An Iterative Generalized Hybrid Decomposition for Soil Moisture Retrieval Under Vegetation Cover Using Fully Polarimetric SAR

An iterative, generalized hybrid polarimetric decomposition, combining model-based and eigen-based techniques together with a generalized vegetation model, is developed for soil moisture retrieval under agricultural vegetation cover. The algorithm is physically based without the need of empirical calibration or fitting with auxiliary data and runs in two iterations. The algorithm is applied on L-band fully polarimetric data sets acquired by DLR’s E-SAR sensor. The flights were conducted within the AgriSAR, OPAQUE, and SARTEO campaigns carried out between 2006 and 2008 on three different test sites. The algorithm achieves inversion rates always higher than 95% for a variety of crop types at different phenological stages. The validation is performed against in situ time-domain reflectometry (TDR), frequency-domain reflectometry (FDR), and gravimetric measurements. The moisture levels range from 5 vol.% to 40 vol.%. The achieved root-mean-square error (RMSE) levels stay between 4.0 vol.% and 4.4 vol.% for all three sites across different vegetation and soil types, comprising the entire phenological cycle (e.g., April–July 2006).