Adaptive filtering in airborne gravimetry with hidden Markov chains.

Abstract

AbstractThe paper studies filtering of airborne gravimetry data on a survey line. The aim is to build a filter that adapts to the nonhomogeneous gravity structure, which is induced by the unknown structure of Earth core, topography, etc. Gravity is modeled by a hidden Markov chain with finite number of states, each state corresponds to a certain type of gravity profile. Filtering is done in three steps: estimate the parameters of the Markov chain, determine moments of transitions between states, and construct the corresponding non-stationary Kalman filter. The algorithm takes into account the type of noise of the global positioning system. Tests on simulated and experimental data show that the filter diminishes over-smoothing and under-smoothing effects.