Calibration of the Satellite Only Solution

The entire satellite only solution was calibrated using the method of Lerch [1991]. This method involves generating subset solutions that specifically exclude that satellite from the general or master solution. Then the calibration factor is computed for that satellite, where this factor is the ratio of the sum of the coefficient differences between the master solution and the subset, divided by the sum of the coefficient sigma differences squared. The weight of the data in the normal equations can be adjusted so that aggregate changes in the coefficients are commensurate with the changes in the coefficient sigmas.

• All data sets were calibrated using subset solution testing.
• Select data sets, in particular those of important satellites ( TOPEX , Starlette or those satellites that had continuous tracking (i.e. GPS/MET, EUVE went through a second and a third iteration of calibrations. That is, after a new master solution was computed using the updated weights (normal equation scale factors), subset solutions were computed for these select satellites.
• We used comparisons with altimeter derived 5 deg by 5 deg gravity anomalies and GPS/Leveling tests to validate the weights selected.
• In general, to avoid deriving an error model that was too optimistic, we aimed for calibration factor, K, of between 0.8 to 0.9.
• Examples of the calibration factors for EUVE, and for Spot-2/ Stella are shown in the attached figures.

Average calibration factors:
• per degree (to 20) = 0.796;
• per degree (to 70) = 0.710;
• per order (to 70) = 0.661.

Click on image for full size

Calibration factors, and coefficient differences for two satellite only solutions, with and without the EUVE data (TDRSS and GPS).

Average calibration factors:
• per degree (to 20) = 0.713;
• per degree (to 70) = 0.881;
• per order (to 70) = 0.604.

Click on image for full size

back to table of contents.
back to egm96 home page