Accelerometer calibration for thermospheric neutral density estimation with GRACE data by dynamic Precise Orbit Determination (POD) with tailored parametrization

The neutral mass density of the upper atmosphere (thermosphere) can be determined by orbit and accelerometer data from Low Earth Orbit (LEO) satellites. Especially the accelerometers of geodetic satellites, measuring the non-gravitational accelerations acting on the satellites, are a very useful observation for precise density determination also on very short time scales.

Nevertheless, the accelerometers of geodetic satellites are affected by bias and drift. Therefore a calibration of the data is indispensable. A time dependent bias and scale factor are to be determined. Usually calibration parameters are estimated by dynamic Precise Orbit Determination (POD) or Gravity Gield Recovery (GFR), together with all other parameters of interest. In both cases, the estimated accelerometer calibration parameters are not the major interest. With the used parametrizations and weighting of the observations, good gravitational field and orbit solutions, do not necessarily give good or physical accelerometer calibration solutions. This is unsatisfying, especially for the anticipated use in density determination and the direct comparison to modeled non-gravitational accelerations.

In this contribution we use dynamic POD and investigate different parametrization strategies tailored for an accurate and physical accelerometer calibration for thermospheric density determination. For example we investigate the effect of constraining the accelerometer calibration parameters in that way, that a continuous calibration over all arcs is achieved, where normally each arc is treated locally separated from all other arcs, leading to jumps in the calibration. The scale factor, which is highly correlated to the estimated bias, is concurrently estimated but over a longer batch of arcs. We compare different bias parametrizations, arc lengths, as well as different observation data and weighting strategies.

Finally we show some preliminary density estimation results with our approach and the influence of the accelerometer calibration.