Landsat-7 Long Term Acquisition Plan Radiometry—Evolution over Time

Brian Markham , NASA Goddard Space Flight Center

Samuel Goward, University of Maryland College Park

Terry Arvidson, Lockheed Martin

Julia Barsi, Science Systems & Applications, Inc.

Pat Scaramuzza, SAIC

 

Photogrammetric Engineering & Remote Sensing, vol. 72, no. 10, pp: 1129–1135.

The Landsat-7 Enhanced Thematic Mapper Plus (ETM+) instrument has two gain states for each spectral band. One gain state has higher saturation radiance, allowing imaging brighter targets without saturation; the other has higher precision, allowing greater discrimination between similar targets. For any given acquisition, only data from one of the two gain states is downlinked for any given spectral band, with the exception of the thermal band. Both gain states of the thermal band are downlinked all the time. The Landsat-7 Long-Term Acquisition Plan team has gone through several iterations on acquisition strategies to effectively use the two gain states. At launch, the plan was based on setting the gains for each scene to maximize the entropy in that scene. One unintended consequence of this plan was that saturation was occurring in densely vegetated targets (high reflectance in ETM+ band 4) and desert targets (high reflectance in all bands). To reduce this saturation, a revised strategy was implemented based on a simple land-cover classification scheme and sun angle thresholds. This procedure did reduce saturation, but resulted in gain changes occurring repeatedly within the same scenes. As the gain changes cause a small loss of data and difficulties for some of the ground processing systems, a procedure was set up to shift the location of the gain changes to the nearest predicted cloudy scenes, where they would cause the least problems. Although the gain setting strategy has improved with time, the results are still not totally satisfactory as gain changes still impact some scenes and saturation still occurs, particularly in ephemerally snow-covered regions. A primary conclusion of this experience with variable gain on Landsat 7 is that such an approach should not be employed on global monitoring missions such as Landsat in the future.