Multi-frequency radiometer-based soil moisture retrieval and algorithm parameterization using in situ sites

Authors: GAO, Y. - Monash University; COLLIANDER, A. - Jet Propulsion Laboratory; BURGIN, M.S. - Jet Propulsion Laboratory; WALKER, J. - Monash University; DINNAT, E. - Goddard Space Flight Center; CHAE, C. - Jet Propulsion Laboratory; Cosh, Michael; CALDWELL, T. - University Of Texas; BERG, A. - University Of Guelph; MARTINEZ-FERNANDEZ, J. - University Of Salamanca; Seyfried, Mark; Starks, Patrick; Bosch, David; MCNAIRN, H. - Agriculture And Agri-Food Canada; SU, Z. - University Of Twente; VAN DER VELDE, R - Collaborator

Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/4/2022
Publication Date: 9/22/2022
Citation: Gao, Y., Colliander, A., Burgin, M., Walker, J., Dinnat, E., Chae, C., Cosh, M.H., Caldwell, T., Berg, A., Martinez-Fernandez, J., Seyfried, M.S., Starks, P.J., Bosch, D.D., Mcnairn, H., Su, Z., Van Der Velde, R. 2022. Multi-frequency radiometer-based soil moisture retrieval and algorithm parameterization using in situ sites. Remote Sensing of Environment. 279. Article 113113. https://doi.org/10.1016/j.rse.2022.113113.
DOI: https://doi.org/10.1016/j.rse.2022.113113

Interpretive Summary: Several soil moisture satellite data records are now available for analysis, but each is based on a different remote sensing algorithm. A single harmonized soil moisture data record would be a desirable product for application and model development. This study calibrated algorithm parameters across satellite products to generate a new data record with relatively equal or better accuracy as compared to validation sites when compared to individual satellite records. It was also demonstrated that the newest generation satellites which are based upon L-band technology provide the best accuracy. This study is useful for climatologists and modelers for long time series studies of land surface water budgets.

Technical Abstract: This study focused on the parametrization of the single channel algorithm for soil moisture retrieval at L-, C- and X-band using brightness temperature observations from four satellite missions and in-situ data from core validation sites across various land cover types. Several satellite missions carrying microwave radiometers have been launched over the past years. Among them are Japan Aerospace Exploration Agency’s (JAXA’s) Advanced Microwave Scanning Radiometer 2 (AMSR2) onboard the GCOM-W satellite, and NASA’s Soil Moisture Active Passive (SMAP) mission. Simultaneous calibrations of the vegetation parameter b and roughness parameter h at both horizontal and vertical polarizations were performed using in-situ data from 2015-2016. A validation of the calibrated parameters was performed with data of 2017. Calibrated model parameters to successfully retrieve soil moisture at 12 different validation sites at L-, C- and X-band are presented and their dependence on frequency are investigated. Results indicate that for soil moisture retrieval accuracy, L-band V-pol performs the best among all channels, with an unbiased root-mean-square difference (ubRMSD) ranging from 0.020-0.080 m3/m3. The retrieval accuracy for C- and X-band at V-pol is reasonable with an ubRMSD of 0.030-0.104 m3/m3. The validation results show that the SMAP default parameters have a varying performance at the 12 study sites (RMSD ranges from 0.027-0.097 m3/m3 considering both H- and V-pol). Our calibrated parameters improve the L-band accuracy by 0.002-0.041 m3/m3 while the C- and X-band accuracies remain similar to calibration. Overall, this study using near-simultaneous spaceborne observations over a number of sites showed that while the C- and X-band can provide soil moisture at a reasonable accuracy, the accuracy offered by the L-band sensitivity to soil moisture is clearly superior.