Performance of the proposed stray light correction algorithm for the Thermal Infrared Sensor (TIRS) onboard Landsat 8

摘要

Imagery from the Landsat 8 Thermal Infrared Sensor (TIRS) have exhibited scene-dependent non-uniform banding and absolute calibration artifacts since the instrument began operation in 2013. These artifacts have been attributed to a stray light effect in which radiance from outside the nominal field-of-view of the instrument enters the optical system and adds a non-uniform signal to the focal plane detectors. A major effort was launched to characterize the stray light sources and derive an operational software correction that could easily be applied to the ground processing system. The proposed solution relies on a regression analysis in which TIRS scene imagery is used in combination with a detailed optical model to calculate the extra stray light signal on the detectors. The predicted signal is then subtracted from the scene data to remove the stray light artifacts. The resulting imagery from the correction algorithm displays a vast improvement in both banding and absolute error over the current TIRS product. The algorithm has the added benefit of being able to run in 'real time' with no additional data needed. Comparisons to MODIS thermal imagery have demonstrated high performance for scenes all over the world and over different material types and temperatures. A summary of these validation studies will be discussed here.