Comparative study of soil moisture estimations from SMOS satellite mission, GLDAS database, and cosmic-ray neutrons measurements at COSMOS station in Eastern Poland

• Cosmic-ray neutrons agreement with SMOS and GLDAS data is high.
• Triple Collocations confirms similar random error for three data sets.
• Subsurface soil moisture from cosmic-ray neutrons is comparable with subsurface from SMOS.

The paper presents a detailed comparative study of three surface soil moisture datasets retrieved from Cosmic-Ray Soil Moisture Observing System (COSMOS) in situ neutron measurement, CATDS Soil Moisture Ocean Salinity (SMOS) satellite microwave observations, and modelled data of the Global Land Data Assimilation System (GLDAS). Subsurface datasets were also calculated from the in situ and satellite measurements by using an exponential filter and were compared with the GLDAS estimates. For these comparisons, the Triple Collocation (TC) method, Nash–Sutcliffe Efficiency (NSE) coefficients, trend charts, and scatterplots were used.The main goal of this work was to verify the concordance of cosmic-ray neutron measurements with low-resolution soil moisture data from the CATDS SMOS and GLDAS products. The second objective was to determine the possibility of obtaining comparable subsurface soil moisture products from the aforementioned soil moisture data source.The obtained results show that data from the COSMOS sensor, which assesses soil moisture in an area 600 m in diameter, agree reasonably well with the CATDS SMOS and GLDAS data having spatial resolution of about 25 km. These conclusions suggest that COSMOS Derlo measurements can be particularly useful for validation of low-resolution satellite soil moisture observations as well as modelled values.The results obtained by using the TC method also revealed satisfactory agreement among all studied surface soil moisture data.However, the performed analysis shows some preponderance of SMOS and COSMOS data over GLDAS products. Although GLDAS data show a noticeable smoothing effect, COSMOS and CATDS SMOS more effectively reveal temporal soil moisture changes. Thus, for some applications, the use of CATDS SMOS estimates rather than GLDAS products may be more appropriate.Results retrieved by using an exponential filter are significant and encouraging. In particular, subsurface soil moisture values calculated from CATDS SMOS show stronger correlation with COSMOS Derlo data than those of GLDAS. Furthermore, COSMOS Derlo data also respond more intensively to surface soil moisture changes.