A probabilistic model for fracture events of Petermann ice islands under the influence of atmospheric and oceanic conditions

• Main Authors: Zeinali-Torbati, Reza, Turnbull, Ian D., Taylor, Rocky S., Mueller, Derek
• Format: Text
• Language: English
• Published: 2021
• Subjects: Arctic; Newfoundland; Petermann glacier
• Online Access: https://doi.org/10.5194/tc-2021-83; https://tc.copernicus.org/preprints/tc-2021-83/

Four calving events of Petermann Glacier happened in 2008, 2010, 2011, and 2012, which resulted in the drift and deterioration of numerous ice islands, some reaching as far as offshore Newfoundland. The presence of these ice islands in the eastern Canadian Arctic increases the risk of interaction with offshore operations and shipping activities. This study used the recently developed Canadian Ice Island Drift, Deterioration and Detection database to investigate the fracture events that these ice islands experienced, and presented a probabilistic model for the conditional occurrence of such events by analyzing the atmospheric and oceanic conditions that drive the causes behind the ice island fracture events. Variables representing the atmospheric and oceanic conditions that the ice islands were subjected to were extracted from reanalysis datasets and then interpolated to evaluate their distributions for both fracture and non-fracture events. The probability of fracture event occurrence for different combinations of input variable conditions were quantified using Bayes theorem. Out of the seven variables analyzed in this study, water temperature and ocean current speed were identified as the most and least important contributors, respectively, to the fracture events of the Petermann ice islands. It was also revealed that the ice island fracture probability increased to 75 % as the ice islands encountered extreme (very high) atmospheric and oceanic conditions. A validation scheme was presented using cross-validation approach and Pareto principle, and an average error of 13–39 % was reported in the fracture probability estimations. The presented probabilistic model has a predictive capability for future fracture events of ice islands and could be of particular interest to offshore and marine activities in the eastern Canadian Arctic. Future research, however, is necessary for model training and testing to further validate the presented ice island fracture model.