Comprehensive Analysis of CYGNSS GNSS-R Data for Enhanced Soil Moisture Retrieval

Soil moisture, an essential climate variable, is traditionally retrieved on large scales using passive or active microwave sensors, with temporal resolution of 2–3 days and no less than 6 days, respectively. Global navigation satellite system-reflectometry represents an emerging concept to retrieve geophysical parameters, including soil moisture, with an improved spatiotemporal resolution compared to traditional sensors. This article outlines a large-scale near-surface soil moisture product derived from Cyclone GNSS (CYGNSS) observations, provided daily at both 9 and 36 km. The proposed algorithm assumes that soil moisture variations from the Soil Moisture Active Passive (SMAP) mission are linearly correlated with changes in surface reflectivity. Surface reflectivity is computed from a subset of the delay-Doppler maps and subsequently normalized for reflection geometry using linear regression, which correlates reflectivity with incidence angle; this approach accounts for the varying effects of coherent and incoherent scattering. We thoroughly assessed our product using over three years of data. Compared to SMAP, we found a median unbiased root-mean-square error of 0.039 cm³cm⁻³, with varying accuracy depending on the land cover type, and of 0.027 cm³cm⁻³ compared to CYGNSS calibration/validation sites. In addition, we performed a triple collocation analysis using 257 in-situ sites and observed similar behavior in our product and SMAP, with an overall larger random noise component associated with CYGNSS. Available upon request, the University of Luxembourg product provides soil moisture information for applications demanding quicker revisit times than traditional products.