IDENTIFICATION, ATTRIBUTION AND SCREENING OF RFI SOURCES USING NWP DEPARTURE STATISTICS

T. Scanlon¹, D. Duncan¹, A. Geer¹, N. Bormann¹

¹ECMWF

In recent years, ECMWF has been moving towards a full Earth System approach which incorporates information about the surface as well as the atmosphere. This approach benefits from the use of low frequency microwave channels as they can provide additional information about the land, ocean and ice surfaces. The recent inclusion of both the 6.9 and 10.65 GHz (C-and X- band) channels into the ECMWF Integrated Forecast System over open oceans (for Sea Surface Temperature retrievals) identified that there were several regions of the ocean severely affected by Radio Frequency (RFI) interference in these channels.

Using NWP background departures (the difference between the observation and the brightness temperature simulated from the background forecast), two approaches were followed to screen out these sources. The first method allowed the identification of potential sources by modelling the glint angle between theoretical broadcast satellites and the observing satellite. The second utilised the differences between two spectrally close channels.

In the current work, we aim to further understand the RFI signals seen at these lower frequency microwave channels in preparation for the upcoming ESA CIMR mission. The work will use the NWP background departure statistics from AMSR2 to design a system of checks which allow RFI to be identified. The method will include a combination of machine learning and understanding the physics behind such signals. A key focus will be determining how to separate RFI signals from other sources of error in an NWP system such as deficiencies in the forward model. Ultimately, this work will further our understanding of the sources of RFI signals and how they manifest within an NWP system, thereby allowing the effective screening of RFI contaminated data.

This work is funded by ESA under the DANTEX programme.