Evaluation of the Tau–Omega Model for Passive Microwave Soil Moisture Retrieval Using SMAPEx Datasets

The parameters used for passive soil moisture retrieval algorithms reported in the literature encompass a wide range, leading to a large uncertainty in the applicability of those values. This paper presents an evaluation of the proposed parameterizations of the tau–omega model from 1) the soil moisture active passive (SMAP) algorithm theoretical basis document (ATBD) for global condition and 2) calibrated parameters from the National Airborne Field Experiment (NAFE’05) for Australian conditions, with special focus on the vegetation parameter <italic>b</italic> and roughness parameter <inline-formula><tex-math notation="LaTeX">$H_{R}$</tex-math></inline-formula>. This study uses airborne <italic>L</italic>-band data and field observations from the SMAP experiments conducted in south-eastern Australia. Results show that the accuracy with the proposed parameterizations from SMAP ATBD was satisfactory at 100-m spatial resolution for maize (0.07 m³/m³) and pasture (0.07 m³/m³ ), while it decreased to 0.19 m³/m³ for wheat. Calibrated parameters from the NAFE’05 did not provide better results, with the accuracy of wheat degrading to 0.23 m³/m³. After a comprehensive site-specific calibration and validation at 100-m spatial resolution, this result was improved to 0.10 m³/m³. Further calibration and validation were performed at 1-km resolution against intensive ground sampling and at 3-km against <italic>in situ</italic> monitoring stations. Results showed an accuracy over grassland and cropland of 0.04 m³/m³ and 0.05 m³/m³, respectively. This study also suggests that the parameters from SMAP ATBD show an underestimation of soil moisture, with the roughness parameter <inline-formula><tex-math notation="LaTeX">$H_{R}$</tex-math></inline-formula> being too low for south-eastern Australian condition. Therefore, a new set of <italic>b</italic> and <inline-formula> <tex-math notation="LaTeX">$H_{R}$</tex-math></inline-formula> parameters for ten different land cover types was proposed in this study.