Design of Ground Penetrating Radar for Antarctic Snow Thickness Profiling

Abstract

In this study, a linear frequency modulated (LFM) ground-penetrating radar (GPR) system was designed for snow thickness profiling in the Antarctic terrain. Conventional GPR systems have been based on ultra-wideband impulse radar, but there are difficulties in generating high power pulse waveforms for snow thickness profiling. High power waveforms are necessary to profile depths of tens to hundreds of meters underground, but there are significant hardware constraints in generating high power impulses sufficient to secure an adequate signal-to-noise ratio (SNR). Additionally, the hardware complexity is increased due to the application of highspeed sampling techniques required for receiving reflected impulse signals. In contrast, the LFM method can achieve additional SNR gain by using a longer duration of the transmitted signal pulse, allowing implementation with reasonable output power. Also, handling relatively lower beat frequencies makes it possible to apply simpler sampling techniques. For the design of the LFM GPR, the permittivity of dry snow, the medium in the Antarctic terrain, was investigated, and surface reflection was calculated. A finitedifference time-domain (FDTD) method simulation model was built based on the investigated the permittivity, and the surface reflection results were validated. Moreover, typical atmospheric conditions, dry snow, and ice sheet layers were modeled using FDTD to derive the characteristics of the reflected signals and to confirm whether the proposed hardware structure could sufficiently ensure the maximum detection range. The LFM GPR system was implemented based on the designed and analyzed data. Nonlinearity occurring in the voltage-controlled oscillator of the waveform generation module was resolved using the pre-distortion technique. Spurious signal due to digital clock leakage included in the digital-to-analog converter output signal was removed by applying a multi-stage reconstruction filter. Electromagnetic interference noise introduced into the radio frequency part from the power supply module was blocked using shielding techniques. Harmonic distortion and intermodulation noise caused by the nonlinearity of the RF circuit and resulting in clutter signals were reduced by applying the intermediate frequency hopping technique, and the profiling image was improved by applying a coherency algorithm. For design verification, outdoor experiments were conducted, confirming a maximum detection range of 100 m and a range resolution of 10 cm. A twodimensional radar image was created by applying the migration technique to data obtained through B-scan experiments, and the effectiveness of the proposed coherency algorithm was validated. Furthermore, field experiments were conducted in the Browning Pass plains, which is near Jang Bogo Station in Antarctica, creating snow thickness profiling image reaching approximately 12 m.