Evaluation of remotely sensed surface soil moisture and an energy balance model driven by LST data with Italian in situ data

Near-surface soil moisture is a key variable for the description of many hydrological and climatic processes. This variable governs the partitioning of mass and energy fluxes between land surface and the atmosphere, thereby playing an important role for many scientific and operational applications such as flood forecasting, landslide prediction, numerical weather prediction, early drought prediction, climate modelling and water management. Particularly in operational hydrology a correct estimation of soil moisture improves the forecast of the rainfall-runoff response of catchments and consequently improves flood predictions. Soil moisture content can be evaluated through several methods such as classical point measurement (TDR, gravimetric, neutron probes, capacitance probe), remote sensing (passive or active microwave sensors) and hydrological models predictions. Ground-based measurements are direct but very localized and limited in coverage; satellite data provide useful but indirect large-scale observations of surface soil moisture and could have problems in some environments (mountain regions, vegetated surfaces, etc..); model predictions provide a more regional perspective but rely on many parameterizations and approximations aimed to a more efficient representation of the physics of the process. The work considers two different types of satellite-derived data: the HSAF project surface soil moisture (SSM) product, derived from ASCAT scatterometer; and an evaporative fraction index, related to soil moisture content, derived by the ACHAB (Assimilation Code for Heat and moisture Balance) model, an operative energy balance model that assimilates Land Surface Temperatures derived from MSG. These two estimates are compared and validated using in situ soil moisture data available in different Italian regions with the aim of obtaining reliable soil moisture estimates to be used for operational flood forecast. Test has been performed using the HSAF and ACHAB data both as the original estimates and as filtered series applying a linear regression correction and a CDF matching in the way that both CDFs match the CDF of in situ measurements. Further analysis were carried out applying a exponential filter (the so called Soil Water Index) that takes into account the differences between the soil moisture in the uppermost soil layer layer (2-3 cm ) observed by satellite data and the soil moisture in the layer observed by in situ measurements (10-30 cm).