Contributor: Brian Markham

The Landsat Calibration Working Group held their most recent semi-annual meeting at the USGS EROS facility in Sioux Falls, SD on December 3 and 4, 2008.  USGS/EROS Image Assessment System (IAS), NASA/GSFC LPSO, South Dakota State University, Rochester Institute of Technology (RIT), University of Arizona, and Ball Aerospace OLI calibration personnel attended the meeting.  Significant progress in a number of areas of Landsat calibration was reported at the meeting:

  • Continued stability is observed in the radiometric calibration of the Landsat-7 ETM+ reflective bands with no required update to the calibration.
  • Continued good geometric performance is observed for the Landsat-7 ETM+, though the switch to bumper mode and recent alignment variations have degraded performance slightly.
  • A good collection of “hot” Salton Sea scenes and matching data from Simon Hook’s site extended the range of water temperatures available for thermal band calibration. These datasets indicate a small gain error in both Landsat-5 TM and Landsat-7 ETM+ thermal band data which results in a circa 1°C error at 30°C, though within the noise at typical water temperatures (between 4°C and 20°C). Updates to the calibration parameter files will be developed to adjust for these errors.
  • The extension of the IAS and Landsat Product Generating System (LPGS) to processing Landsat-5 TM data is complete and the database is beginning to be populated.  Initial examination of the database shows the value of these results over the sparse results previously used to develop some of L5 TM models, e.g., the outgassing model. Challenges remain in terms of having sufficient resources to take advantage of the new capabilities.
  • Indications are that the extension of the LPGS processing to Landsat-5 TM beginning December 8, 2008 will result in reduced banding artifacts in the distributed data products.
  • The Landsat-5 Thermal Band vicarious calibration history was extended back to launch in 1984 by RIT using Great Lakes and Ocean Buoys.  With sufficient screening the scatter (circa 0.5°C) in the process is only slightly worse than other vicarious methods.  Results indicate an extremely stable data product over the full mission with circa 1°C RMSE; a time dependent linear correction could reduce this to circa 0.5°C RMSE.
  • Another iteration to the Landsat-4 TM lifetime gain model was presented by SDSU.  A sparseness of data limits the ability to achieve high confidence in this result.  A decision was made to accept a flat line for the calibration of all but band 1 of the Landsat-4 TM; a linear function of time was selected for band 1.  A look up table will be developed to implement this change in the LPGS when it goes on-line for Landsat-4 TM data in early 2009.
  • The radiometric calibration histories for all five MSS sensors were developed by SDSU including cross calibrations between them.  Results indicate nearly constant calibration coefficients for each sensor post calibration using the on-board calibration system.  A methodology was developed to make the data consistent between all five systems to about 5% using simple linear equations.