|Title||SMOS brightness temperature forward modelling, bias correction and long term monitoring at ECMWF|
|Series/Collection||ESA Contract Report|
|Type||ESA Contract Report|
|Authors||de Rosnay, P, Munoz-Sabater, J, Albergel, C, Isaksen, L|
This report presents the European Centre for Medium-Range Weather Forecasts (ECMWF) radiative transfer modelling activities conducted to use Soil Moisture and Ocean Salinity (SMOS) brightness temperature observations for Numerical Weather Forecast (NWP) applications. The Community Microwave Emission Modelling Platform (CMEM) is used as the ECMWF SMOS forward operator to simulate L-band brightness temperatures (TBs). In a first part, simulated brightness temperature are compared to the observed SMOS near real time reprocessed brightness temperature product for 2010-2011 for several configurations of CMEM using different set of parametrizations. We show that simulated brightness temperatures are more sensitive to the choice of vegetation opacity and soil roughness models than to the dielectric model. Best configurations of CMEM are shown to be those using the so-called Wigneron vegetation opacity model with the simple empirical Wigneron soil roughness model. The Wang and Schmugge and the Mironov soil dielectric models perform similarly and lead to better agreement with SMOS observations than the Dobson dielectric model. Based on this intercomparison the configuration of CMEM retained for ECMWF SMOS forward modelling activities is the one based the Wang and Schmugge dielectric model, the Wigneron simple roughness model and the Wigneron vegetation model. In a second part, this paper presents the SMOS brightness temperature bias correction developed and used at ECMWF. It is a monthly Cumulative Distribution Function bias correction based on SMOS and ECMWF re-analysis-based brightness temperatures for the period from 1 January 2010 to 31 December 2013. Results show that it efficiently corrects for systematic differences between observations and model, with global root mean square differences (RMSD) and global mean bias for 2010-2013 for 30, ?40 ?, 50 ?incidence angles decreasing from 16.7 K and -2.1 K before bias correction to 7.91K and 0.0016 K after bias correction, respectively. The monthly approach allows to correct for seasonal cycles systematic differences, with correlation values improved from 0.56 before bias correction and 0.62 after bias correction. Residual differences remaining after bias correction correspond to random differences between the model and observations which provide relevant information for monitoring and data assimilation purposes. Finally, in a third part long term monitoring of SMOS brightness temperature monitoring is presented covering a 7-year period 2010-2016 at both polarisations, at 40 degrees incidence angle. RMSD, correlation and anomaly correlation statistics show that SMOS and ECMWF reanalysis-based brightness temperature agreement steadily improves between 2010 and 2016, indicating improvement of SMOS products quality through the SMOS lifetime.