VALIDATION OF SMOS L2 AND L3 SOIL MOISTURE PRODUCTS OVER THE DUERO BASIN AT DIFFERENT SPATIAL SCALES
- 1Universidad de Salamanca/CIALE, Duero 12, 37185 Villamayor, Spain
- 2Universitat Politècnica de Catalunya, UPC/IEEC, and SMOS Barcelona Expert Center, Jordi Girona 08034, Barcelona, Spain
- 3Institut de Ciències del Mar, ICM/CSIC and SMOS Barcelona Expert Center, Pg. Marítim 37-49, 08003 Barcelona, Spain
Keywords: Soil moisture, SMOS, Validation, Remote Sensing, L-Band
Abstract. An increasing number of permanent soil moisture measurement networks are nowadays providing the means for validating new remotely sensed soil moisture estimates such as those provided by the ESA’s Soil Moisture and Ocean Salinity (SMOS) mission. Two types of in situ measurement networks can be found: small-scale (100–10000 km2), which provide multiple ground measurements within a single satellite footprint, and large-scale (>10000 km2), which contain a single point observation per satellite footprint. This work presents the results of a comprehensive spatial and temporal validation of a long-term (January, 2010 to June, 2014) dataset of SMOS-derived soil moisture estimates using two in situ networks within the Duero basin (Spain). The first one is the Soil Moisture Measurement Stations Network of the University of Salamanca (REMEDHUS), which has been extensively applied for validation of soil moisture remote sensing observations, including SMOS. REMEDHUS can be considered within the small-scale network group (1300 km2). The other network started from an existing meteorological network from the Castilla y León region, where soil moisture probes were incorporated in 2012. This network can be considered within the large-scale group (65000 km2). Results from comparison to in situ show that the new reprocessed L2 product (v5.51) improves the accuracy of former soil moisture retrievals, making them suitable for developing new L3 products. Validation based on comparisons between dense/sparse networks showed that temporal patterns on soil moisture are well reproduced, whereas spatial patterns are difficult to depict given the different spatial representativeness of ground and satellite observations.