Combining remote sensing and in situ soil moisture data for the application and validation of a distributed water balance model (HIDROMORE)

An application of the FAO56 approach to calculate actual evapotranspiration (AET) and soil moisture is reported, implemented by means of the HIDROMORE computerized tool, which performs spatially distributed calculations of hydrological parameters at watershed scale. The paper describes the application and validation of the model over 1 year in an area located in the central sector of the Duero Basin (Spain), where there is a network of 23 stations for continuous measurement of soil moisture (REMEDHUS; Soil Moisture Measurement Stations Network) distributed over an area of around 1300 km2. The application integrated a series of Landsat 7 ETM+ images of 2002, from which the NDVI series (Normalized Difference Vegetation Index) and the map of land covers/uses were derived. Validation consisted of the use of the REMEDHUS soil moisture series and their comparison with the series resulting from the application. Two simulations were performed, with soil parameters values at the surface (0–5 cm depth) and at the mean of the profile scale (0–100 cm depth). The behaviour of the simulated soil moisture was described by means of its correlation with the measured soil moisture (determination coefficient, R2 = 0.67 for the surface values and 0.81 for the mean profile values), and the Root Mean Square Error (RMSE), resulting in a range of it for the 23 stations between 0.010 and 0.061 cm3 cm−3. The application afforded an underestimation of the soil moisture content, which suggests the need for a redefinition of the limits of the plant available water used in the calculation. The results showed that HIDROMORE is an efficient tool for the characterization of hydrological parameters at global scale in the study zone. The combination of the FAO56 methodology and remote sensing techniques was efficient in the spatially distributed simulation of soil moisture.

Research highlights▶ Validation of simulations by means of observed soil moisture at the field network. ▶ Automatic integration of remote sensing products data in the model. ▶ Discussion of the critical issues in the model: limits of water content, root depth, Kcb–NDVI relationships.