distributions in an optimal way, is found by cross-correlating the forward vector wavefield and the backward-propagated vectorial residual wavefield. The formulation of the solution is given in a very general, albeit compact and elegant, fashion. Each iteration step of our inversion scheme requires several calculations of propagating wavefields. Novel features of the scheme compared to previous full-waveform GPR inversions are as follows: 1) The permittivity and conductivity distributions are updated simultaneously (rather than consecutively) at each iterative step using improved gradient and step length formulations; 2) the scheme is able to exploit the full vector wavefield; and 3) various data sets/survey types (e.g., crosshole and borehole-to-surface) can be individually or jointly inverted. Several synthetic examples involving both homogeneous and layered stochastic background models with embedded anomalous inclusions demonstrate the superiority of the new scheme over previous approaches.

Soil Surface Water Content Estimation by Full-Waveform GPR Signal Inversion in the Presence of Thin Layers

We analyzed the effect of shallow thin layers on the estimation of soil surface water content using full-waveform inversion of off-ground ground penetrating radar (GPR) data. Strong dielectric contrasts are expected to occur under fast wetting or drying weather conditions, thereby leading to constructive and destructive interferences with respect to surface reflection. First, synthetic GPR data were generated and subsequently inverted considering different thin-layer model configurations. The resulting inversion errors when neglecting the thin layer were quantified, and then, the possibility to reconstruct these layers was investigated. Second, laboratory experiments reproducing some of the numerical experiment configurations were conducted to assess the stability of the inverse solution with respect to actual measurement and modeling errors. Results showed that neglecting shallow thin layers may lead to significant errors on the estimation of soil surface water content(??>0.03 m3/m3), depending on the contrast. Accounting for these layers in the inversion process strongly improved the results, although some optimization issues were encountered. In the laboratory, the proposed full-waveform method permitted to reconstruct thin layers with a high resolution up to 2 cm and to retrieve the soil surface water content with an rmse less than 0.02 m3/m3, owing to the full-waveform inverse modeling. These results suggest that the proposed GPR approach is promising for field-scale mapping of soil surface water content of nondispersive soils with low electrical conductivity and for instances when soil layering is encountered.