ECMWF Final Report on SMOS brightness temperature activities over land: Monitoring and Data Assimilation

This paper is the final report of the SMOS project activities conducted at ECMWF between 2007 and 2018 to use SMOS brightness temperature (TB) observations in the ECMWF Integrated Forecasting System (IFS) for monitoring and data assimilation purposes. These activities were conducted in two phases. The first one, (contract 20244/07/I-LG) focused on the implementation of the SMOS TB data in the IFS and the development of the SMOS TB monitoring system. It included the development of the SMOS forward model CMEM (the Community Microwave Emission Modelling platform) as well as a number of technical tasks such as the SMOS BUFR specification, the SMOS Observation Data Base, and the set-up of the SMOS monitoring pages. The second phase (contract No 4000101703/10/NL/FF/fk) was dedicated to SMOS TB data assimilation. It included the implementation of the SMOS TB data in the Simplified Extended Kalman Filter data assimilation system, further developments of the forward model CMEM, bias correction, long term monitoring, tuning and specification of the background and observation errors specification, and evaluation and SMOS TB data assimilation in the IFS. Activities conducted in 2016-2018 also included the SMOS neural network developments and offline data assimilation as well as ocean activities (sea ice and wind speed), which are reported separately and therefore not included in this paper. Both contracts included technical support activities like the definition of the SMOS DPGS interface and Interface Control Document (ICD) and support to ESA with ECMWF products and changes related to new IFS cycles, resolution and format changes. Long term monitoring results over land showed that SMOS and ECMWF reanalysis-based brightness temperature agreement steadily improved between 2010 and 2016, indicating improvement of SMOS products quality through the SMOS lifetime. Data assimilation results showed that the soil moisture state benefits from the direct assimilation of SMOS TB, especially in better representing the temporal variations of soil moisture in time. The forecasting skills of low level atmospheric variables remains mainly driven by the screen level observations. Despite the clear benefits on the soil state, remote sensing data needs to be used with screen level variables to add value on the state of the atmosphere, pointing to inconsistencies in the physical coupling between the land and near-surface components of the ECMWF Earth system. These activities demonstrated the relevance of the SMOS observations for numerical weather prediction modelling and assimilation developments.