Landsat-8 data processing evolution

Shortly after Landsat-8 launched in February 2013, the U.S. Geological Survey (USGS) Earth Resources Observation and Science (EROS) Center began creating radiometrically and geometrically corrected products. In order to provide these products as soon as possible, the Landsat Product Generation System (LPGS) was developed based on instrument designs and testing prior to launch. While every effort was made to ensure the LPGS produces highly accurate products, some aspects of the sensors are difficult to characterize during testing on the ground. Examples of these characteristics include differences between individual detectors that make up the focal plane array, and the way detectors view radiometric targets in preflight testing versus the way they view the Earth on orbit, and the accuracy of the measurements made on the ground. Once in orbit, more accurate measurements of these sensor characteristics were made that improved processing parameters, resulting in improved quality of the final imagery. This paper reviews the changes that have occurred to the processing of Landsat-8 data products which include parameter changes as well as some modifications to the processing system itself. These changes include: improved linearization of the data, both to parameters and the algorithm used for linearizing the data; improved radiance and reflectance conversion coefficients; individual detector coefficients to improve uniformity; and geometric alignment coefficients to improve the geometric accuracy. These improvements lead to a reprocessing campaign that occurred in early in 2014 that replaced all prior data with improved products.