Enhancing sea ice knowledge through assimilation of sea ice thickness from ENVISAT and CS2SMOS

Abstract. Arctic sea ice extent has declined significantly over the past two decades, opening up the Arctic to shipping and resource extraction while also impacting wildlife and local communities. This has led to an increasing need for skillful sea ice predictions. We focus on furthering the understanding of the role that sea ice thickness plays in the skilfulness of seasonal Arctic sea ice predictions. We look at how observations of sea ice thickness can improve both sea ice reanalyses and predictions. We use the Norwegian Climate Prediction Model (NorCPM), which has previously assimilated ocean and sea ice concentration observations. We additionally assimilate two sea ice thickness products: CS2SMOS, and, for the first time in any study, ENVISAT. This allows us to produce a 2-decade reanalysis with sea ice thickness assimilation focusing on the Arctic Ocean. This reanalysis is then used to initialise and generate a series of year-long seasonal hindcasts for each season of the reanalysis. The reanalysis and hindcasts are compared to observations and other reanalyses to assess the impact of sea ice thickness observations. Assimilation of sea ice thickness data strongly improves the representation of sea ice thickness and volume, primarily in the central Arctic as well as the ice edge location. Although ENVISAT observations have greater uncertainties, the dataset still provides a useful impact on the model. For prediction, sea ice thickness initialisation reduces the model biases of thickness throughout the year as well as errors in the detrended anomalies. Ice thickness bias correction results in improvements in the representation of the ice edge location, i.e., the timing and extent of the summer melting. Thickness initialisation has little improvements for detrended sea ice extent anomalies, but yields some skill in the Beaufort Sea and Central Arctic during summer. Overall, we show the impact of sea ice thickness assimilation has a positive effect on prediction skill in NorCPM.