The Cryospheric Science Group of the Interdisciplinary Centre on Climate Change (IC3) is pleased to welcome PhD candidate Lei Wang, of the VIP lab within UW Systems Design Engineering, to present his research at our seminar this month. His presentation is entitled:
“Learning to Estimate Ice Concentration from SAR Images”
ABSTRACT
High-resolution ice concentration maps are of great interest for Arctic operations and climate change research. Synthetic aperture radar (SAR) is the main data source for high resolution ice monitoring. Due to the complexity of SAR signatures of sea ice, the automatic interpretation of SAR images has been a difficult task. In this study, a convolutional neural network (CNN) is adapted to estimate ice concentration from dual-polarized SAR images. Trained by ice concentration maps generated by ice experts, the CNN is demonstrated to produce ice concentration maps with more details than produced operationally. Reasonable ice concentration estimations are made in melt regions and in regions of low ice concentration.