Abstract
Accurate and timely forecasts of sea ice conditions are crucial for safe shipping operations in the Canadian Arctic and other ice-infested waters. Given the recent observations on the declining trend of Arctic sea ice extent over the past decades due to global warming, machine learning (ML) approaches are deployed to provide accurate short-term to long-term forecasting. This study unlike previous ML approaches in the sea-ice forecasting domain provides a daily spatial map of the probability of ice in the study domain up to 90 days of lead time. The predictions are further used to predict freeze-up/breakup dates and show their capability to capture these events within a valid time period (7 days) at specific locations of interest to communities.
Highlights
Sea ice presence is an important variable for northern communities and shipping operators
Given that our models predict spatial maps of sea ice presence probability over a grid at a spatial resolution of 31 km, we first apply a 50% threshold to this probability to convert each pixel to ice or water
E.g., the first top-left box in this figure (Fig 1(b)) corresponds to the average accuracy after 1 day forecast for all forecasts launched between January 1 and January 31, ending in January 2 to April 1 and the second box corresponds to average accuracy of forecasts launched between January 1 and January 31 ending in January 3 to April 2
Summary
Sea ice presence is an important variable for northern communities and shipping operators. A recent approach that is closer to what is proposed here is IceNet (Andersson et al, 2021), which applies an ensemble of CNNs, each with a U-Nets architecture, to produce monthly maps of ice presence (probability SIC > 15%) for the 40 6 months Input to this model consists of climate variables, which is similar to other studies. The method is similar to operational forecasting studies (Chevallier et al, 2013; Sigmond et al, 2013) except we are using a data-driven statistical approach, as compared to a physics-based model, and our forecasted variable is a number between 0 and 1 that indicates an (uncalibrated) probability of ice at a grid location, as compared to sea ice 50 concentration. All the input variables except sea ice concentration and landmask are normalized before being input to the network
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
