부산대학교 도서관

The ocean circulation is typically constrained in operational analysis and forecasting systems through the assimilation of sea level anomaly (SLA) retrievals from satellite altimetry. This approach has limited benefits in the Arctic Ocean and surrounding seas due to data gaps caused by sea ice coverage. Moreover, assimilation of SLA in seasonally ice-free regions may be negatively affected by the quality of the Mean Sea Surface (MSS) used to derive the SLA. Here, we use the Regional Ice Ocean Prediction System (RIOPS) to investigate the impact of assimilating Absolute Dynamic Topography (ADT) fields on the circulation in the Arctic Ocean. This approach avoids the use of a MSS and additionally provides information on sea level in ice covered regions using measurements across leads (openings) in the sea ice. RIOPS uses a coupled ice-ocean model on a 3-4 km grid-resolution pan-Arctic domain together with a multi-variate reduced-order Kalman Filter. The system assimilates satellite altimetry and sea surface temperature together with in situ profile observations. The background error is modified to match the spectral characteristics of the ADT fields, which contain less energy at small scales than traditional SLA due to filtering applied to reduce noise originating in the geoid product used. A series of four-year reanalyses demonstrate significant reductions in innovation statistics with important impacts across the Arctic Ocean. Results suggest that the assimilation of ADT can improve circulation and sea ice drift in the Arctic Ocean, and intensify volume transports through key Arctic gateways and resulting exchanges with the Atlantic Ocean. A reanalysis with a modified Mean Dynamic Topography (MDT) is able to reproduce many of the benefits of the ADT but does not capture the enhanced transports. Assimilation of SLA observations from leads in the sea ice appears to degrade several circulation features; however, these results may be sensitive to errors in MDT. This study highlights the large uncertainties that exist in present operational ocean forecasting systems for the Arctic Ocean due to the relative paucity and reduced quality of observations compared to ice-free areas of the Global Ocean. Moreover, this underscores the need for dedicated and focused efforts to address this critical gap in the Global Ocean Observing System.