In the fall of 2003, NOAA/NESDIS proposes to make a change in the calibration of the GOES Imagers to correct for the deleterious effects of solar heating on the calibration during the eight hours surrounding local midnight (0500 UT for GOES-12, 0900 UT for GOES-10). At certain times of the year (see below), data users may notice changes in measured radiances reaching a maximum of approximately 1K (for a scene at 300K) within an hour of midnight and decreasing with increasing time before and after midnight. A brief description of the midnight calibration errors and the correction algorithm follows.

The midnight blackbody calibration error affects the calibration slopes that are computed in real time on orbit. The slope is defined as the coefficient of the linear term in the radiometric calibration equation, the equation relating scene radiance to sensor output counts. It is computed once every 30 minutes from data collected when the Imager views space and its internal blackbody. At certain times of the year, we find that the computed slopes may be in error by as much as 3% during the period around local satellite midnight. The effect is most pronounced in channels 2 and 3 (3.9 and 6.7 μm, resp.) of the Imager, but it is also noticed to a lesser extent in the channels at longer wavelengths. For GOES-8, which was operational at 75W longitude from 1994 until April 1, 2003, the effect occurred primarily in the months from April through October. For GOES-10, operational at 135W longitude since 1998, the effect occurs almost year-round. For GOES-12, operational at 75W longitude since April 1, 2003, the effect has been present continuously so far, but at the time of this writing (July 2003), we do not yet know if it occurs year round or only during part of the year. An example of the effect is shown as the upward spikes in the time series of calibration slopes displayed in this figure:

Such slope errors artificially depress measurements of brightness temperature of the Earth by as much as 1K. This effect is believed to be caused by extraneous radiation, probably from hot components in the imager's scan-mirror cavity, reaching the Imager's detectors during its calibration cycle.

The correction algorithm is based on the observation that when this effect is absent, there is a very high correlation between the calibration slope and the temperature of several optics components, particularly the telescope's primary mirror. Accordingly, when the effect occurs, we replace the bad slope values by estimates computed by regression on the primary-mirror temperature. The figure shows the original slopes that have been replaced (x's inside squares) by the regression estimates (diamonds), which we believe are better estimates of the instrument's true slopes.

An accompanying report presents details of the correction algorithm.

For even more information, see Johnson, R.X. and M.P. Weinreb, GOES-8 Imager Midnight Effects and Slope Correction, in GOES-8 and Beyond, Edward R. Washwell, editor, Proc. SPIE 2812, pp 596-607 (1994).