Visible calibration of the Imager and Sounder uses an integrating sphere that is also used for the AVHRR instrument on the NOAA polar orbiters. Traceablility to absolute standards is unknown to me. There is no on-board source, active or passive, for visible calibration in orbit. Relative response is specified as S/N of 150/1 from 0.5% to 100% albedo, and the instruments achieve approximately 250/1 S/N in pre-launch testing. The 8 individual silicon detectors are significantly more stable than the previous GOES' generation of PMTs, but they still require histogram normalization to compensate for faint striping. NOAA expects to adjust the relative visible response a few times per year.
Pre-launch calibration to determine non-linearity in the IR detectors is performed using the vendor-built external target in thermal-vac. On-orbit infrared calibration of the Imager and Sounder uses an internal blackbody and frequent looks to space. Neither target is traceable to absolute standards. Absolute accuracy is specified to be 1 K, but that appears to be unachievable. Determination of the non-linearity is uncertain to within uncertainties in the test set-up (approximately 1%). Space is used for a DC-restore once every 36.6 seconds, The internal target is maintained near 320 K, and used to determine gain approximately once every 15 minutes. Gain is determined between frames on the Imager, during frames on the Sounder. Trends in gain are monitored by NOAA. Detector-to-detector IR striping is noticeable in the noisier channels -- a fraction of the single-sample noise. NOAA plans to model gain-changes between detectors and during the diurnal cycle to reduce relative error.
Regular de-icing of the detector windows on the Sounder is required to maintain acceptable performance. On the Sounder, the contaminant is unknown, with a spectrum dominated by absorption in the 9,7 micron channel. On the Imager, water-icing of detector windows was observed in pre-launch testing, but it does not occur on-orbit after the set-temperature for the window was raised.
Both instruments suffer from angle-dependent emissivity of the glass-overcoated scan mirror. The SiO2 reststrahlung effect peaks around 8 microns, and is most noticeable in the middle infrared (6.7 to 12 microns for the Imager, 6 to 14 microns for the Sounder), where the apparent brightness of outer space varies along the east-west scan. Fortunately, the effects are predictable and measurable as a function of position and channel, and the angle-dependence is being corrected in real-time software before being broadcast to the users. The effect is only a few percent for the ±5 degree oscillation of the scan mirror about its normal degree position. The calibration look to the internal black body is also performed at 45 degrees. The use of the east or west-most scan positions for the space-look DC-restore is significant, and NOAA looks to the side farthest from the Sun.
Both instruments suffer from stray light and erratic calibration near sub-satellite midnight, when sunlight streams into the face of the telescope. NOAA may "flywheel through" this period, rather than use obviously corrupted gain factors.
NOAA performs on-orbit validation of the sensitivity and stability of the GOES sensors using the same radiometric under flights made for AVHRR.
GOES-EAST and GOES-WEST should routinely scan the same regions under the same solar illumination during the duration of the years of operations to cross-characterize the two platforms.
Each GOES-I/M Imager and Sounder on each satellite should routinely scan the same region of the Earth simultaneously to cross-characterize the instruments on the same platform.
GOES-EAST and GOES-WEST should routinely scan the same regions being observed by the EOS platforms to cross-characterize all similar radiometers.
Can GOES achieve 2% uncertainty in the pre-launch absolute calibration of the visible channels? Can relative changes between detectors and on the optical surfaces be controlled? Can 1 K absolute calibration be actually achieved in the infrared?