The Gravity Probe B Data Analysis Filtering Approach

Abstract

A simple pre-flight strategy of the Gravity Probe B (GP-B) data analysis has evolved in the elaborate multi-level structure after the discovery of the complex polhode motion, and of the patch effect torques. We describe a cascade of estimators (filters) that reduce the science data (SQUID and telescope signals) to the estimates of the relativistic drift rates. Those estimators, structured in two "floors", are based on the polhode-related models for the readout scale factor and patch effect torque. Results of the 1 st Floor processinggyro orientation profiles-manifest clearly the strong geodetic effect but also the presence of classical torque. Modeling of the patch effect torque at the 2 nd Floor provides a successful compensation of the torque contributions, and leads to consistent estimates of the relativistic drift rates.

Gravity Probe B data analysis: II. Science data and their handling prior to the final analysis

“…As for data compression (2), the majority of science signals had high sampling rates, like 10 Hz for the telescope signals, or 5 Hz for the four SQUID signals. Since the signals typically varied much more slowly, the rates and the corresponding large sets of data points were redundant for the analysis (in fact, even the chosen L2 time step of 2 s was initially thought to be too short, until the need to accurately model patch effect torques was established [8]). This was why the number of data points of frequently sampled signals was reduced between L1 and L2.…”

“…Its 1st floor was a modification of the pre-launch scheme [24] that used batch processing. At the 2nd floor, the more accurate modeling of the scale factor and misalignment torque coefficient took place, integrating information from all the batches.…”

Misalignment and Resonance Torques and Their Treatment in the GP-B Data Analysis

“…One method is to estimate the uniform components of the drift rate as well as the misalignment torque coefficient, using an appropriate model of the time variation of the latter. This method along with preliminary results is described in the accompanying paper by Heifetz et al (2009). A second, complementary method takes advantage of the linear combination of the two drift rate equations (11) which eliminates the torque coefficient.…”

GP-B Systematic Error Determination

“…In addition, this model improves the numerical fit of the model to the data, thereby reducing the residuals. Initial science results making use of the improved model are encouraging (Heifetz et al 2009). …”

“…Readout scale factor induced experiment uncertainty has been probed by comparing the science results obtained using the harmonic expansion method described above with the results using two other techniques. The first of these, the γ C g model (Heifetz et al 2009), uses the low frequency gyroscope signal to determine a body fixed representation of the trapped field contribution to the scale factor. The other, TFM, also determines a body fixed representation of the trapped field contribution to the scale factor, but it does so using the high frequency gyroscope signal.…”