Assessment of five SMAP soil moisture products using ISMN ground-based measurements over varied environmental conditions
Chuanxiang Yi,Xiaojun Li,Jiangyuan Zeng,Lei Fan,Zhiqing Xie,Lun Gao,Zanpin Xing,Hongliang Ma,Antoine Boudah,Hongwei Zhou,Wenjun Zhou,Ye Sheng,Tianxiang Dong,Jean-Pierre Wigneron
DOI: https://doi.org/10.1016/j.jhydrol.2023.129325
IF: 6.4
2023-03-08
Journal of Hydrology
Abstract:Since the launch of Soil Moisture Active Passive (SMAP) mission in 2015, several advanced global surface soil moisture (SM) products derived from different retrieval schemes have been released. However, the factors contributing to the performance differences of these SMAP products have not been fully investigated. This study tries to fill the gap by assessing and intercomparing five SMAP products, namely, the single channel algorithm V/H-pol (SCA-V/H), dual channel algorithm (DCA), multi-temporal dual channel algorithm (MT-DCA) and the new SMAP-INRAE-BORDEAUX (SMAP-IB), using the International Soil Moisture Network (ISMN) ground measurements worldwide during 2016–2020. After reviewing the retrieval algorithms, the impact of five potential perturbing factors on the skills of all products is reported; these factors including mean surface soil temperature (MSST), surface roughness (log(SR)), vegetation density (characterized by vegetation optical depth (VOD)), mean surface soil wetness (MSSM) and soil organic matter (SOM). Results shows that the SMAP-IB product has the highest overall R value of 0.75 with ISMN ground SM, and the smallest unbiased root mean square difference ( ub RMSD) values same as SCA-V and DCA (∼0.058 m 3 /m 3 ). The performance of all five products in terms of R and ub RMSD rapidly degrades over the regions with high VOD, rough surface and low MSST. In contradiction with the fact that product performance degrades with increasing roughness, the ub RMSD of all five products increases with decreasing log(SR) on a relatively smooth surface (log(SR) < 6) with larger growth rates for SMAP-IB, SCA-H and MT-DCA. Regarding characterizing the temporal variations of in situ SM anomalies ( R anom ), R anom increases as the ground becomes wet for all products within a certain range (MSSM < 0.3 m 3 /m 3 ). In particular, SMAP-IB and MT-DCA present higher R anom values than the other three products under wet soil surfaces (MSSM > 0.3 m 3 /m 3 ). The two metrics R and R anom decrease significantly with increasing SOM for SMAP-IB compared to the other products, and SMAP-IB obtains the lowest values for both metrics when SOM > 20%. These findings are expected to provide useful information for appropriate product selection and improvement.
geosciences, multidisciplinary,water resources,engineering, civil