Characterizations of the Antarctic Sea Ice Using Multiple Satellite Remote Sensing Data

• Main Author: Kim, Miae
• Other Authors: Im, Jungho
• Format: Doctoral or Postdoctoral Thesis
• Language: English
• Published: Graduate School of UNIST 2018
• Subjects: Amundsen Sea; Antarctic; The Antarctic; West Antarctica; Antarc*; Antarctica; Ice Shelf; Ice Shelves; Iceberg*; Sea ice
• Online Access: https://scholarworks.unist.ac.kr/handle/201301/24580; http://unist.dcollection.net/common/orgView/200000109019

Department of Urban and Environmental Engineering (Environmental Science and Engineering) Sea ice closely interacts with the atmosphere and ocean systems. Land fast sea ice (fast ice) is a kind of sea ice attached to the shore, ice shelves, or grounded icebergs. It is widely distributed along the Antarctic coast and acts as an interface between the atmosphere and the ocean, affecting heat balance feedback, thermal insulation effects, and deep water formation depending on the temporal and spatial effects of the environmental conditions. It also plays an important role in the biological aspects of Antarctica. Attached to the Antarctic glacier is strongly associated with calving events of ice shelf as it is physically coupled with glaciers at the terminus. The existing Antarctic fast ice has been mainly focused on the East Antarctic, especially for the research on long-term fast ice. Several case studies for West Antarctic fast ice with satellite images were performed in local areas. Various types of satellite data and detection techniques were utilized to successfully detect fast ice. In addition, long-term fast ice maps specifically focused on the Amundsen sea of West Antarctica were generated to investigate the distribution and variability of fast ice. This thesis reports the results of fast ice detection algorithms that have been developed using various satellite images that can be used for fast ice detection. Along with the use of multiple satellite data, the proposed fast ice detection algorithms can more effectively detect fast ice, which then allows to obtain more accurate fast ice detection and produce long-term fast ice with high accuracy. Especially, the distribution and variability of time-series fast ice in West Antarctica, which is more concentrated in the Amundsen Sea, were analyzed together with bathymetry data and the distribution of glacier icebergs. In order to detect fast ice, machine learning techniques were basically used in this thesis. Two classes (i.e. fast ice and non-fast ice) were ...