Gravity Field Re ̋nement by Radial Basis Functions from In-situ Satellite Data

In this thesis, an integrated approach is developed for the regional renement of global gravity eld solutions. The analysis concepts are tailored to the in-situ type character of the observations provided by the new satellite missions CHAMP, GRACE, and GOCE. They are able to evaluate data derived from short arcs of the satellite's orbit and, therefore, oer the opportunity to use regional satellite data for the calculation of regional gravity eld solutions. The regional character of the approach will be realized at various stages of the analysis procedure. The rst step is the design of specically tailored space localizing basis functions. In order to adapt the basis functions to the signal content to be expected in the gravity eld solution, they will be derived from the covariance function of the gravitational potential. To use the basis functions in gravity eld modeling, they have to be located at the nodal points of a spherical grid; therefore investigations will be performed regarding a suitable choice of such a nodal point distribution. Another important aspect in the regional gravity eld analysis approach is the downward continuation process. In this context, a regionally adapted regularization will be introduced which assigns dierent regularization matrices to geographical areas with varying signal content. Regularization parameters individually determined for each region take into account the varying frequency behavior, allowing to extract additional information out of a given data set. To conclude the analysis chain, an approach will be described that combines regional solutions with global coverage to obtain a global solution and to derive the corresponding spherical harmonic coecients by means of the Gauss-Legendre quadrature method. The capability of the method will be demonstrated by its successful application to real data provided by CHAMP and GRACE and to a simulation scenario based on a combination of GRACE and GOCE observations. verdeutlicht.