Recent Improvements in AMSR2 Ground-Based RFI Filtering

Abstract:

Passive satellite radiometer measurements in the microwave frequencies (6-89 GHz) are useful in providing geophysical retrievals of sea surface temperature (SST), atmospheric water vapor, wind speed, rain rate, and more. However, radio frequency interference (RFI) is one of the fastest growing sources of error in these retrievals. RFI can originate from broadcasting satellites, as well as from ground-based instrumentation that makes use of the microwave range.

The microwave channel bandwidths used by passive satellite radiometers are often wider than the protected bands allocated for this type of remote sensing, a common practice in microwave radiometer design used to reduce the effect of instrument noise in the observed signal. However, broad channel bandwidths allow greater opportunity for RFI to affect these observations and retrievals.

For ground-based RFI, a signal is broadcast directly into the atmosphere which may interfere with the radiometer – its antenna, cold mirror, hot load or the internal workings of the radiometer itself. It is relatively easy to identify and flag RFI from large sources, but more difficult to do so from small, sporadic sources. Ground-based RFI has high spatial and temporal variability, requiring constant, automated detection and removal to avoid spurious trends leaching into the geophysical retrievals.

Ascension Island in the South Atlantic Ocean has been one of these notorious ground-based RFI sources, affecting many microwave radiometers, including the AMSR2 radiometer onboard JAXA’s GCOM-W1 satellite. Ascension Island RFI mainly affects AMSR2’s lower frequency channels (6.9, 7.3, and 10.65 GHz) over a broad spatial region in the South Atlantic Ocean, which makes it challenging to detect and flag this RFI using conventional channel and geophysical retrieval differencing techniques.

The authors have developed a new method of using the radiometer’s earth counts and hot counts, for the affected channels, to detect an Ascension Island RFI event and flag the data efficiently and accurately, thereby reducing false detections and optimizing retrieval quality and data preservation.