Development Of H-Alpha Feature Decomposition Technology For Full-polarimetric Ground Penetrating Radar

In many applications, being able to know the subsurface topology or structures is very useful and can greatly improve the safety and economic efficiency of operations. For instance, in civil engineering constructions, the poor construction quality associated with the notorious "jerry-built projects” which uses some inferior materials or cuts corners can be detected and, thus preventing tragedy from happing. In addition, the maintenance of key infrastructures such as bridges, tunnels, and dams could be conducted nondestructively and more efficiently, and thus preventing development of structure failures. In archaeology, ability to map underground structures reduces unnecessary destructive excavation and minimizing damaging artifacts. In agriculture, knowing subsurface soil topology and water drainage pattern efficiency help increasing crop yields. Current near-surface geophysical survey methods include gravitometer, magnetometer, electromagnetic induction, resistivity meter, seismic sensors, boring, and grond penetrating radar(GPR) etc. But some of these methods are destructive.Some of the equipment operations may be complicated, low efficiency, or require extensive training to interoperate the data. Among these near-surface geophysical technologies, ground penetrating radar(GPR) is probably the most popular one due to its non-destructive, cost effective, and ease of use.GPR utilizes radio waves to investigate and characterize subsurface distribution of discontinuity in wave propagation property which is often related to change in permittivity. GPR is a non-intrusive survey methodology, which has the advantages of fast data acquisition speed, easy to carry, easy to operate, reasonable price, and high spatial and depth resolution. In recent years, GPR has been widely used in the field of civil engineering construction, environment monitoring, archaeology surveys, geological explorations, hydrogeology surveys, geological surveys, locating agricultural irrigation and drainage systems, discovering underground tunnels, graben, mineral exploration, landmines and unexploded ordnance(UXO) detection, and planetary exploration, and have produced good results.Most GPR systems are still single polarization systems. Therefore identifying different types of scattering targets only relies on 2D slice of survey data at different depth. The single polarization systems can only obtain single polarization data, which cannot easily discriminate some common linear subsurface targets such as pipes, cables, tunnels, and UXO etc. from other types of non-linear targets.Although fully-polarimetric radars and the associated signal processing techniques have been widely used in remote sensing, fully-polarimetric GPR systems and the associated data processing techniques are still uncommon. Being able to collect full-polarization scattering data should be of great benefit in subsurface targets classification.Different common man-made or natural underground targets often have different polarimetric scattering characteristics. For instance, pipes and cables, and tunnels produce linearly polarized scattering features. Subsurface layers produce only orientation-independent co-polarization scattering. Subsurface corner structure produces unique dihedral type of depolarized scattered fields, which is different from the polarimetric characteristics of layers and pipes. These targets with different polarimetric characteristics can be easily discriminated using fully polarimetric GPR systems which produce fully polarimetric scattering matrix from which polarimetric scattering features associated with targets can be extracted. These different polarimetric scattering signatures correspond to different types of subsurface targets, and thus improving the accuracy and effectiveness of GPR surveys. Unfortunately, fully polarimetric GPR systems and associated technical issues have not been thoroughly studied. As a result, fully polarimetric GPR systems and associated signal processing technology have not been widely used.Therefore, our main research contributions include:(1) Using the experimental results to demonstarte that the method of fully polarimetric target identification based on entropy(that is, the H- alpha decomposition method) is effective for the identification of the target in GPR. This method has been applied in the field of remote sensing and borehole radar, but it has not yet been used in the surface-based GPR. These results showed that the classification of full- polarimetric target could help identify subsurface targets, so this paper first chose four different types of targets, metal ball, metal plate, metal dihedral and volume scattering with a plurality of linear branches. These targets were buried in dry sand in the laboratory. Full-polarization scattering data were subsequently obtained by using the robotic full-polarization data collection system in the laboratory. We successfully demonstrated the feasibility of H- alpha polarization decomposition method in surface GPR application, and get better classification results.(2) Developed a novel, improved fully polarimetric target classification method which is more suitable for GPR data which often suffer from clutter and noise. In theory, the original H- alpha decomposition method can classify different targets such as spherical symmetry targets(subsurface layers), dihedral(subsurface corner structure, fault), linear targets(pipelines, cables and UXO etc.), but this is limited to low noise and low clutter. In generally, the actual surface polarimetric GPR data usually contain more noise, and its strength is not weaker than the signal intensity, so in this case it is difficult to use the original H-alpha classification method to classify the targets. Based on this limit, through the study on the effect of different noise and clutter intensity on H-alpha value, developed a set of improved H-alpha target identification method. This method can be applied to high noise and clutter environment to classify dihedral targets, linear target and spherical symmetry targets.(3) Developed a novel data collection and processing methodology which allows for extracting fully polarimetric target features from single-polarization GPR systems. This dissertation and other scholars have confirmed that full-polarimetric targets identification should be helpful for identifying subsurface targets, especially for linear targets such as pipes, cables and other unexploded ordnance, it is difficult to classify this type targets only using single polarization data. But the vast majority of GPR users are using single polarization radar, only obtaining single polarization data, in order to help the users to carry out full-polarization target identification using single polarization system, this paper developed a new method for using single polarization radar system obtain full-polarimetric data.(4) Experimentally demonstrated the effectiveness of the improved full polarimetric identification method for classifying the different subsurface target types. This paper applied the improved H- alpha method of target identification to classify UXO and agricultural irrigation and drainage tube, and obtained good classification results. This paper also uses single polarization commercial radar(Sensor & Software) obtained full polarimetric data method. Drainage pipeline were measured by this new method and obtained full-polarimetric data of drainage pipeline, eventually, applied fully polarimetric data to the improved H-alpha target identification method and get better classification results.