| Summary: | Some 250000 point gravity data for a 900x1 000km area covering the British Isles and the North Sea have been combined with a detailed topographic model on land and with a selection of global geopotential models in a high resolution, high precision geoid computation using two different techniques. The gravity data, comprising bouguer anomalies on land and free air anomalies at sea, together with the long wavelength gravity effect of the topographic model, were detrended with free air anomalies calculated on the ellipsoid from a high degree and order spherical harmonic geopotential model. The residual anomalies and the condensed short wavelength topography were transformed to the potential using both a fouricr technique and a classical Stokes' algorithm. The results were retrendcd with the geopotential model to produce absolute geoid heights. For each method, long wavelength residuals in the detrended gravity anomalies, caused by errors in the geopotential models, generate significant long wavelength errors which propagate through to the final geoid solution. It was found that these errors could be minimised by tapering the data and altering the shape of the data set, in the case of the fourier technique, and by using Meissl's modification to Stokes' function in the case of the classical technique. Long wavelength geoid discrepancies between different algorithms and detrending with different global gravity models involved slopes of typically less than 2cm/100km. For the cumulative effect of wavelengths between about 4 and 200km, computed geoids differed at the 1cm level. For local gravity field computations, the fourier technique was found to be 104 quicker than the classical method, for no appreciable loss in accuracy. Comparisons with short-baseline GPS-derived relative heights showed them to agree with the gravimetric geoid solutions at the 4cm level. Preliminary relative geoid heights derived from longer baseline GPS surveys agree with relative gravimetric geoid heights to 5cm/100km. Internal estimates give absolute long wavelength errors in gravimetric geoids to be approximately 2cm/100km, suggesting that the GPS observations so far available have not always achieved the accuracy of the gravimetric geoid computations. A comparison of geoid solutions from this project with the previous best geoid over the British Isles, EGG1, showed the new solutions to have a greatly improved resolution.
|
|---|
