Journal of Engineering Geology

---

Ground-Penetrating Radar (GPR) is a non-destructive and high-resolution geophysical method which uses high-frequency electromagnetic (EM) wave reflection off buried objects to detect them. In current research this method has been used to identify geometrical parameters of buried cylindrical targets such as tunnel structures. To achieve this aim, relationships between the geometrical parameters of cylindrical targets with the parameters of GPR hyperbolic response have been determined using two intelligent pattern recognition methods known as artificial neural network and template matching. To this goal GPR responses of synthetic cylindrical objects produced by 2D finite-difference method have been used as templates in the neural network and template matching algorithms. The structure of applied neural network has been designed based on extracting discriminant and unique features (eigenvalues and the norm of eigenvalues) from the GPR images and predicting all geometrical parameters of the targets, simultaneously. Also the template matching operation carried out using two diverse similarity approaches, spatial domain convolution and normalized cross correlation in 2D wave number domain. The results of the research show that both two employed intelligent methods can be applied for in situ, real-time, accurate and automatic interpretation of real GPR radargrams, however in general the neural network method has led to less error and better estimation than template matching to predict the geometrical parameters of the cylindrical tar

---

Introduction
     Ground-penetrating radar (GPR) method is a pretty new, non-destructive and high-resolution geophysical method that is widely used to identify the thickness of snow and ice layers and glaciers bed, because snow and ice are transparent for electromagnetic (EM) waves. Therefore, this method has been used to determine the thickness and basement topography of Alam-kooh glacier. In this research, only the GPR acquired data using unshielded antenna with central frequency of 25 MHz along one line in Alam-kuh glacier, Kelardasht- Mazandaran, have been processed and interpreted. The GPR data acquisition has been done by using common offset method, and transmitter-receiver separation of 6 meters. The final real radargram related to one of the surveyed GPR profiles in this region has been prepared after applying various processing operations containing signal saturation correction, amplitude gain, f-k migration filtering and static (topography) correction on the raw data. After applying processing sequences on the acquired data, the EM waves reflection off the interfaces of different layers including the reflections from the glacier basement have been detected, and by assigning a suitable EM wave velocity in the ice (0.16 m/ns), the thickness of 50 m for the ice layer laid under the survey line has been estimated. Also, in present research, forward and inverse modeling of GPR data have been performed to employ for snow, ice and glaciological investigations in the AlamKooh region of Mazandaran. To achieve this goal, GPR response of synthetic model corresponding to the real radargram was simulated first, by 2-D finite-difference time domain (FDTD) method. Afterward the inversion method by solving an optimization problem was employed to validate the interpretation of real GPR data.
Methodology and Approaches
     Based on the nature, physical and geometric properties of the subsurface target in the field data, their synthetic model have been built and their two-dimensional GPR responses forward modeling using ReflexW software and finite difference algorithms improved in the frequency domain, have been obtained. Also, it has been used an effective algorithm, coded in GUI environment of MATLAB programming software and as a result, a reliable and accurate inverse modeling has been carried out. In the present study, to simulate the behavior of the propagation of EM waves in GPR method, two-dimensional finite difference method has been used. The main advantage of this method is its comparative simplicity of the concept, high accuracy and simple implementation for complex and arbitrary models as well as easily adjusting the antenna when applied. In this study, acquisition of GPR field data and synthetic data modeling have been carried out in TM mode. The radargrams of the GPR data have been demonstrated using ReflexW software after performing necessary processing sequences.
Results and Conclusions
     The obtained results reveal that moraine materials covering the surface of the area are mainly fine-grained granite. The bed-rock or basement in the area is also granite. The polarity representing ice-bed rock is clearly seen on the GPR profiles. The topography of the glacier basement has successfully been detectable using just by GPR method. The electrical resistive nature of the glacier has caused large penetration depth of GPR pulses in this research. Furthermore, the results of the research for presented profiles on the basis of forward and inverse modeling output of GPR data in comparison with real GPR radargrams in the region validated the accuracy of GPR investigations in the area. Although with a quick glance, the error obtained by the inverse modeling for real GPR data seems unexpected and unacceptable, absolutely the high rate of error depends on many factors influencing on the real earth models containing various limitations existing in all forward modeling algorithms and software packages, impossibility of making forward modeling exactly according to the real models (due to the complex nature of the ground), taking into account the homogeneity and uniform host environment and targets in the modeling process unlike the diversity, the presence of different types of noises and so on. Therefore, making a controlled geophysical test site and trying performance of inverse modeling algorithm for field GPR data in this site, as well as determining the important physical parameters such as dielectric permittivity and electrical conductivity by experimental method through sampling from different depths for complex geological environments are suggested../files/site1/files/124/1ahmadpur%DA%86%DA%A9%DB%8C%D8%AF%D9%87.pdf

---

Introduction
Ground-penetrating radar (GPR) is a high-resolution geophysical method which uses electromagnetic waves with high-frequency in order to map structures and objects buried in subsurface without any destruction of the medium. In present research, choice of optimum parameters of real data acquisition for this method has been studied. The governed behavior on the GPR fields can be simulated by solving the Maxwell’s equations and the appropriate boundary conditions that form the basis of electromagnetic theory. Among the variety of available numerical methods, the finite-difference time-domain (FDTD) method has paid more attention due to having the simple understanding of the concepts, flexibility, simulation and modeling of complex environments and the acceptability of its responses in the applied cases. The purpose of this study is to identify what reasonable information can be obtained from field data under different environmental conditions and different survey parameters.
 
Materials and methods
To achieve the goal, first forward modeling of GPR data has been carried out for several synthetic models corresponding to common targets in subsurface installations, using 2-D finite-difference time-domain method by means of GPRMAX, ReflexW and Radexplorer softwares. The main purpose of the simulations is investigation of the effect of survey parameters such as spatial sampling intervals (trace interspacing) and temporal sampling frequency on the GPR response of targets with various physical and geometrical parameters. Also to select and design the most appropriate conditions and survey parameters for real GPR data, numerous field traverses were performed in Isfahan University of Technology campus over the pre-known buried cylindrical targets containing power cable, petro-gas pipe, water pipeline and waste water pipeline with diverse host media. In this operation due to having one monostatic GPR system equipped by shielded antenna with central frequency of 250 MHz, some of the survey parameters containing central frequency, antenna separation and antenna directivity are invariant. The most important investigated survey parameters are temporal sampling frequency, spatial sampling distance (trace intervals), time window and number of stacked traces.
 
Results and discussion
Regarding carried out investigations through field data acquisition, in only one case the GPR system failed to detect any understated targets which this mode is related to choice a sampling distance of 1 cm and a sampling frequency of 504 MHz. The sampling frequency of 504 MHz is just capable to detect the surface water pipeline (due to its low burial depth). Also only in three cases the GPR system is capable to detect all subsurface targets so that the first mode of the trace interval is 2 cm and the sampling frequency is 1954 MHz, whereas in the latter two, the trace interval is 1 cm and the sampling frequencies have been selected 1563 and 1954 MHz. At the end success or failure of the targets detection was investigated on the basis of selected survey parameters and the probability of successful target detection was determined depending on the temporal and spatial sampling frequency so that the maximum probability of target detection is regarding to temporal sampling frequency of 1954 MHz and trace interval of 1 cm. Regarding GPR field data acquisition, considering the relations between the central frequency of GPR measurement systems, the depth of penetration and resolution, the diversity of materials and various components of the host media of targets and their surface overburdens a range of dierse equipments with a variety of frequencies is needed, which all of them are not generally available.
 
Conclusion
As a general conclusion of this study, in order to reduce the risk in GPR data acquisition operation, optimal survey parameters are suggested as follows:
The sampling frequency should be about 7 to 8 times the central frequency of the employed system (should not be less than this value in order to avoid aliasing and on the other hand, due to reduction in the amount of data and thus the memory needed for storage and processing), trace interspacing equal to 1 cm (in order to detect all buried targets especially targets with small size), the number of stacked traces equal to 16 (to reduce the amount of computer memory required for processing and storing data) and time window according to the computational-empirical relation (1).
                                                                                                                                                                (1)
Where W is time window, D is the maximum depth and V is the minimum velocity.
The results of this research are not restricted to the investigated case, but in practice are applicable for cases with similar host environments, especially in urban areas (which application of non-destructive methods such as GPR is necessary)../files/site1/files/131/6Extended_Abstract(1).pdf