A study of non-stochastic IMU errors in strapdown airborne gravimetry

Abstract

Strapdown airborne gravimetry is the determination of the Earth's gravity acceleration from an aircraft, using a strapdown inertial measurement unit (IMU) in combination with an aiding sensor (typically GNSS). Several publications show, that a modern navigation-grade IMU in combination with two-frequency phase-differential GNSS enables the gravity determination to the level of several micro-g. We hereby present a quantitative analysis of the impact of non-stochastic IMU errors on the gravity determination, based on simulated gravity flights with realistic dynamics. Several error types are discussed, including biases, scale factors, cross-couplings and misalignments of the sensor triads, each for both the accelerometers and gyroscopes. In particular, it is shown that the on-line estimation of in-flight accelerometer bias changes is not possible for this application, due to the inseparability against changes in the gravity signal. For each type of stochastic IMU errors, a threshold value is provided where the average error of the estimated gravity reaches 1 mGal. As an example, these threshold values are compared to actual accelerometer calibration results of an iMAR RQH unit. It is shown for this device, that a proper calibration is required in order to reach gravity estimate accuracies at the 1 mGal level.