Assimilation of RCM data in the Canadian ice concentration analysis system

The sea and lake ice concentration pan-Arctic analysis system at Environment and Climate Change Canada (ECCC) initializes both the short-range Arctic sea ice forecasting models and numerical weather prediction tools. In this study, our previously developed approach for deriving ice concentration from RADARSAT-2 was adjusted to become applicable to the RADARSAT Constellation Mission (RCM) data. This algorithm adaptation was done by incorporating RCM beam-specific wind speed retrieval models and enhancing the quality control procedure. We found an excellent agreement (correlation = 0.996) between ice concentrations derived from nearly 4000 RCM images and ice concentrations from the image analyses manually produced by the Canadian Ice Service (CIS) over a one-year period between August 1, 2020 and July 31, 2021. Then, for data assimilation experiments in the ECCC ice analysis system, we extracted ice concentrations at 1.6 km resolution from >65,000 RCM images acquired over the same test time period and over the whole Arctic. We demonstrated that assimilation of RCM-derived observations significantly improves ice analyses especially over lakes, in the vicinity to the coasts, and marginal ice zones. Accuracy of the ice analyses in the areas that are not covered by the CIS charts such as Alaska, High Arctic, and Eurasian Arctic substantially increased after the assimilation of RCM data. Visual assessment of the ice analyses against optical data showed that high-resolution details in ice concentration fields were successfully introduced by the RCM retrievals. Operational implementation of near-real-time RCM data assimilation in the ice concentration analysis system that initializes the ECCC Regional Ice-Ocean Prediction System is underway.