Deep learning (DL) has gained traction in ground-penetrating radar (GPR) tasks. However, obtaining sufficient training data presents a significant challenge. We introduce a structure-adaptive GPR-generative adversarial network (GAN) to generate GPR defect data. GPR-GAN employs double normalization for stabilizing parameters and convolution outputs, an adaptive discriminator augmentation (ADA) module for small dataset training stability, and a modified self-attention (MSA) module to generate GPR defects with complex features. We evaluated the performance of GPR-GAN using three datasets in conjunction with three state-of-the-art detection networks (faster region-based convolutional neural network (FasterRCNN), single-shot multibox detector (SSD), and YOLOv5). Our results reveal that GPR-GAN exhibits strong generalization skills, adeptly adapting to GPR data generation tasks that encompasses a variety of targets, frequencies, and equipment. GPR-GAN generated data increased the $F1$ score for void recognition in simulation data by at least 5.27%, improved the average $F1$ score for highway pavement defect detection by at least 7.68%, and enhanced the average $F1$ score for railway subgrade defect detection by at least 9.22%. GPR-GAN offers a powerful data support tool for DL research in GPR.

Ground-penetrating radar (GPR) is an efficient and nondestructive geophysical method with great potential for detecting soil water content at the farmland scale. However, a key challenge in soil detection is obtaining soil water content rapidly and in real-time. In recent years, deep learning methods have become more widespread in the earth sciences, making it possible to use them for soil water content inversion from GPR data. In this paper, we propose a neural network framework GPRSW based on deep learning of GPR data. GPRSW is an end-to-end network that directly inverts volumetric soil water content (VSWC) through single-channel GPR data. Synthetic experiments show that GPRSW accurately identifies different VSWC boundaries in the model in time depth. The predicted VSWC and model fit well within 40 ns, with a maximum error after 40 ns of less than 0.10 cm³ × cm⁻³. To validate our method, we conducted GPR measurements at the experimental field of the Academy of Agricultural Sciences in Gongzhuling City, Jilin Province and applied GPRSW to VSWC measurements. The results show that predicted values of GPRSW match with field soil samples and are consistent with the overall trend of the TDR soil probe samples, with a maximum difference not exceeding 0.03 cm³ × cm⁻³. Therefore, our study shows that GPRSW has the potential to be applied to obtain soil water content from GPR data on farmland.

Ground-penetrating radar (GPR) is an important tool for measuring soil water content (SWC) at the field scale. The amplitude analysis of the early time signal (ETS) of GPR may provide a rapid way to estimate SWC. By assuming a homogeneous medium, various studies have been conducted on the relationship between the amplitude of ETS and the topsoil layer’s electromagnetic parameters (dielectric permittivity and conductivity) through numerical simulations, laboratory experiments, and field experiments. Soil is a typical inhomogeneous medium, and soil cultivation is a factor affecting its heterogeneity. In this context, we discuss the ability of the amplitude of ETS to estimate the water content of heterogeneous soil. First, we establish a multi-scale stochastic medium model with the inhomogeneous distribution of dielectric permittivity and conductivity and simulate the GPR response by the finite-difference time-domain (FDTD) method to observe the influence of medium heterogeneity on the GPR response. The heterogeneity of the soil models is evaluated by a geostatistical analysis described by two parameters, correlation length and variability. Then, we analyze the relationship between variability and the average envelope amplitude (AEA) of ETS. A strong soil heterogeneity increases the error of the AEA method in estimating SWC. Finally, the AEA method is used to estimate the SWC of two adjacent fields with different heterogeneities, which were caused by different cultivation methods. The results of the numerical simulation and field experiment indicate that the soil heterogeneity can have an impact on the estimation of SWC using EST, with an error lower than 3% within a depth range of 1/2 λ to λ (wavelength). This suggests that the EST of GPR can be applied to soil layers with relatively large lateral changes in water content.

Geologic interpretation results from conventional ground-penetrating radar (GPR) detection methods tend to have a certain degree of uncertainty. In order to improve the reliability of ground-penetrating radars in the detection of rock mass fissures in grottoes, this study proposes a ground-penetrating radar detection method with the logging data constraint, which is applied to detect the fractured rock mass in the Baodingshan Scenic Area of Dazu Rock Carvings, Chongqing, China. First, conventional logging and borehole televiewer data were compared and verified, yielding detailed lithological and wellbore fissure information. Next, electromagnetic wave velocity was calibrated using GPR profile and the depth of the stratigraphic interface determined by borehole data. Utilizing this calibrated velocity, we are able to accurately calculate the depth values of anomalies in GPR interpretation profiles. Subsequently, we compared the preliminary GPR interpretation profile with the borehole televiewer images. After eliminating false anomalies caused by interference, we obtained more reliable location information for detection targets such as fissures, fracture zones, and weak interlayers. The results of fissure detection in the Dazu Rock Carvings indicate that the detection results of ground-penetrating radar are verified and supplemented under the constraints of stratigraphic and well-wall fissure information obtained by logging. This effectively mitigates the influence of multiplicity and false anomalies of GPR detection on interpretation results. GPR with the logging data constraint enhances the accuracy of the fissure detection results, providing novel technical means for the protection and restoration of grotto relics.

With continuous improvements in oil production, the environmental problems caused by oil exploitation are becoming increasingly serious. Rapid and accurate estimation of soil petroleum hydrocarbon content is of great significance to the investigation and restoration of environments in oil-producing areas. In this study, the content of petroleum hydrocarbon and the hyperspectral data of soil samples collected from an oil-producing area were measured. For the hyperspectral data, spectral transforms, including continuum removal (CR), first- and second-order differential (CR-FD, CR-SD), and Napierian logarithm (CR-LN), were applied to eliminate background noise. At present, there are some shortcomings in the method of feature band selection, such as large quantity, time of calculation, and unclear importance of each feature band obtained. Meanwhile, redundant bands easily exist in the feature set, which seriously affects the accuracy of the inversion algorithm. In order to solve the above problems, a new method (GARF) for hyperspectral characteristic band selection was proposed. It combined the advantage that the grouping search algorithm can effectively reduce the calculation time with the advantage that the point-by-point search algorithm can determine the importance of each band, which provided a clearer direction for further spectroscopic research. The 17 selected bands were used as the input data of partial least squares regression (PLSR) and K-nearest neighbor (KNN) algorithms to estimate soil petroleum hydrocarbon content, and the leave-one-out method was used for cross-validation. The root mean squared error (RMSE) and coefficient of determination (R²) of the estimation result were 3.52 and 0.90, which implemented a high accuracy with only 8.37% of the entire bands. The results showed that compared with the traditional characteristic band selection methods, GARF can effectively reduce the redundant bands and screen out the optimal characteristic bands in the hyperspectral data of soil petroleum hydrocarbon with the method of importance assessment, which retained the physical meaning. It provided a new idea for the research of other substances in soil.

In railway tunnel inspections using train-mounted ground penetrating radar (GPR), catenary support devices on the tunnel lining often introduce significant clutter into GPR profiles. This external clutter distorts GPR data and complicates the interpretation of internal defects. Due to its strong reflections, extensive coverage, and similarity to target features, catenary clutter lacks effective suppression methods that simultaneously preserve weak signals of interest. In this work, we propose a catenary clutter elimination generative adversarial network (2C-GAN), which conceptualizes clutter suppression and profile reconstruction as a transformation problem from the cluttered domain to the clutter-free domain. The 2C-GAN accepts unpaired clutter-free and clutter-laden GPR data, iteratively training a generator to remove clutter and reconstruct GPR profiles. Its main advantage lies in its unsupervised approach with unpaired data, enhancing its practical applicability in engineering contexts. To address the challenge of limited training data, 2C-GAN employs a lightweight design with residual Cross Stage Partial modules, improving network performance and reducing generator parameters. Synthetic data testing shows that 2C-GAN significantly outperforms other unsupervised methods in catenary clutter suppression and related metrics. Furthermore, applying 2C-GAN to a real high-speed railway tunnel dataset with only 192 unpaired samples produced remarkable results, demonstrating the effectiveness of 2C-GAN in suppressing catenary clutter and its potential for application in similar above-surface diffraction scenarios.