An analysis of ice drift data collection methods using marine radar imagery

Abstract

This thesis presents an analysis of the use of marine radar imagery for the collection of ice drift data, which can be used for scientific, operational and safety purposes. A general analysis of state-of-the-art approaches for ice drift data collection is performed. This discusses the usage of GPS-buoys, satellites and marine radar images. In addition this the various marine radar image ice drift data collection techniques are discussed in-depth. Three distinct techniques to collect ice drift data with marine radar images are identified from this analysis. These three techniques are correlation, feature detection and optical flow based approaches and are incorporated in proof of concept algorithms. These algorithms are configured and tested with a dataset collected by the Oden icebreaker in 2015(OATRC15). Each algorithm’s performance, qualities and deficiencies are assessed by validating it with available ice drift data from GPS-buoys and ship positional data. Based on this validation shortcomings and weaknesses in the algorithms are identified. Finally a number of expansions and improvements for each algorithm type are proposed, aimed at solving their shortcomings and weaknesses. The proposed algorithms lay the groundwork for improved algorithms that can provide accurate ice drift data collection performance using marine radar images.