Session: 07-02-01 Arctic Sea Transportation

Paper Number: 101757

101757 - Assist Observations 2017-2021: Uncertainty, Comparison With Sea Ice Charts, and Ice Concentration From Deep Learning Models 

With receding ice edge due to climate change, both research and commercial interest in the Arctic has increased leading to an increase in the number of marine operations in icy waters. With that, there has been an increase in the number of ASSIST observations (https://icewatch.met.no/assist) from ~300 observations a year in 2006 to ~1500 observations in 2020. These are manual observations taken at high resolution (typically covering one nautical mile around the ship) and made available freely on https://icewatch.met.no. While this is great for better localized monitoring of the Arctic ice cover, the manual nature of these observations makes them highly subjective. The experience level of the ice observer can significantly affect their observations. The present study will evaluate the uncertainty of total/partial ice concentrations reported in ASSIST observations in two ways. First, ASSIST observations for the year 2017-2021 will be compared against the sea ice charts from the Norwegian Meteorological Institute (MET Norway), the National Ice Center (NIC) and the Arctic and Antarctic Research Institute (AARI). And second, as a test case, images of sea ice scenes from the north of Svalbard (Oct-Nov 2022) will be shown to multiple ice experts and non-experts, and their reported concentrations will be compared: against each other, against the sea ice charts as well as against the sea ice concentration calculated by a deep learning model from optical images. The results from this study will enhance the usefulness of ASSIST database by supplementing it with uncertainty information and help in formulation of more comprehensive guidelines for manual sea-ice observations. Furthermore, the comparison between ice concentrations from manual observations and deep learning model will establish a benchmark in automated image analysis for extraction of sea ice concentration. 

 

Presenting Author: Nabil Panchi Norwegian University of Science and Technology

Presenting Author Biography: I received my master's degree in Ocean Engineering and Naval Architecture from Indian Institute of Technology, Kharagpur. My summer stay at NTNU inspired me towards research at the intersection of Computer Science and Arctic Engineering. I am currently pursuing a PhD in AI's application and challenges in the Arctic marine environment.

Authors:

Nabil Panchi Norwegian University of Science and Technology
Ekaterina Kim Norwegian University of Science and Technology
Roger Skjetne Norwegian University of Science and Technology
Nick Hughes Norwegian Meteorological Institute