Decoupling Flat-fielding and Non-linearity Correction of a Pushbroom Radiometer – Analysis of Landsat 9 Operational Land Imager-2 Prelaunch Test Data

One of the goals in any calibration effort for pushbroom imaging radiometers is to flat field the instrument data across the full dynamic range. A challenge when approaching levels of 1% and lower in a 14 bit system is that the non-uniformity and non-linearity become a coupled variable set. The Landsat-9 Operational Land Imager-2 (OLI2) prelaunch radiometric calibrations conducted at Ball Aerospace utilized spectral sources, large integrating spheres and rotation stages. These tools, combined with unique collects such as the integration time sweeps and yaw collects at multiple illumination levels provided the basis for improvements in the calibration of both the full field of view and the full dynamic range for all of spectral bands. In an integration time sweep, a constant source level is observed while varying the detectors’ integration time; in a yaw collect the instrument is rotated so that each detector views the same part of the illumination source. While previously we reported on the characterization of the calibration source used, this presentation will focus on how the multiple datasets were utilized to arrive at flat fielding and the non-linearity corrections. The method used enables a reduction in the uncertainty of the uniformity correction throughout the dynamic range. The uncertainty in the source non-uniformity, the source stability and the instrument under test stability are the three limiting factors. The data sources, the types of non-linearities, the differences between integration time sweeps and radiance collects, the representations of the non-linearity and the validation of the relative gain corrections throughout the dynamic range will be presented.