Soil moisture content retrieval over meadows from Sentinel-1 and Sentinel-2 data using physically based scattering models

Soil moisture content (SMC) information at field scale could have important applications in agricultural and regional water management. This study presents an operationally applicable scheme for SMC retrieval over meadows from synthetic aperture radar (SAR) backscatter (σ0) observations. We parameterized the vegetation scattering and absorption model developed at the Tor Vergata University of Rome (TV) and the integral equation method (IEM) surface scattering model for grass-covered soil surfaces. Leaf area index (LAI) estimates from a Sentinel-2 product provide field-scale vegetation information, as is demonstrated by validation against in situ measurements. The SMC retrieval scheme is applied with field-averaged Sentinel-1 σ0 observations from November 2015 to November 2018 and evaluated on 21 meadows against adjacent in situ station measurements, without (IEM) and with a vegetation correction (TV-IEM). Masking the IEM and TV-IEM SMC retrievals for dense vegetation conditions improves their performance, but this is a trade-off with the number of retrievals. By setting the SMC retrievals that exceed the upper retrieval limit of 0.75m3m−3 during the wet period to the maximum SMC, the performance metrics improve to mean Pearson correlation coefficients of 0.55 for IEM and 0.64 for TV-IEM, root mean square deviations (RMSD) of 0.14m3m−3 for IEM and 0.13m3m−3 for TV-IEM, and RMSDs relative to the range of the SMC references of 24% for both IEM and TV-IEM. The slightly better SMC retrieval performance with TV-IEM is caused by invalid SMC retrievals under dense vegetation conditions, and the performance metrics for IEM and TV-IEM are practically equal by considering the same retrieval–reference pairs. The IEM and TV-IEM retrieval performances are also similar to the performances of two other Sentinel-1 based products at field scale. They are, on average, outperformed by NASA’s Soil Moisture Active Passive (SMAP) 9km and 36km products evaluated at field scale, but these products are expected to deviate if larger regional differences are present and in field-specific situations.