Wavelet Spectral Techniques for Error Mitigation in the Superconductive Angular Accelerometer Output of a Gravity Gradiometer System

A superconductive angular accelerometer (SAA) is an important sensor that supplements a superconductive angular gradiometer on moving platforms. The SAA measurements compensate one of the main errors, i.e., angular velocity squared in the gradiometer output. While the angular accelerations can be measured, angular velocity squared is computed by integration. However, a major difficulty arises when the angular accelerations are noisy because of translational accelerations of the platform and temperature fluctuations of the environment during the operation. Therefore, filtering of angular accelerations is essential before integration to derive angular velocities. We have implemented wavelet de-noising and de-trending techniques in order to mitigate these errors. The wavelet results are compared with the results of bandpass filtering. Since stationary observations are used, the best filtering method is selected based on its improvements in reducing the root mean square of the known noise floor. The best de-noising results showed 99% improvement in reducing the known noise level, whereas wavelet de-trending and de-noising results indicate a 99.4% improvement. Some bandpass digital filtering are also designed and applied to the angular acceleration, but wavelets perform slightly better than the best bandpass filter. The filtered angular acceleration can improve a gravity gradient error by almost 50%, which can ultimately affect the geophysical interpretation of the gradients. Despite the encouraging results in the laboratory, the expected superiority of wavelet filtering under real dynamic conditions has yet to be demonstrated.