Calibration of SMOS soil moisture retrieval algorithm: a case of tropical site in Malaysia

Abstract

Soil Moisture and Ocean Salinity (SMOS) mission has successfully contributed to global soil moisture products since 2009. Validation and calibration activities were conducted worldwide, yet some of the validation results do not fulfill the targeted accuracy of ±0.04 m3m-3. This paper presented the site-specific calibration of the V620 retrieval algorithm with in situ data collected at selected agricultural sites in the humid tropical regions, Malaysia. This set of data has been validated where low accuracy of SMOS soil moisture products was found. To improve the SMOS soil moisture retrieval, calibration of SMOS soil moisture retrieval algorithm based on the L-band Microwave Emission and Biosphere model and SMOS Level 1C TB products, considering the local parameters was conducted. The calibration proves that these site-specific parameters improve the product's accuracy. Validation of SMOS Level 2 product with in situ data showed bias, root-mean-square error (RMSE), and unbiased RMSE (ubRMSE) ranging from 0.050 to 0.118 m3m-3 , 0.068 to 0.142 m3m-3 , and 0.069 to 0.103 m3m-3 , respectively. The soil moisture retrieval based on the calibrated model showed an improved bias of 0.020-0.056 m3 m-3 and RMSE of 0.026-0.065 m3m-3. The ubRMSE ranges from 0.017 to 0.034 m3m-3. Recently released SMOS-IC V105 product was also validated, where small improvements were noticed when compared to the accuracy of SMOS Level 2. This paper shows the importance of local parameters in retrieving soil moisture with higher accuracy compared to the use of global generalized parameters that are used in the original SMOS soil moisture retrieval algorithm.