Qiang Wang, UiT The Arctic University of Norway will talk about Pixel-wise sea ice mapping with deep learning.
Monitoring sea ice in polar regions is critical for understanding global climate change and supporting marine navigation. Currently, ice charting is the main resource for understanding the sea ice status. However, the spatial resolution of ice-charting is in the order of km, which is not enough for some applications, for instance, search -and-rescue, offshore operation close to the coast , etc. Here, we present a deep learning-based method to generate sea ice map in the polar region by using Sentinel-1 dataset. Meanwhile, an incidence angle based data augmentation scheme is being utilized to enrich the training dataset in our methodology.