Sea ice floe size distribution in the Arctic Ocean

: satellite observations and model evaluation

Abstract

Floe size distribution (FSD) of sea ice plays a crucial role in thermodynamic and dynamic processes related to sea ice, and it has wide-ranging implications for the polar climate system. For a credible projection of Arctic climate and associated changes in the Arctic, it is essential to improve our understanding of FSD-related feedback processes and incorporate it into sea ice/climate models. FSD-related parameterizations have recently been implemented in sea ice models, yet model development and validity are still at an early stage. In particular, limited data from high-resolution FSD observations and lack of model evaluation have been major factors hindering the advance of FSD models. In this thesis, I aim to address these issues by generating and analysing a new high-resolution FSD dataset and using it to evaluate the three current FSD models. The high-resolution FSD dataset has been derived from Measurements of Earth Data for Environmental Analysis (MEDEA) images with a resolution of 1 metre and Worldview (WV) images with a resolution of 0.5 metres in the Arctic. The new FSD data have been used in this PhD project to investigate the statistical characteristics and seasonal evolution of FSD and the validity of the three existing model parameterisations of FSD in the Arctic.

My results show that the spring-summer transition of sea ice floes from continuous ice cover into distinct ice floes is evident, which can be conceptualised by distinctive stages of the life cycle of the FSD. In this thesis, the life cycle is characterised by the spring fracturing stage, transition stage and summer melt/wave fragmentation stage. At the spring fracturing stage, FSD exhibits a single power-law behaviour, associated with sea ice brittle fracture and deformation. During the transition stage, the distribution of floes with mean floe calliper diameter 𝑑 smaller than 200–300 m exhibits a similar power-law slope to the spring fracturing stage, while for larger floes, FSD is fitted by a steeper power-law slope. The results imply that large floes are more vulnerable to fragmentation processes than small floes. During the melt season, melt ponds and weakened ice accelerate the split of ice into smaller floes. Meanwhile, moderate ocean surface waves could also easily fracture sea ice floes that have been already weakened by thermodynamic processes in the melt season. The FSDs follow single power-law distribution at this stage again but with a steeper slope than the one at the spring fracturing stage.

The model evaluation study reveals that the three FSD models generally overestimate floe perimeter density in comparison to observations. It was discovered that the models have a much higher percentage of small floes with a radius smaller than 10–30 m but a much lower proportion of large floes with a radius larger than 30– 50 m. In a comparison of the FSD models against the observations, the WIPoFSD model demonstrates a consistent negative correlation between the concentration of sea ice and the floe perimeter per unit ice area, which is in agreement with the observations. However, CPOM-FSD and FSDv2-WAVE models exhibit contrasting patterns. The discrepancies that exist between these models and observations may result from the limited resolution of the observations in identifying small floes or inaccurate model parameterisations that include the use of a constant global powerlaw exponent in the WIPoFSD model and weak floe welding and overactive wave–ice interaction fractures in the CPOM-FSD and FSDv2-WAVE.

Date of Award	30 Nov 2023
Original language	English
Supervisor	Byongjun (Phil) Hwang (Main Supervisor) & Ryan Wilson (Co-Supervisor)