Classification of Sea Ice

The ultimate objective of this work is to improve characterization of the ice cover distribution in the polar areas and to develop a new real-time high spatial resolution satellite-based ice cover product for use in operational applications. Our particular intent is to identify and adequately reproduce small-scale inhomogeneities in the ice cover distribution be- hind the ice front such as leads and polynyas. To achieve this goal we will use observations in the optical and infrared spectral range made with MODIS sensors onboard the Terra and Aqua satellites and VIIRS instrument onboard NPP satellite platform. The technique for mapping ice cover will combine observations from all three instruments and will make use of multiple observations per day over high latitudes. Leads and polynyas will be identified using both spectral and textural features derived from satellite imagery. A recently developed technique to restore MODIS Aqua band 6 data, critical for ice identification, will be applied to improve ice mapping, cloud mask and ice/cloud discrimination. We will also enhance the current ice identification technique used with MODIS and VIIRS data by accounting for angular anisotropy of the ice reflectance in the visible and near infrared spectral bands.

The primary focus of the study will be on the Beaufort and Chukchi Seas, which are both identified as Navy areas of primary interest. The developed techniques will be used to generate daily maps of ice cover distribution for these areas at 500m spatial resolution. Accuracy assessment of the maps will be conducted through visual inspection of the original satellite imagery as well as through the analysis of high spatial resolution satellite data. Once tested and proved efficient, the technique will be applied to generate daily ice cover products over Arctic and Antarctic basins.

Computational components of the project, including reading and pre-processing data, computing snow and ice products, geographic re-projection, data integration, as well as leads and polynyas features will be factored into online learning modules for use in courses, including the PI’s Satellite Image Processing class. These learning modules will utilize technology for web- based automated testing and evaluation, developed by project senior personnel. In addition, some of the proposed sub-tasks will form the basis for student projects, and student research will result in students giving presentations at appropriate conferences. Web-based visualization tasks will also provide the basis for projects in a web applications class offered by the senior personnel.

Completion of this project will lead to a substantial improvement in the characterization of the ice cover distribution in the high latitude areas. It is anticipated that the new ice cover product generated within this project will contribute to and facilitate interactive ice mapping at the NOAA-Navy National Ice Center and eventually to Navy operations.