Estimation of Soil Moisture Content Using Air-Launched GPR Techniques in Variable Soil Conditions

Air-launched Ground Penetrating Radar (GPR) techniques have most frequently been used for infrastructure characterization, but these techniques show promise for soil moisture estimation in the near subsurface. Air- launched GPR data can be acquired very quickly, and data processing can be easily automated, so these techniques have potential for efficient estimation of water content in the shallow subsurface over large areas. In this experiment, we investigate the efficacy of air-launched GPR techniques for estimating soil water content under saturated and dry conditions in both sandy and organic-rich soils. Data were also acquired to investigate the depth of penetration of air-launched data in these soils using multiple GPR frequencies. The experiment was performed in a large tank under controlled climatic conditions. Initially, the tank was filled with wet sand to a depth of 24-cm, and GPR data were acquired over the sand using 250-, 500-, and 1000-MHz antennas. Then, a thin plastic tarp was placed on the wet sand, a 3-cm layer of dry sand was placed on the tarp, and data collection was repeated. Additional 3-cm layers of dry sand were placed in the tank, with data acquisition after each layer, until the dry sand layer was 15-cm thick. The tank was then excavated, and a basal layer of dry sand was added. Data were again acquired over the dry sand, and the incremental filling of the tank and data acquisition were repeated using 3-cm layers of wet sand. Finally, the entire process was repeated using a basal layer of wet organic soil overlain by dry organic soil and using a basal layer of dry organic soil overlain by wet organic soil. For all air-launch data, the dielectric constant was determined using the amplitudes of the reflection from the soil surface, and Topp's equation was used to convert the dielectric constant to water content. Data analysis is ongoing, but preliminary results indicate that water content can be estimated with reasonable accuracy in both wet and dry soils, although some frequencies tend to overestimate the water content in wet soils. The penetration depth in dry soil was approximately 6- to 9-cm for all frequencies; data analysis for the penetration depth in wet soils is in progress.