In-Situ and Triple-Collocation Based Evaluation of Global Root-Zone Soil Moisture Products Across China

Soil moisture (SM) is a key variable in the interaction between land and the atmosphere, playing an important role in various fields such as agriculture, ecology, atmospheric science, and drought/flood risk assessment. Satellite-based remote sensing technologies typically provide surface SM, while root-zone SM needs to be simulated using model-based assimilation techniques. It is essential to evaluate the accuracy and reliability of these root-zone soil moisture products to provide applicable data support for relevant fields such as agricultural water resource management and climate change research. In this study, in situ observations from 2,766 sites across China from 2016 to 2020 and a triple collocation (TC) method are used to evaluate five commonly used root-zone SM products, including ERA5, GLDAS-2.1, GLDAS-2.2, GLEAM v3.8a and GLEAM v3.7b datasets. The evaluation is conducted across China and in different climate regions. The evaluation result based on in situ observations indicates that all products tend to overestimate but can generally capture the temporal variation, and GLDAS-2.2 exhibits the best performance with the highest median correlation coefficient (R=0.44) and the lowest median unbiased root mean square error (ubRMSE = 0.0377m³.m^-3) against in situ observations. The evaluation result based on TC method indicates that GLDAS-2.2 has the lowest median error standard deviations (ESD =0.0014m³.m^-3), followed by GLEAM v3.7b(ESD=0.0026m³.m^-3). GLEAM v3.7b has the highest median TC-derived R(0.887), followed by GLDAS-2.2 (0.807). The applicability of different products varies in different climate regions. In arid and semiarid, Tibetan Plateau, and temperate humid regions, the evaluation results are comparable to that obtained nationwide. However, in tropical/subtropical humid regions, the result shows that the values of GLEAM v3.7b and v3.8a are closer to in situ SM observations, and TC-based evaluation indicates that GLEAM v3.7b is the most suitable product for this region. GLDAS-2.1 and -2.2 have consistently demonstrated high accuracy. Overall, the results of TC-based evaluation of rootzone soil moisture products are highly consistent with that of insitu based evaluation.