Regularization of regional gravity fields from GOCE data
Wednesday, 17 December 2014
Regional gravity field recovery in spherical radial base functions is a strongly ill-posed problem. The ill-posedness is mainly due the restriction of observations as well as the base functions to a specific area. We investigate this ill-posedness as well as the related regularization process. We compare four different methods for the choice of regularization parameter and discuss their characteristics. Moreover, two different kinds of shape coefficients for the spherical radial base functions are used to assess the impact of shape coefficients on the regularization process. These shape coefficients are the Shannon coefficients with no smoothing properties and the Spline kernel with smoothing features. As the data set, we use two months of real GOCE ultra-sensitive gravity gradient components which are Txx, Tyy, Tzz and Txz. The regional solutions are considered in Amazon. Our results indicate, that the method used for the choice of regularization parameter is directly influenced by the shape of spherical radial base functions. In addition we conclude, that the Shannon kernel along with a proper regularization approach delivers satisfactory results with physically meaningful coefficients.