6B.1 SMOS Neural Network Soil Moisture Data Assimilation (Invited Presentation)

Tuesday, 14 January 2020: 10:30 AM
Nemesio Rodríguez-Fernández, CNRS, Toulouse, France; CESBIO, Toulouse, France; Centre d'Etudes Spatiales de la Biosphère (CESBIO), Toulouse, France; and P. de Rosnay, F. Aires, C. Albergel, M. Drusch, Y. Kerr, C. Prigent, S. Mecklenburg, J. Muñoz Sabater, and P. Richaume

In this presentation we will summarize several years of research on the use of neural networks to retrieve soil moisture from passive microwave radiometers such as the European Space Agency (ESA) Soil Moisture and Ocean Salinity (SMOS) satellite. Neural networks have been found to be an efficient alternative to radiation transfer modelling leading to a new ESA near real time (NRT) soil moisture product for operational applications in hydrology and meteorology. We will also describe how to use a neural network as offline observation operator for efficient data assimilation (DA). This approach was tested using the offline surface-only Land Data Assimilation System (so-LDAS) at the European Centre for Medium Range Weather Forecasts (ECMWF) and using the analyzed soil moisture fields to initialize atmospheric forecasts (Rodriguez-Fernandez et al. 2019, Remote Sensing, 11). The positive results of these experiments led to the operational data assimilation of a dedicated SMOS neural network soil moisture dataset in the ECMWF Integrated Forecast System (IFS) starting in cycle 46r1.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner