Automatic High-Accuracy Sea Ice Mapping in the Arctic Using MODIS Data

The sea ice cover is changing rapidly in polar regions, and products with high temporal spatial resolution are of great importance studying global climate change navigation. In this paper, an map generation model based on Moderate-Resolution Imaging Spectroradiometer (MODIS) reflectance bands constructed to obtain data a resolution. By constructing training sample library using multi-feature fusion machine learning algorithm for classification, the high-accuracy recognition cloud regions achieved. first product provided by near real-time single-scene presence map. Compared photo-interpreted ground truth, verification shows that can higher accuracy ice, clouds, water, exceeds 98%. second daily weekly clear sky map, which provides synthetic maps one day or seven consecutive days. A filtering method motion used make more accurate. third optical images. comparison Advanced Microwave Scanning Radiometer 2 (AMSR2) concentration performed August 2019 September 2020, these composite images showed consistency over time, suggesting they be many scientific practical applications future.