Determination of the spectral weights of satellite models, and terrestrial and airborne gravity data for local geoid computations – case studies from GSVS11 and GSVS14 area

English Abstract:

For local geoid computations, the least squares collocation has been the standard method to combine several sets of gravity data. An alternative is to use the spectral combination which takes into account the error characteristics of each data represented by their error degree variances (EDV). However, the EDV of terrestrial and airborne gravity data are not known; thus limiting wide use of the method in practice. One way to estimate EDV of terrestrial and airborne gravity data is to use statistical models of red and white noise. This paper explores the feasibility of computing EDV directly from gravity data without using the statistic models. The EDV at low degrees are estimated by the aid of a satellite gravity model, while a leave-out-one error estimation technique is introduced to estimate those at the medium and high degrees directly from the local data themselves. The EDV implied spectral weights restrict the satellite models to long wavelength, the contribution of airborne gravity to the medium wavelength, and terrestrial data to the medium to short wavelengths of the gravity field. The spectral weights are applied in local geoid computations in GSVS11 and GSVS14 areas. The study shows not only the contribution of each data set to the geoid in the frequency domain, but also reveals the degree of correctness of the satellite gravity models.