Mitigating the GOES-17 ABI Thermal Anomaly Using Predictive Calibration

The GOES-17 Advanced Baseline Imager (ABI) has an anomaly with its on-board cooling system that prevents it from maintaining its Focal Plane Modules (FPMs) at cold, optimal temperatures. Because of this, during certain times of the year the FPMs and their detectors warm and cool throughout the day. Changing the detectors' temperature changes their response to incoming radiance, which leads to calibration errors over time and degrades the imagery. Numerous mitigation strategies have been implemented to reduce the solar insolation on the instrument and to mitigate image degradation, including semi-annual yaw flips and changing the integration time of the detectors twice daily. These and other mitigations all work with the baseline calibration algorithms currently in place on the GOES-R Ground System. In an attempt to reduce the image degradation even further, the ABI vendor designed a new calibration scheme that predicts key parameters forward in time to account for the drifting FPM temperatures. These parameters, the linear gain term and dark current scene, are nominally updated on orbit every 5 minutes and 30 seconds, respectively. However, even at these relatively short cadences the detectors can change temperature, thereby rendering the parameters invalid for accurate calibration. By projecting these parameters forward in time the radiometric bias is reduced and image quality improves. This Predictive Calibration modification was deployed to operations on July 25, 2019, following several months of extensive testing and optimization by the GOES-R science teams. During this time several parameters and thresholds were tuned to ensure Predictive Calibration was turning on and off at the optimal times. Since going into operations users have seen noticeable improvement to the imagery and its calibration. This paper will discuss the fundamental assumptions behind the baseline equations and highlight the changes introduced by Predictive Calibration. Results will show the improvements to the calibration of the operational L1b products and reduction in image degradation.