A multi-temporal and multi-angular approach for systematically retrieving soil moisture and vegetation optical depth from SMOS data

Authors: BAI, Y. - Chinese Academy Of Sciences; ZHAO, T - Chinese Academy Of Sciences; JIA, L. - Chinese Academy Of Sciences; Cosh, Michael; SHI, J. - Chinese Academy Of Sciences; PENG, Z - Chinese Academy Of Sciences; LI, X - Chinese Academy Of Sciences; WIGNERON, J.P. - National Research Institute For Agriculture, Food And Environment

Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/19/2022
Publication Date: 8/13/2022
Citation: Bai, Y., Zhao, T., Jia, L., Cosh, M.H., Shi, J., Peng, Z., Li, X., Wigneron, J. 2022. A multi-temporal and multi-angular approach for systematically retrieving soil moisture and vegetation optical depth from SMOS data. Remote Sensing of Environment. 280:113190. https://doi.org/10.1016/j.rse.2022.113190.
DOI: https://doi.org/10.1016/j.rse.2022.113190

Interpretive Summary: Microwave radiometry is able to provide an estimate of a variety of surface variables, including surface roughness, vegetation optical depth, and surface soil moisture. But it is challenging to account for all three of these features with a single satellite look, and requires assumptions or alternate data sources. Using multi-temporal and multi-angular data from the Soil Moisture Ocean Salinity (SMOS) mission, an improved set of parameters are estimated, all from the same satellite platform, without the accompanying assumptions. There was an improvement in the estimation of soil moisture as compared to in situ resources. This study is of value to remote sensing scientists and modelers who are attempting to interpret features of the land surface within the microwave spectrum.

Technical Abstract: Soil moisture retrieval is an example of no-linear and ill-posed problems, as microwave emission from landscape is affected by a variety of surface parameters including soil moisture, soil texture, soil surface roughness, vegetation optical depth (VOD), vegetation structure, and atmospheric properties etc. It is an effective means to make the retrieval results more robust or enable more parameters to be retrieved by increasing observation information. In this study, a multi-temporal and multi-angular approach is developed with SMOS (Soil Moisture and Ocean Salinity) data for simultaneously retrieving soil moisture, VOD, effective single scattering albedo '_eff, and surface roughness. The 2015-2016 SMOS L-band multi-angle observations at the horizontal polarization are used to develop the methodology proposed in this study. The retrieved VOD (Re-VOD) is found to have a global spatial distribution of different vegetation types that is consistent with SMOS Level 3 (SMOS-L3) and SMOS-IC VOD products. The Re-VOD over selected validation sites reflects the vegetation seasonal variation characteristics and shows the phenomenon of peak lag compared with the normalized difference vegetation index (NDVI). The spatial distribution of retrieved '_eff generally shows a dependence on both VOD and land cover type. Results of retrieved surface roughness, Z_s, range from 0.04 to 0.22 cm, and its spatial distribution is consistent with most existing roughness products except for parts of the Sahara Desert. The retrieved soil moisture (Re-SM) shows generally high correlations with in-situ measurements with correlation coefficients greater than 0.6. The average absolute value of Bias for SMOS-IC is 0.056 cm3/cm3, which is slightly higher than that of Re-SM (0.045 cm3/cm3) and SMOS-L3 (0.042 cm3/cm3). The ubRMSE of Re-SM varies from 0.022 to 0.059 cm3/cm3 and is a better than that of SMOS-IC (0.036 to 0.069 cm3/cm3) and SMOS L3 (0.053 to 0.074 cm3/cm3). Therefore it is concluded that by inclusion of multi-temporal SMOS data, the proposed method is able to retrieve more surface parameters of effective single scattering albedo and surface roughness with accurate soil moisture and VOD products as compared with SMOS-L3 and SMOS-IC.