The use of GPS sensors and numerical improvements in Aeromagnetic Compensation

Abstract

SummaryAeromagnetic compensation for the magnetic effects (noise) of the aircraft is now a significant limiting factor in the final data quality. Fundamentally, the residual noise from the aircraft movement after compensation is still 100 s of times larger than the signal to noise capabilities of a modern cesium sensor. Users who require highly accurate data for interpretation of subtle features can utilize acquisition systems capable of the required sensitivity and software capable of utilizing the required sensitivity but the final data is flawed by relatively high noise from the aircraft movement effects.Surveys over magnetic terrains face the additional problem that the aircraft attitude required for the compensation techniques are often incorrect. This is due to the measurement of the aircraft attitude via the use of a vector fluxgate magnetometer and the assumption that the Earth s field is uniform. This problem is envisaged to increase as surveying heights get lower and lower, increasing the anomalous magnetic field sensed by both the vector fluxgate and the airborne magnetometer. Non-magnetic orientation devices, such as the use of three GPS receivers on the aircraft are an obvious solution to this problem.The results of this work demonstrates that it is possible to use three well spaced GPS receivers on an aircraft to measure the aircraft attitude to the precision required to compensate for the aircraft effects.The experiments with the different highpass filters used on the data and with the different solvers indicate that with the proper selection of filters and solvers, better compensation results can be obtained.

An Adaptive Filter for Aeromagnetic Compensation Based on Wavelet Multiresolution Analysis

“…However, considering only the frequency characteristics of the aircraft maneuvers and filtering the data with a frequencydomain filter cannot always guarantee an expected result. In fact, the Gaussian filter applied in time-domain turns out to be more appropriate [16]- [18] than the frequency-domain filter, which is also confirmed by our simulation results in Section IV. This may be caused by the shortcomings of the discretetime Fourier transform.…”

“…The results of the frequencydomain filter are comparatively worse; about half of the cases fall in the range between 0.04 and 0.05 nT, and about 30% of the cases exceed 0.05 nT. This verifies the relevant inferences in [16]- [18]. For the proposed adaptive filter, more than 70% of the cases fall in the range between 0.01 and 0.02 nT, and all cases are within 0.04 nT.…”

“…. , q}, (2 ≤ p ≤ q ≤ J, p, q, J ∈ Z) (16) as the consecutive subscript group of the selected coefficient vectors. The corresponding reconstructed results of Δ and H T are denoted by bpf Γ (Δ) and bpf Γ (H T ), respectively.…”

“…To address this problem, many scholars have presented different solutions to estimate the coefficients based on the advanced regression theory, e.g., [22]- [24]. Meanwhile, considering that a bandpass or high-pass filter should be applied to the system of linear equations to extract the signal component related only to the aircraft maneuvers, Jia, Lo, and Groom pointed out that the issue of multicolinearity could be resolved by selecting an appropriate filter and recommended a Gaussian filter (see [16]- [18]). Unfortunately, however, the selection of this filter is not justified in detail.…”