Soil Moisture Estimation From Spire GNSS-Reflectometry Reflectivity: Semi-Empirical and Linear Regression Methods
Year: 2026
Type: article
Abstract: This study leverages ∼ 6-month Spire GNSS reflectometry (GNSS-R) reflectivity observations, from late Jan to Jul 2024, to develop semi-empirical (denoted as Phy) and linear regression (Linreg) algorithms to estimate near-surface soil moisture (SM). A semi-empirical model of GNSS-R reflectivity is initially developed, considering soil properties, vegetation opacity depth (VOD), and surface roughness. The Soil Moisture Active Passive (SMAP) VOD data at a resolution of 9 km are used to predict vegetation attenuation throughout this study. Effective surface roughness attenuation maps at different incidence angles at 9 km resolution are determined using CYGNSS reflectivity and SMAP SM data. After correcting for vegetation and roughness impacts, corrected Spire reflectivity observations are gridded at 36 km resolution daily. We then use these daily measurements to invert soil dielectric permittivity and SM based on our semiempirical model at each grid. Linear regression parameters are also determined by correlating corrected reflectivity measurements to SMAP SM data for each grid that are then applied to Spire data to estimate SM. The performance of our algorithms is assessed over Australia. Spire-Linreg SM retrievals show better agreement with the SMAP SM observations than Spire-Phy retrievals. Spire-Linreg SM data have a correlation coefficient of 0.85 and a root mean squared difference (RMSD) of 0.05 cm 3 /cm 3 , however, Spire-Phy data have a correlation coefficient of 0.64 and a RMSD of 0.07 cm 3 /cm 3. Both Spire-Phy and Linreg SM retrievals show larger RMSD over denser and higher vegetation canopies. The RMSD of Spire-Phy SM retrievals is less than 0.06 cm 3 /cm 3 over barren and sparsely vegetated regions. The RMSD of Spire-Linreg SM data is less than 0.06 cm 3 /cm 3 over nearly all landcover types. This study highlights our effort to develop a semi-empirical method to invert SM from GNSS-R reflectivity. In the future, improved correction of vegetation and surface roughness is necessary to enhance the performance of both semi-empirical and linear regression methods.
Source: IEEE Access
Institution: University Corporation for Atmospheric Research
Language: English
Subfield: Environmental Engineering
Field: Environmental Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,075
Award: G7727620887
OpenAlex ID: W7128314707
