Criteria of iceberg hazard in the seas of the Russian Federation Arctic zone

• Main Authors: I. A. Bychkova, K. G. Kortikova, V. G. Smirnov, И. А. Бычкова, К. Г. Кортикова, В. Г. Смирнов
• Other Authors: The work under this project is supported by the Russian Science Foundation through the Project № 23-27-00122, Исследование выполнено при финансовой поддержке Российского научного фонда (грант № 23-27-00122)
• Format: Article in Journal/Newspaper
• Language: Russian
• Published: Государственный научный центр Российской Федерации Арктический и антарктический научно-исследовательский институт 2024
• Subjects: спутниковый мониторинг; convolutional neural network; degree of iceberg hazard; satellite monitoring; visible spectral range; арктические моря; видимый диапазон; сверточная нейронная сеть; Arctic; Severnaya Zemlya; Iceberg*
• Online Access: https://www.aaresearch.science/jour/article/view/615; https://doi.org/10.30758/0555-2648-2024-70-2-238-252

The aim of the study was to develop quantitative criteria of iceberg hazard for the seas of the Russian Arctic based on the results of statistical processing of the data of regular satellite surveys. The study used non-commercial data of the visible spectral range from the Landsat-8 satellite (spatial resolution of 15 m) for 2014–2023. It is proposed that the criteria of iceberg danger be expressed in points. The main criterion is the value of the average daily density of icebergs in squares of a regular grid obtained for a particular month (season) of the year and expressed on a 5-point scale. Additional criteria included registration in the iceberg grid with a length of more than 200 m (extensive icebergs according to the classification of the World Meteorological Organization, WMO), icebergs with a surface height of more than 25 m, grounded icebergs (iceberg traps), and icebergs with high drift speeds (>1 km/hour). The additional criteria are estimated at 1 point on the iceberg hazard scale. To detect icebergs in the images, a technique was used based on statistical criteria for searching for gradient zones in the analysis of two-dimensional fields of satellite images. A convolutional neural network based on the U–Net architecture was used to automate the detection of grounded icebergs among drifting ice floes. Examples are given of the application of the criteria developed to the Severnaya Zemlya region. The possibility is discussed of zoning the water area according to the degree of iceberg hazard using the criteria suggested. Задачей исследования была разработка количественных критериев айсберговой опасности для морей Российской Арктики, основанных на результатах статистической обработки материалов регулярной спутниковой съемки. В качестве источника информации использовалась некоммерческая информация видимого диапазона с ИСЗ Landsat-8 (пространственное разрешение 15 м) за 2014–2023 гг. Критерии айсберговой опасности предложено выражать в баллах. Основным критерием является величина среднесуточной ...