PhD Defence Andreas Rønne Stokholm

Hovedvejleder

• Seniorrådgiver Sine Munk Hvidegaard, DTU Space

Medvejledere

• Professor René Forsberg, DTU Space
• Lektor emeritus Leif Toudal Pedersen, DTU Space

Eksaminatorer:

• Lektor Rasmus Tage Tonboe, DTU Space
• Lektor Ekaterina Kim, Norsk Universitet for naturvidenskab og teknologi, Norge
• Seniorforsker Juha Karvonen, Finsk Meteorologisk Institut, Finland

Formand ved forsvar

• Afdelingsleder Allan Hornstrup

Summary

The ever-changing sea ice is critical to map for safe and efficient navigation in the remote and cold Arctic, as ships can get stuck and capsize in the ice. Sea ice mapping is also important for monitoring the state of the climate and as information input to weather and climate models because the sea ice acts as an insulating material between the ocean and the atmosphere. Global warming is causing the amount of sea ice in the Arctic to be diminishing, which makes this region more hospitable and navigable. New economic opportunities are emerging, such as adventure tourism, resource extraction and the opening of new trade routes to connect the Pacific and Atlantic oceans through the Arctic. However, less sea ice is believed to result in a more dynamic sea ice environment, so hazardous conditions will remain present. For these reasons, sea ice mapping will continue to be relevant and can be viewed as a critical infrastructure component in the Arctic.

At many national ice services, professional ice analysts draw sea ice charts daily based on satellite radar images, which allow sea ice observation throughout the year in relatively high resolution, independent of sunlight and clouds. However, the radar images are challenging to interpret because the radar measurements depend on the measurement angle and the observation surface, where the roughness and material composition influence the measurements, which can cause ocean and sea ice to appear identical. Consequently, the sea ice analysts analyse the radar images manually with their in-depth knowledge and understanding to create precise and detailed sea ice charts. The manual task is a time and resource-demanding task that limits the number of produced ice charts and delays the delivery of the critical information. The work carried out in this PhD thesis investigates the opportunities to automate the production of sea ice maps using deep-learning methods within artificial intelligence applied to satellite radar images.

Sea ice analysts chart various sea ice information, such as the sea ice concentration, describing the amount of sea ice in the ocean in an area, the stage of development, indicating the age of the ice, a proxy for its thickness, and the floe size, which describes the number of floes and the degree of sea ice breakup. All these sea ice parameters are valuable information for maritime navigation. During the project, Convolutional Neural Networks were developed to create models that can automatically map sea ice. Different aspects of improving model performance to map sea ice concentration were investigated, such as, how the noise correction in the radar images and the number of pixels viewed by the model. Furthermore, different model goals and how they can improve the models' mapping capability have been explored. Methods to combine these different goals in the model mapping process have also been developed with multiple models responsible for different subtasks within sea ice concentration mapping. Finally, an international competition was conducted where participants were tasked with mapping both sea ice concentration, stage of development and the floe size.