Sea Ice Mapping from SAR: An overview of methods and approaches developed during 8 years of CIRFA
Johannes Lohse
After operating for 8 years, the Center for Integrated Remote Sensing and Forecasting for Arctic Operations (CIRFA) ended in December 2023. Located at the intersection between research and industry, the main objective of CIRFA was to use satellite remote sensing for the analysis of ocean surface conditions such as sea ice and icebergs, ocean wind and currents, and marine pollution, in order to provide information for safer shipping and operations in the Arctic as well as data assimilation into numerical models for improved sea ice and ocean forecasts.
In this presentation, I will give an overview of different classification and segmentation approaches for sea ice SAR imagery that were developed within CIRFA’s work package 2 (“Remote sensing of sea ice and icebergs”). The methods include both traditional algorithms, such as statistical classifiers that were tailored towards the particular challenges of SAR data and sea ice applications, as well as machine-learning approaches based on convolutional neural networks. I will briefly present the basic ideas and concepts of each method, outline how the different approaches can be combined to obtain better or more useful results, and cover the advantages as well as remaining challenges of each algorithm to hopefully help direct future research in the upcoming years. Finally, I will show an application example that uses one of the algorithms in near-real-time operational support for an Arctic expedition, pointing out advantages and drawbacks in comparison to operational ice charts.