Reverse Time Migration For Subsurface Imaging

Ground-penetrating radar (GPR), as one of the promising non-destructive detection and imaging tools, has been popularly applied in various fields. In this paper, we present a milti-input and multi-output (MIMO) GPR system for subsurface imaging of lunar regolith structure. This system is equipped with twelve Vivaldi antennas, each of which can be used as transmitter and receiver. Unlike a traditional GPR system, this system works in a stationary mode, and would not move along the survey line. We developed a two-dimensional reverse time migration algorithm for high-resolution subsurface imaging using this MIMO GPR system.RTM algorithm mainly consists three steps, i.e. the forward and backward modeling of the electromagnetic fields, and the imaging condition. As a mature technology, the finite-difference time-domain (FDTD) method simulates the transient electromagnetic wave field. Those antennas in the computation region are simplified as point sources at corresponding position. In the underground area. Cross-correlation of the forward and backward electromagnetic fields in the computation domain is used as the imaging condition.Fig. 1 shows the model for a two-dimensional numerical experiment. Only ten antennas are applied in the 2-D RTM algorithm since the other two antennas are out of the imaging plane. The relative dielectric permittivity and electric conductivity of the lunar soil are respectively set to be 2.5 and 0.01 mS/m. Since no prior information about the subsurface structure and the buried objects could be obtained, a half-space initial model is employed for the 2D RTM. After the image reconstruction by RTM, we can clearly identify the buried object as well as the air/soil interface, as shown in Fig. 2.In the laboratory experiment, as shown in Fig. 3, a marble slab of 60 cm x 60 cm is buried at the depth of 2 m in a volcanic ash pit. The dielectric properties of the volcanic ash are close to those of lunar soil. The relative dielectric permittivity of the volcanic ash is about 2.5 and the electric conductivity is negligibly small. The thickness and relative dielectric permittivity of the marble slab are 3 cm and about 8, respectively. At the depth of 2.5 m, e.g. the bottom of the ash pit, there is a metal plate. A frequency domain filter is applied to the acquired three-dimensional laboratory dataset to the two-dimensional counterpart required for the 2D RTM. Fig. 4 depicts the RTM result of the buried marble slab. We can identify the upper and lower interfaces of the slabs, although they are only 3 cm in thickness. This means that the MIMO GPR system has a depth resolution better than 3 cm in marble rock. Considering the velocity difference in marble and lunar soil, we can conclude that the MIMO GPR system has a depth resolution of about 5 cm in lunar soil.