Gravity Vector Estimatiton from a Large and Densely Spaced Heterogeneous Gradient Data Set Using Closed-Form Kernel Approximations

摘要

An estimator was recently developed by R. Rummel that captures most advantages of least squares collocation and the integral model approach without many of their disadvantages. It was developed as an alternative to the inverse Stokes integral for processing large, densely spaced satellite altimetry data sets by accounting also for measurement errors. This idea is extended here to heterogeneous gravity gradient data sets at altitude. The appropriate kernel function is expressed as a Fourier transform of its spectrum which in the flat earth approximation is fitted by pieces of powers of spatial frequency and thereby can be integrated analytically. A further enhancement to the estimator is suggested which allows the suboptimal incorporation of isolated data of possibly different kinds.