GNSS Interferometric Reflectometry (GNSSIR) is a technique that estimates areal surface soil moisture at the field scale by analyzing interference between direct and ground-reflected GNSS signals received by near-surface antennas. In recent years, low-cost GNSS antennas and receivers have become increasingly widespread, and their application to GNSS-IR has been desired. However, low-cost receiving systems face challenges such as low data accuracy and limitations in the available satellites and frequency bands. In this study, we developed a low-cost GNSS-IR system using an L1-band GPS antenna, receiver, and Raspberry Pi microcomputer, with a total cost of approximately ¥ 10,000. Field observations were conducted using this low-cost receiving system to evaluate the accuracy of soil moisture estimation and the observation coverage provided by the tracked satellites. As a result, the calculated Fresnel zones of GPS signals captured by the system covered most of the field area, excluding approximately 60◦ in the northern direction where satellite flyovers were absent. From the Signal-to-Noise Ratio (SNR) interference patterns, we extracted three variables, phase offset (φo), signal amplitude (As), and effective reflector depth (Hdep) ̶ and compared them with in-situ volumetric water content (θm) obtained using dielectric probes. The correlation coefficients between θm and φo, As, and Hdep were r = 0.79,−0.43, and −0.63, respectively. Restricting the analysis to signals from azimuths of 120 – 240◦, where the influence of nearby buildings was expected to be smaller, the correlations improved to r = 0.78 (φo), −0.75 (As), and −0.82 (Hdep). Soil moisture estimates based on φo yielded an RMSE of 0.03 m3 m−3. These results suggest that surface soil moisture can be estimated using this low-cost GNSS-IR system, despite limitations such as the restriction to single-band L1 signals and NMEA format output. Limiting the analysis to unobstructed azimuths improved the correlations, suggesting that careful selection of satellite signal directions can improve retrieval performance.
View full abstract