Dynamic estimation of drifting ice topography using underwater mobile measurements

In this paper we present an estimation design for a drifting two dimensional ice topography. Under the assumption that the topography can be accurately represented by a truncated Fourier series, and given a set of measurements distributed discretely along the topography, the steepest descent method is proposed to solve the problem. The chosen method allows different weighting on the various measurements and coefficients such that the weight on possible faulty measurements can be reduced, while important coefficients can be prioritized. In cases where measurements of the ice topography are available only periodically, an observer is proposed such that non-measured nodes can be dynamically updated. In both cases the effectiveness of the estimation algorithm is simulated in a constructed ideal scenario, as well as a scenario where a set of real ice measurements are taken from the underside of an ice sheet off the cost of Greenland.