Deep Learning-Based Soil Moisture Retrieval in CONUS Using CYGNSS Delay–Doppler Maps

NASA's Cyclone Global Navigation Satellite System (CYGNSS) mission has gained significant attention within the land remote sensing community for estimating soil moisture (SM) by using Reflectometry (GNSS-R) technique. CYGNSS constellation generates Delay-Doppler Maps (DDM)s, containing important earth surface information from GNSS reflection measurements. Many previous studies considered only designed features DDM such as peak value of DDM, whereas whole image is affected SM, topography, inundation, and overlying vegetation. In this paper, a deep learning (DL)-based framework presented SM products in Continental United States (CONUS) leveraging spaceborne GNSS-R observations provided along with other remotely sensed geophysical data products. A data-driven approach utilizing convolutional neural networks (CNNs) developed to determine complex relationships between reflected measurements parameters which can help provide improved estimation. The CNN model trained jointly three types processed images Analog Power, Effective scattering area, Bistatic Radar Cross-section (BRCS) auxiliary elevation, properties, normalized difference vegetation index (NDVI) water content (VWC). evaluated Soil Moisture Active Passive (SMAP) mission's enhanced at 9 km × resolution VWC less than 5 kg/m$^{2}$. mean unbiased root-mean-square (ubRMSD) concurrent SMAP retrievals 2017 2020 0.0366 $m^{3}/m^{3}$ correlation coefficient 0.93 over 5-fold cross-validation 0.0333 0.94 year-based spatial temporal same mission.