Fast, flexible, focused: the case for a single-column sea ice data assimilation framework

Assimilating sea ice observations into numerical sea ice and climate models has garnered increasing interest, driven by a demand for more comprehensive sea ice records and forecasts in response to a rapidly changing cryosphere. The development of data assimilation (DA) techniques targeted specifically for sea ice, however, has been comparatively limited.  The computational requirements and structure of many modern sea ice models, the physical characteristics of key sea ice variables, and the uncertainty and relatively limited scope of assimilated sea ice observations all pose significant challenges for the development and tuning of sea ice DA systems. This work presents a new, lightweight framework for sea ice DA development that couples a flexible ensemble DA software to a single-column, multi-category sea ice model, and reviews several recent applications. Key results include the variable impact of common sea ice observation kinds across different sea ice regime types; the benefits of tailoring DA algorithms to the physical and modeled characteristics of sea ice; and the efficacy of assimilating new kinds of observations, including the ice thickness distribution and sea ice albedo. Collectively, these results highlight the ease of experimentation proffered by this new framework, which enables both novel research and more accessible development in sea ice state estimation and forecasting contexts.