Stray Light Corrections in GVAR Signal for GOES-East
Four IR Channel from imager with stray light contamination
There are periodic, and predictable, errors within the raw signal on the GOES satellites that arise when sunlight hits the Satellite so that it emits radiation that is detected by the sensor, or when satellite structures reflect energy towards the sensors. There errors usually arise when the Sun is close to being viewed directly by the sensor near “Satellite Midnightâ€. NOAA/NESDIS has recently (22 February 2012) implemented a series of corrections to mitigate these errors on the GOES-13 Imager. Not only does this increase the number of useable images, but it makes derived products – cloud top pressure, for example – more accurate. Parameters pertinent to the correction are included within Block 0 of the GVAR signal. In McIDAS, these bits relating to the stray light status are included as part of the AREA line prefix.
An example of the error in the raw (or un-corrected) signal is shown at top, with data from the four infrared channels (3.9 6.5, 10.7 and 13.3 micrometers) shown. Note the comparative magnitude of the extra radiation: it is far stronger and more widespread in the 3.9 micrometer image because the sun emits so much more radiation at that wavelength. (The Imager band most affected is the visible band (click here to see two contaminated — and uncorrected — and one clean image), the images above are at night). Options to deal with the stray light errors included: (1) Send all imagery , regardless of solar position/contamination, and let users decide; (2) Cancel images if the sun is within 6 degrees (currently) or 10 degrees of the frame boundary; (3) Scan away from the sun – for example, scan only the Northern Hemisphere if the solar contamination is in the Southern Hemisphere during the Spring eclipse season; and (4) Apply an L1B algorithmic correction to minimize stray light in the images prior to GVAR broadcast. Option (4) has been implemented for GOES-13. Currently option (3) is being implemented for GOES-15.
3.9 micrometer images showing stray light contamination (left) and corrected version (right)
The figure above shows a 3.9 micrometer image with a significant amount of stray light contamination in the southwest part of the image. The corrected version is also shown. Note that the contamination extends throughout the picture – brightness temperatures are too warm even in regions away from the large contamination (over the central United States, for example; compare the brightness temperatures of the cloud tops in the scene). The contaminated 3.9 micrometer data are corrected using two sources of information. For regions outside 6 degrees, the known amount of additional stray light is subtracted from the signal. If the sun is within 6 degrees of the pixel and the stray light signal is overwhelming, signals from the longer wavelength channels are used in combination with the 3.9 micrometer signal to estimate the true 3.9 micrometer signal. Linear relationships between the IR channels will vary with geographical location. Other thermal channel data that contain much less stray light are used in each of 256 geographic bins as input into multiple linear regressions relating 3.9 micrometer data (or 6.5 micrometer data) to 10.7 and 13.3 micrometer data. The hybrid image that results is uniformly cooler with a clear signal in a region formerly overwhelmed by stray light. The algorithm was developed by ITT and implemented by NOAA/NESDIS.
Current plans call for correcting the GOES-15 Imager during the fall 2012 eclipse season.
This ftp site contains more information. The GOES Eclipse schedule is here. This is the ‘White Paper’ on Stray Light. Finally, click here for more information on GVAR.
Finally, here is the notification from SSD that the Stray Light Correction was implemented.