A key quantity in oceanography is the ocean's dynamic topography. While satellite altimetry has provided the oceanographic community with a good picture of the time variable dynamic topography for some time now, direct determination of the mean dynamic topography (MDT), which is simply the difference between the mean sea-surface and the geoid, has been restricted by the lack of a sufficiently accurate geoid. From an oceanography perspective the primary benefit from GOCE will be the resolution of this problem. However, because of the contrasting representations of the constituent surfaces, and their differing error characteristics, care is needed in the choice of algorithms used to derive the MDT. The objective of this paper is to assess a number of elementary approaches for computing and filtering the MDT. For the GOCE User Toolbox (GUT) determination of good elementary procedures is important since, compared with more sophisticated methods, such procedures are less computationally expensive, and are easier for a user to understand and implement. Two main computational algorithms are considered: conventional differencing of the mean sea-surface height and geoid height anomaly in the geographical domain, and differencing in the spectral domain. It is shown how the latter approach significantly reduces numerical artifacts in the resulting MDT. We also consider simple methods of filtering the MDT to remove noise. The problems of naive filtering, including coastal artifacts and the attenuation of oceanographic signals are considered, and methods to resolve these issues without incurring a high computational burden are presented.
Bingham et al. (Thu,) studied this question.