Developments to 2D modelling of ice island deterioration using the CI2D3 Database

The calving of ice tongues and ice shelves can generate large, tabular icebergs that have climatological implications given their role in dispersing freshwater from the Greenland and Antarctic ice sheets. These ‘ice islands’ also pose potential risk to marine industry. It is therefore critical that influential deterioration mechanisms be accurately represented in simulations of ice island drift and deterioration, both for risk mitigation in offshore industry and for climatological studies that are focused on the Polar Regions. The majority of ice island deterioration is the result of sidewall erosion, and specifically that which results from waterline wave-erosion leading to ram growth and buoyancy-forced fracture. This study therefore focuses on the inclusion of the buoyancy-driven “footloose” calving mechanism (Wagner et al., 2014) in simulations of ice island length and areal change. Using size and lineage information of ice islands tracked in the Canadian Ice Island Drift, Deterioration and Detection (CI2D3) Database, we quantitatively assess the performance of the footloose calving model by simulating the deterioration of 172 ice islands. The mean model error was +15 (+/- 400) m over 20 d and increased to +401 (+1400/-800) m for simulations that ran to 80 d. The performance of the footloose calving model is a substantial improvement when compared to simulations that did not include this calving mechanism. For example, a thermal-melt model had mean errors of -252 and -1403 m at 20 and 80 d of simulation, respectively, and the mean error of a zero-melt model was -281 and -1545 m over the same time periods. We also present a new approach to modelling ice island areal change resulting from footloose calving. This simple, two-parameter approach simulates discrete footloose calving events and adjusts the ice island surface area to maintain a constant aspect ratio. Mean model error remained under 1 km² over 80 d of simulation, showing that the model performs well over numerous months. Using the CI2D3 Database, we were able to conduct the first large-scale assessment of the footloose model’s performance in simulating change to the ice island length dimension. The morphological data included in the database also provided the opportunity to develop an approach for modelling areal deterioration resulting from footloose calving events. The model assessments would benefit from more observations of long-duration ice island tracks, as there were a limited number of ice islands that were tracked in the CI2D3 Database for over 40 d. Future work can look to implement the presented approaches in operational and climatological modelling while the iceberg modelling community also develops an approach to simulate larger-scale ice island fracture.