Field-Scale Soil Moisture Sensing Using GPS Reflections: Description of the PBO H2O Soil Moisture Product

Abstract:

Data from NSF’s EarthScope Plate Boundary Observatory (PBO), and similar GPS networks worldwide, can be used to monitor the terrestrial water cycle. GPS satellites transmit L-band microwave signals, which are affected by water at Earth’s surface. GPS signals take two paths: (1) the “direct” signal travels from the satellite to the antenna; (2) the “reflected” signal interacts with the Earth’s surface before travelling to the antenna. The direct signal is used by geophysicists to measure position of the antenna, while the effects of reflected signals are generally ignored. Recently, our group has developed a technique to retrieve terrestrial water cycle variables from GPS reflections. The sensing footprint is intermediate in scale between in situ and remote sensing observations. Soil moisture, snow depth, and an index of vegetation water content are estimated from data collected at over 400 PBO sites. The products are updated daily and are available online. This presentation provides a description of the soil moisture product.

Near-surface soil moisture is estimated at more than 100 sites in the PBO H₂O network. At each site, a geodetic-quality GPS antenna records the interference pattern between the direct and ground-reflected GPS signals in signal-to-noise ratio (SNR) interferograms. The ground-reflected GPS signal is altered by changes in the permittivity of the ground surface, which is primarily a function of its water content. Temporal changes in the SNR interferogram, primarily its phase, are indicative of changes in soil moisture. SNR phase data are converted to soil moisture using relationships determined using an electrodynamic model. Soil moisture is not retrieved when there is snow or significant vegetation (> ~1 kg m^-2 of vegetation water), as both affect SNR phase. When there is moderate vegetation, a correction is applied to the phase data before conversion to soil moisture. The effect of vegetation on SNR phase and the exact relationship between SNR phase and soil moisture are the primary sources of error. At sites where seasonal variations in vegetation water content are < 0.5 kg m^-2, validation surveys show that the RMSE between in situ and GPS soil moisture estimates is < 0.04 cm³ cm^-3. Therefore, soil moisture data from these sites could be utilized for satellite validation and other applications.