Soil Moisture Estimation Using the Correlation between Dual-Polarization GNSS-R Interference Patterns

Soil moisture is a key variable routinely used to understand and predict the behavior of Earth’s climate and water cycle. Over the past decade, Global Navigation Satellite Systems Reflectometry (GNSS-R) techniques have emerged as a “signal of opportunity” for continuous and near real-time soil moisture estimation. Since 2007, multipath signals have been used to estimate soil moisture primarily through three ground-based GPS receiver setup configurations. The first configuration uses two antennas, one looking toward the zenith to acquire the direct signal, and the other looking toward the ground to acquire the reflected (multipath) signal. With this ground-based GPS receiver configuration, soil moisture can be estimated from the reflection coefficient computed by dividing the averaged waveforms from direct and reflected GNSS signals. The second configuration employs an interferometric GNSS-R ground-based receiver with a single antenna, and it estimates soil moisture by analyzing the phase, amplitude, and frequency of the interference pattern between the direct and reflected signals. The third ground-based receiver configuration is known as the interference pattern technique (IPT). It employs a dual-polarized antenna oriented horizontally to measure the power fluctuations of the interference of direct and reflected signals at horizontal polarization (H-pol) and vertical polarization (V-pol). With the IPT, soil moisture is currently estimated by determining the so-called notch position, θB: the angular elevation value (θ) of the smallest interference power (IP) oscillation at V-pol. Accurate determination of θB is challenging in real GNSS-R acquisitions, especially when the IP waveform exhibits low-frequency oscillations or maintains constant amplitude over a wide range of θ. Here, we investigate the potential of a ground-based GNSS-R receiver with two linearly polarized antennas that measure the IP of direct and reflected signals in H-pol and V-pol to estimate soil moisture in a patch of very smooth bare soil and an irrigated grassland field harvested every three months. This study provides a practical method to estimate the soil moisture, through the use of the coefficient of determination between the IP waveforms at H-pol and V-pol (R²v/h). A coherent specular reflection model was employed to first explore the relationship between  and soil moisture for different values of soil roughness and vegetation water content. That relationship was applied to estimate soil moisture from  determined from GPS signals acquired continuously between May and December 2022 of bare soil (vegetation water content equal to zero). The results show that the proposed method can estimate the soil moisture of the upper 10 cm layer of bare soil with high accuracy (RMSE of 1.5 vol.%). The use of  in the irrigated grassland field produced inaccurate estimates of soil moisture, likely due to the presence of vegetation causing V-pol and H-pol to consistently oscillate out-of-phase (R²v/h= 0). The work is currently focusing on the use of amplitude and frequency of V-pol and H-pol to improve soil moisture estimation in the irrigated grassland field.