DYNAMIC ESTIMATION OF DRIFTING ICE TOPOGRAPHY OVER A 2D AREA USING MOBILE UNDERWATER MEASUREMENTS

摘要

We present an estimation design for monitoring a drifting underwater ice topography, for realtime acquisition of high resolution ice thickness data in a local operations area. Additionally, the feature of detection and tracking of significant ice features such as ridges, growlers, bergy bits, and icebergs frozen into the sea ice comes for free from the setup. For continuous underwater monitoring of the drifting sea-ice we are challenged by two obstacles: 1) the ice topography is a dynamic landscape, and 2) we need to communicate the measurement data in real time through a limited hydroacoustic communication channel. Representing the topography by a truncated 2D Fourier series and a spatially discretized set of measurement nodes distributed over the surveillance area, a steepest descent method is proposed to make the underwater ice topography estimate converge to the real topography. Since only a small number of Fourier coefficients is needed to represent the topography sufficiently well, it becomes possible to transmit the estimated topography in real time through the hydroacoustic communication channel and to online reconstruct and present it in a surface-based operation center. Since the nodes in the observation grid will only be measured while the mobile sensor is in range, a nonlinear observer is proposed to filter the measurements and update them dynamically in dead-reckoning when measurements are not available. The effectiveness of the estimation algorithm is simulated based on an extensive and highly varying bottom topography, emulating in our case a typical ice topography.