On-orbit Calibration of GRACE and GRACE-FO accelerometers

The onboard GRACE and GRACE-FO accelerometers measure the non-gravitational accelerations of the spacecraft that are indispensable for the modelling of the Earth’s gravity field since they are subtracted from the GPS POD total accelerations to isolate the pure gravitational accelerations. The accelerometers, like any other instrument, must be calibrated very accurately to serve their purpose but the calibration cannot be done as per standard metrological methods, in a laboratory environment due to the influence of the gravitational signal which is almost 10 orders of magnitude larger. Many researchers have proposed different approaches for the estimation of the bias and the scale factor of the accelerometers, most of which are based on the physical models of the non-gravitational accelerations The drawback of these methods is the dependency of the calibration on the theoretical models and specifically on the drag models that exhibit the highest uncertainty due to the complexity of the upper atmosphere. An alternative, on-orbit calibration of the accelerometers is proposed that is commensurate with standard metrological methods and thus it is based only on the satellite measurements. The idea behind this method lies in a time-reversal method widely used in radar applications for the detection and measurement of known but distorted pulses hidden in the scattered signal from the reflectors. By analogy to radar applications, the total accelerations estimated from GPS through numerical double POD differentiation, correspond to the ‘transmitted signal’ (calibrated) and the accelerometer measurements (uncalibrated) comprise the ‘return signal’ (scattered signal).  The key to this method is that the penumbra transitions are present in both signals and play the role of the known calibration pulse. Examples of 30 daily scale factors are calculated for different operational periods during lower and higher solar activity of GRACE and GRACE-FO and the accelerometer bias is computed by a daily polynomial fit. The results show a robust estimation of the scale factor and bias of the accelerometers. The thermal sensitivity of the accelerometer and its correlation with the β’ angle is investigated.