Radio Frequency Interference

Introduction to RFI

Ocean reflected radio frequency interference (RFI) is arguably the fastest growing source of errors in passive microwave measurements.  RFI is largely caused by media broadcasting (including radio and television)  from commercial satellites in geostationary orbits.  The geostationary broadcasting signal reflects off the Earth's ocean surface into the microwave instrument's antenna causing an alteration in the measurement. Ground-based instrumentation in the micowave range can also produce RFI.  In this case, the signal is broadcast directly into the atmosphere and is picked up by the radiometer antenna.    It is relatively easy to identify and flag RFI from large sources, but more difficult to do so from smaller sources.  In addition, the spatial and temporal variability of RFI requires constant monitoring for new or stronger sources.  If one does not remove RFI from ocean products, spurious trends in data products may result.  Correcting microwave data for RFI contamination requires automated detection and removal.  This is not a trivial task.

Here we give examples of RFI on microwave data, discuss our methods for RFI removal and provide a list of geostationary and ground-based RFI sources that we know of.

Identifying RFI in Microwave Data

The microwave data errors due to RFI from geostationary broadcast sources are dependent on several characteristics, including:  communication broadcast frequency, signal power and direction, microwave instrument bandwidth, signal glint angle, and ocean surface roughness. The observation bandwidths of microwave instrument channels are typically wider than the protected bands allocated for microwave remote sensing. Thus, microwave instruments can receive RFI from nearby approved communication frequency bands. 

AMSR-E and WindSat are most affected by RFI, while SSM/I and TMI both appear to be relatively unaffected. This is likely due to the lower frequency channels of AMSR-E and WindSat (10.7 GHz and 18.7 GHz measurement channels) which are sensitive to frequencies used extensively for media broadcasting.  WindSat has more significant RFI than AMSR-E due to  a wider observation bandwidth. Observing more bandwidth tends to yield less radiometer measurement noise, but also leads to more interference from frequencies further from the channel’s center observation frequency.

Signal power and direction are also important factors affecting RFI. Satellite media broadcasts appear to direct most of their signal power very carefully to specific markets. These powerful signals can result in large RFI errors within certain regions. To serve smaller but more geographically dispersed markets, media satellites also broadcast wide, low power beams to cover much larger, less populated regions. These lower power beams produce more subtle RFI effects that are difficult to detect and remove. Assuming the Earth observation point is within the footprint of the geostationary broadcast, the magnitude of RFI is highly dependent on the glint angle, or how close the microwave observation reflection vector comes to pointing at the RFI source.

We detect regions of RFI by differencing AMSR-E SSTs derived using all microwave channels (6.9 GHz – 36.5 GHz) from those SSTs derived without using the  6.9 GHz channel (10.7 GHz – 36.5 GHz), as well as by differencing wind speeds derived using all microwave channels (10.7 GHz – 36.5 GHz) from those wind speeds derived without the 10.7 GHz (18.7 GHz – 36.5 GHz).  This difference method can be used for strong RFI sources.

Geostationary examples

These two examples demonstrate the type of errors in microwave data caused by RFI.

AMSRE 10.7 GHz  RFI

The following figure shows RFI in AMSR-E sea surface temperature descending passes near Europe.  In this example, you can see the errors in SST increasing yearly from 2002 when AMSR-E on Aqua was launched.  The errors increased in intensity and coverage over the 6 years shown here.

     

AMSRE 18.65 GHz RFI

AMSR-E ocean products derived from descending 18.7 GHz observations are potentially impacted along all U.S. coastal waters from September 2007 forward. The location, timing, and frequency of the interference are highly consistent with HDTV broadcasting activities from DirecTV satellite 10 in geosynchronous orbit at 102.8 west longitude.  DirecTV-10 launched July 7, 2007, and began broadcasting operationally after about 2 months of testing (https://en.wikipedia.org/wiki/DirecTV-10). The first significant DirecTV interference we detected in AMSR-E data occurred on September 15, 2007. The locations of interference are consistent with geosynchronous satellite signals reflecting off the ocean surface into the AMSR-E field of view. DirecTV-10 is reportedly broadcasting at 18.648 GHz, well within AMSR-E bandwidth range centered on 18.7 GHz.

Having similar broadcasting capabilities, DirecTV-11 (https://en.wikipedia.org/wiki/DirecTV-11) launched March 19, 2008 and began broadcasting on July 31, 2008, from geosynchronous orbit at 99.2 West Longitude. DirecTV-11 also impacts AMSR-E at 18.7 GHz. Thsi likely results in an increase the amplitude of the RFI more than the spatial extent, as the 2 broadcasting satellites are only 3.6 degrees apart in longitude.

In the following example, AMSR-E surface wind retrievals from 3 descending passes from April 10, 2008 are shown.  The wind speeds are affected by geostationary RFI. The position of each reflection within the swath indicates a geosynchronous source is the likely cause for the errors. In the Atlantic and Gulf of Mexico, the RFI contaminated 18.7 GHz values are mostly out of bounds and no wind retrievals are made. Erroneously low wind speeds remain on the edge of the Atlantic data exclusion. In the Pacific, the RFI-contaminated 18.7 GHz observations were realistic enough to retrieve spuriously low wind speeds.

In the following figure, we demonstrate how the RFI off the western US coast can appear very much like rain in the microwave observations.  Here, AMSR-E 18.7 GHz brightness temperatures are shown on the left for vertical and horizontal polarizations.  The RFI results in somewhat realistic looking, but result in erroneous rain retrievals as shown on the right side of the figure.

 

The DirecTV RFI problem was first detected in WindSat data by NRL. The 18.7 GHz channels on WindSat have significantly wider bandwidth than AMSR-E, thus the RFI contamination is more substantial and obvious. WindSat has been impacted by the DirecTV nationwide beams, which broadcast at frequencies below the sensitivity of AMSR-E at 18.7 GHz. The DirecTV spot beams, however, are using frequencies that impact AMSR-E. Initially, it was hoped that the spot beams would limit the spatial extent of impact on AMSR-E. Further investigation revealed that DirecTV spot beams cover nearly the entire U.S. coastline. Spot beams are used to beam different local programming to different local markets, thereby allowing reuse of the same set of frequencies. As the U.S. coast is well populated, the constellation of spot beams provides coastal coverage similar to the nationwide beams. Spot beams also serve Alaska and Hawaii, and subtle RFI effects can be seen in those locations.

Ground-based examples

A ground RFI source off the Netherland coast has concurrently increased power to become more prominent as seen by the small distinctive dot forming over the years. The ground source produces SST errors of opposite sign compared to the geostationary RFI in the region. In this small region, two prominent sources of RFI error tend to cancel each other, potentially complicating detection and removal. The striping visible in Figure 6 is not due to any cross-swath problem with the SSTs or wind speeds, but is due to the glint angle geometry which results in a heavily stripped glint angle pattern caused by AMSR-E’s ground track repeat pattern every 233 orbits.

Removing RFI Errors

Glint angles and broadcast footprints are together highly predictive of potential RFI bias. Therefore, to remove RFI errors from the AMSR-E SST and wind products we calculate the signal glint angles using the longitude of the geostationary orbits. These glint angles, together with analysis of broadcast footprints, are used to remove retrievals with high probability of RFI error.

AMSR-E RFI Flag: Geostationary Satellite Glint Angle

RFI intensity is affected by glint angle, geographic location, and surface roughness.

Glint Angle: The magnitude of RFI from these sources is highly correlated with how directly the broadcast signals reflect off of Earth's surface into AMSR-E's field of view. We find the Geostationary Satellite Glint Angle by computing the scalar angle difference between the reflection vector (boresight, or observation vector reflected off the Earth observation point) and the vector from the Earth observation point to the geostationary orbit position. A glint angle of 0° is a direct reflection; glint angles above 20° are unlikely to have significant RFI.

Geographic Location: The broadcasting satellites direct their power very specifically to serve markets. Outside of these geographic areas, low glint angles do not correlate with intense RFI. Ascending passes are not affected at all. On descending passes, ocean RFI is most intense near coastal areas.

Surface Roughness: RFI is also correlated with surface roughness of the ocean. Low wind, smooth waters yield intense RFI confined to lower glint angles. Rough, windy waters result in RFI of more moderate intensity, but over a larger area including higher glint angles. Unfortunately, the RFI makes wind retrievals unreliable, complicating any attempt to utilize this correlation. We have not studied the impact over land.

AMSR-E Geostationary Satellite Glint Angle Angles less than 20° are shown in red.

Ongoing List of RFI Sources

Geostationary RFI Sources

More detailed information and coverage maps for each of these geostationary RFI sources can be found at www.satbeams.com/satellites or www.lyngsat.com.    

Spacecraft	Longitude	Launch date	Areas Affected	Channel
DirecTV-10	102.8°W (257.2°E)	2007-07-06	North America	18.7 GHz
DirecTV-12	102.8°W (257.2°E)	2009-12-28	North America	18.7 GHz
DirecTV-11	99.2°W (260.8°E)	2008-03-19	North America	18.7 GHz
Intelsat 3R (Sky Brazil)	43.0°W (317.0°E)	1996-01-12	Brazil	10.65 GHz
Intelsat 11 (Sky Brazil)	43.0°W (317.0°E)	2007-10-05	Brazil	10.65 GHz
Hispasat 1E	30.0°W (330.0°E)	2010-12-29	Europe	10.730 GHz
Atlantic Bird 4	7.2°W (352.8°E)	1998-02-27	Middle East	10.65 GHz
Eutelsat W3A	7.0°E	2004-03-16	Europe	10.65 GHz
Hot Bird 7A -> Eurobird 9A	13.0 -> 9.0°E	2006-03-11	Europe	10.65 GHz
Eutelsat W2A	10.0°E	2009-04-03	Europe	10.65 GHz
Hot Bird 6	13.0°E	2002-08-21	Europe	10.65 GHz
Hot Bird 8	13.0°E	2006-08-05	Europe	10.65 GHz
Astra 1KR	19.2°E	2006-04-20	Europe	10.65 GHz
Astra 1E	19.2 -> 23.5°E	1995-10-19	Europe	10.65 GHz
Astra 2D	28.2°E	2000-12-19	Europe	10.65 GHz
Astra 2C	28.2 -> 31.5°E	2001-06-19	Europe	10.65 GHz

 

Ground-based RFI Sources

Region Affected	Frequency	Period of Interference	Effect on Data
Ascension Island	6.9 GHz	pre-2002 to present	lower SSTs
Gulf of Aden	10.7 GHz	Mar 2009 to present	lower SSTs, increased winds
Coastal Netherlands Coastal Norway	6.9 GHz	2004 to present	higher SSTs
coast off Mumbai	6.9 GHz	2003 - present	higher SSTs

More Information

Presentations showing more examples of RFI

RFI Detection and Mitigation for AMSR-E Ocean Retrievals, a presentation given at the 13th Specialist Meeting on Microwave Radiometry and Remote Sensing of the Environment, Pasadena, CA, March 24-27, 2014

RFI 2.0 and Adaptive Algorithms presentation given at the AMSR Science Team Meeting, Portland, OR, September 2012

RFI in Passive Remote Sensing of Ocean Parameters presentation given at the IGARRS GEOSS Workshop XXXVII – Data Quality and Radio Spectrum Allocation, July 2010

AMSR-E Geostationary RFI presentation given at the Joint AMSR Science Team Meeting, Arlington, VA, June 2009