Landsat 8 Thermal Infrared Sensor (TIRS) Stray Light Correction Algorithm Development and Assessment Using Coincident Terra MODIS Data

Abstract:

From first light in early 2013, the Thermal Infrared Sensor (TIRS) onboard Landsat 8 exhibited obvious indications that stray light was contaminating the thermal data collected from the instrument. Although traditional techniques were applied to calibrate the data, these efforts did not mitigate the non-uniform banding introduced by the stray light. The development of an operational technique to remove the effects of the stray light became a high priority to enhance the utility of the TIRS data, which in some cases had errors in absolute estimates of temperature in excess of 9 K. A potentially operational world-wide solution was developed that reduces non-uniform banding and absolute errors significantly. Finding “truth” data to train and validate the algorithm, however, was a major challenge. The scarcity of truth data represents a potential limiting factor for fully understanding the utility of the algorithm.

This presentation gives an overview of the stray light correction algorithm developed for the Thermal Infrared Sensor onboard Landsat 8. An emphasis is placed on the challenges in identifying truth data used in the training and analysis of the algorithm. Terra MODIS data from March 28^th, 2013 – March 30^th, 2013 is currently being used as truth data in this work. This time period is significant as Landsat 8 had not yet achieved its final orbit and for a brief few days, flew only minutes ahead of MODIS and through the center of its field-of-view. A presentation of how this data was used to train the algorithm, how it was used to assess the utility of the stray light correction algorithm versus the nominal calibration, and the need for additional truth data sources will be discussed.