Analysis of Low-frequency Drone-Borne GPR for Soil Electrical Conductivity Mapping

In this study, we analysed the low-frequency drone-borne ground-penetrating radar (GPR) and full-wave inversion method for soil electrical conductivity characterization. The sensitivity of parameters (the relative dielectric permittivity εr and electrical conductivity σ) is investigated based on the full-wave inversion method. For the full-wave inversion analysis, numerical experiments were conducted within the frequency range of 15–45 MHz. The results show that the soil surface reflexion is more sensitive to the soil electrical conductivity than the soil permittivity to be retrieved using full-wave inversion within this frequency range, with a sensitivity depth of around 0.5 m. Yet, the permittivity also affects the results. For the field experiment, a 5-meter long dipole was mounted on a drone to provide the low frequency range of 15–45 MHz. The measurements were conducted during two different days. The conductivity maps obtained by the drone-borne GPR and by electromagnetic induction (EMI) are presented for comparison. The agreement between the two drone-GPR experiments demonstrated the feasibility of soil conductivity mapping using drone-borne GPR and full-wave inversion. Nevertheless, the conductivity maps were opposite in terms of spatial distributions compared to EMI. This was attributed to the different depth sensitivities of the two methods in relation to the local soil pedology.

I. Introduction

Ground-penetrating radar (GPR) has proven to be one of the promising methods for road inspection [1] , [2] , tree trunk imaging [3] , [4] , buried objects detection [5] , [6] , soil surveys [7] – [9] etc. Nowadays, with the drone technology development, and its features of being low-cost, easy-controlled, high-effective etc. [10] , a growing number of new possibilities are available for scientific applications. A commercial GPR was set up on a drone for snow thickness measurement [11] , and a time-domain GPR was used for landmine detection [12] etc. For soil properties characterization, we recently proposed a lightweight frequency-domain GPR and used it for soil moisture mapping [13] .