A new high-resolution Coastal Ice-Ocean Prediction System for the East Coast of Canada

Abstract. This paper describes the Coastal Ice Ocean Prediction System for the East Coast of Canada (CIOPS-E) running operationally at Environment and Climate Change Canada (ECCC). CIOPS-E uses a one-way downscaling technique on a 1/36° horizontal grid (~2 km) to simulate high-resolution ice and ocean conditions over the northwest Atlantic Ocean and the Gulf of St. Lawrence (GSL). CIOPS-E is forced at its lateral boundaries with ECCC’s Regional Ice-Ocean Prediction System (RIOPS) and tidal conditions from the Oregon State University TPXO model. The three-dimensional temperature and salinity fields are spectrally nudged towards the RIOPS solution offshore of the 1500 m isobath to, effectively constrain mesoscale features in the Gulf Stream area. Over the continental shelf and the GSL, the CIOPS-E solution is free to develop fully according to model dynamics.

CIOPS-E is evaluated over one year from March 2019 to February 2020. Overall, the CIOPS-E improves the representation of tides compared to ECCC’s lower resolution systems: RIOPS (1/12°) and the Regional Marine Prediction System – Gulf of St. Lawrence (RMPS-GSL, 5 km). The accuracy of the tides are comparable to the TPXO at most coastal tide gauges. Sub-tidal water levels from CIOPS-E agree well with the observed seasonal variability and show improved errors statistics at all stations compared to RIOPS and RMPS-GSL. Improvements are especially noted for the GSL.

Sea surface temperatures (SSTs) from CIOPS-E are lower (higher) in spring (fall) over most of the GSL compared to satellite-derived analyses and RIOPS. Comparison with in-situ observations of SST show significant improvement in CIOPS-E with respect to the RMPS-GSL. Lastly, sea ice conditions in the GSL are compared with the Canadian Ice Service (CIS) charts and the RMPS-GSL model. The sea ice cover and thickness from the pseudo-analysis component (without data assimilation) shows an overestimation compared to the CIS estimates, which is subsequently corrected in the forecast phase through the direct insertion of a Radarsat image analysis product.