Assimilation of SMOS Brightness Temperatures at ECMWF: Impact on the Surface and Atmospheric Fields
Muñoz Sabater, Joaquín1; de Rosnay, Patricia1; Albergel, Clement1; Isaksen, Lars1; Balsamo, Gianpaolo1; Drusch, Matthias2; Fouilloux, Anne1

In the past few years the Soil Moisture and Ocean Salinity (SMOS) mission has confirmed to be able to provide a valuable source of information very sensitive to the variations of soil water content of the first few centimeters of soil. Land Surface Data Assimilation (LDAS) systems can make use of this information to improve soil moisture content predicted by a Land Surface Model (LSM). As Numerical Weather Prediction (NWP) centre, the European Centre for Medium-Range Weather Forecasts (ECMWF) has developed the capability to assimilate SMOS data, aiming at better initializing soil moisture used as input for the forecast run.

Soil moisture is analysed at ECMWF by using an advanced LDAS, consisting of a point-wise Extended Kalman Filter (EKF). SMOS brightness temperatures have recently been incorporated in this system. The main advantage of assimilating a level-1 product for NWP systems, as SMOS brightness temperatures, is the ability to use data as close as possible to the raw measurements, the quick availability of these data and the possibility of producing a Near Real Time (NRT) product based on their assimilation. This is specially suited for NWP systems. The objective at ECMWF is using the information provided by SMOS data to obtain more accurate soil moisture fields, and investigate the potential impact on the forecast skill.

In this paper several activities aiming to prepare the observations before assimilation will be addressed. In particular, the bias correction approach, based on matching the cumulative distribution function of SMOS observations to that obtained by running the ECMWF forward model operator over the same period of the observations, will be discussed. The impact of assimilating SMOS data on the analysis and forecast of surface and near-surface atmospheric fields, as well as in the synoptic forecast skill, will be assessed by evaluating several experiments that combine the assimilation of screen-level observations (i.e., 2 m air temperature and relative humidity) and SMOS data. In this paper, also the first-version of a new soil moisture product (SMOS-SM-v1.0), based on the assimilation of SMOS brightness temperatures, will be presented. Its quality will be validated against ground-based observations of the International Soil Moisture Network.