The ocean’s role in driving Antarctic sea ice predictability

Antarctic sea ice is a critical component of the climate system and acts as a vital habitat for Southern Ocean ecosystems. Understanding the drivers and physical processes influencing Antarctic sea ice predictability is of broad interest. This thesis presents a comprehensive investigation of the physical processes underlying Antarctic sea ice predictability.
This thesis investigated the predictability of sea ice and the upper ocean in the Weddell Sea region using outputs from a high-resolution global coupled sea ice ocean model; the Australian Community Climate and Earth System Simulator (ACCESS-OM2). Predictability in the ice and the ocean are consistent with each other and follow similar evolution. Oceanic predictability or ‘oceanic memory’ is the major source of memory underlying sea ice predictability. Oceanic memory in our analysis (i.e., the correlation between sea ice area and lagged upper ocean temperature), can be interpreted as residual thermal anomalies from sea ice processes lingering in the ice-ocean system at inter-seasonal timescales. This implies that sea ice predictability is a signature of local ice-ocean interaction. Oceanic memory of sea ice anomalies is largely confined to the Winter Water layer (WW) bounded by the permanent pycnocline (PP), which prevents deep water from mixing into the WW and helps sustain the ocean memory. Once the mixed layer deepens in winter to reach the PP, further deepening induces entrainment of warm Circumpolar Deep Water which has no sea ice-related memory into the WW and causes the dissipation of predictability (termed as the predictability barrier). Our analysis shows that sea ice and upper ocean predictability is closely tied to the vertical ocean structure and its seasonal evolution.
Further, the regional variability of sea ice and ocean predictability in six sectors around Antarctica is assessed using model data. Individual sectors produced unique sea ice and ocean predictability patterns and showed strong regional variability. Summer persistence and spring re-emergence are the two main predictability regimes found among the sectors. A major regional contrast in sea ice predictability is the lack of summer persistence in the eastern sectors (King Haakon Sea, East Antarctica, and Ross Sea). This contrast is also seen in the ocean predictability, where the western sectors produced high correlations extending deep (~200 m) into the ocean and persisting longer, while in the eastern sectors, correlations form only in a shallow layer (~50 m) and dissipate earlier. It is inferred that the lack of summer persistence in the eastern sectors is due to the rapid dissipation of oceanic memory to the atmosphere due to weak stratification of the seasonal pycnocline and lack of summer sea ice cover in the eastern sectors.
Finally, the model sea ice predictability is compared with observed sea ice predictability. Sea ice predictability in the model (ACCESS-OM2) is found to be more persistent than observed. On evaluating the climatological ocean conditions in the model and observations, the model biases in sea ice predictability are traced back to systematic biases in the representation of the upper ocean vertical structure. The model has an overly stratified upper ocean, which preserves upper ocean memory over longer timescales and produces longer predictability skill. This over?stratified upper ocean is a consequence of the model ocean being too fresh above the permanent pycnocline, and too warm below. Our study shows that a better representation of the upper ocean in the sea ice zone of the Southern Ocean can produce sea ice predictability of better agreement with observation. Since ocean vertical structure, especially changes in stratification can have a strong influence on predictability, the disparity between observed and model predictability presents a priority area for future model improvements and developments.
This thesis builds on existing research on Antarctic sea ice predictability and presents further evidence on the dominant role of the ocean in retaining ice-ocean memory and driving sea ice predictability. It presents evidence on how the memory in the ocean follows the seasonal evolution of the upper ocean and sheds light on how variability in the upper ocean structure leads to regional differences in sea ice predictability. The findings in this thesis not only elucidate the physical processes in the upper ocean underlying Antarctic sea ice predictability but also present the nature of these processes in the form of an ‘ice-ocean interaction’. The investigation in this thesis is one of the first to provide a detailed assessment of the ocean’s role in modulating seasonal predictability. Our results also imply that changes to upper ocean conditions in a warming climate might alter sea ice predictability patterns in the future.