Near Surface Soil Moisture Vertical Profile Retrieval Based on GNSS Interferometric Reflectometry

Abstract:

The soil moisture content (SMC) is a key environmental variable, whose anomalies is shown to have strong connections to climate. One branch of methods attracting attention recently is to estimate the SMC utilizing the signal of opportunity from the Global Navigational Satellite System (GNSS). The GNSS Interferometric Reflectometry (GNSS-IR) technique makes use of the multi-path interference between the direct signal and the reflected signal to retrieve SMC from the angular pattern of the received signal to noise ratio (SNR). GNSS-IR takes advantage from existing geodetic GNSS antennas in place across the globe and does not require newly deployed hardware. The GNSS-IR technique has been demonstrated for measuring surface soil moisture. However, earlier work ignores the rapidly varying SMC vertical profile in the analysis, and the sensitivity of the L-band GNSS-IR signal to the vertical profile of the soil moisture is yet to be examined.

In this paper, we examine the potential and performance of GNSS-IR for near surface soil moisture vertical profile retrieval through a physical scattering model. The vertical profile of soil moisture is parameterized based on the physical solution to the Richards’ equation for unsaturated flow in soils. By considering the Fresnel reflection coefficients derived from Maxwell’s equations on flat multi-layered media (in this case the multi-layer soil with different moisture), and using the Mironov soil permittivity model to link the soil permittivity at L-band to its moisture content and clay fraction, our model comprehensively connects the soil moisture and GNSS SNR data received by antenna. The soil moisture profiles are then retrieved by matching the GNSS SNR angular patterns to our model predictions with a least-mean-square-error (LMSE) based retrieval algorithm. The retrieval algorithm is applied to both synthesized data and the UNAVCO experimental data to test its performance, and it shows reasonably good performances when applied to model-synthesized SNR data with an average moisture bias in the top 20cm of soil in general less than 0.03. It also yields promising near-surface soil moisture profile information when applied to UNAVCO SNR measurements.