Noise Reduction in Full Tensor Gradiometry Data Using the Fourier Method

Abstract:

The full tensor gravity gradiometer independently measures 5 components of gradient. Although independently measured, these fields are connected to one another by relations imposed on them by the underlying field equation. This enables noise reduction. If gradient components can be calculated from one another multiple versions of each measured channel can be obtained. These alternate versions of the same signal are then combined, or stacked, together in a weighted average to generate a reduced noise product.

When gradient components are measured on a flat surface and transformed to Fourier domain they can be calculated from each other using simple algebraic expressions. Therefore the Fourier method offers a direct means of reducing noise by converting gradient components. It is also fast due to the ready availability of implementations of the Fast Fourier Transform algorithm. However, application of the Fourier method to airborne gradiometry survey data presents two challenges: (1) How to Fourier transform an arbitrary shaped survey area and (2) How to take into account changes in altitude. After a review of the theory behind the Fourier method an outline will be given of how these issues have been solved.

The effectiveness of noise reduction will be demonstrated using both real and simulated data. This will be accompanied by a discussion of why the degree of noise reduction differs between gradient tensor components. The use of the Fourier method to interpolate, continue and integrate survey data will also be demonstrated.